Está en la página 1de 240

Elena Rubei

Algebraic Geometry

Also of Interest
Knots
Burde/Zieschang/Heusener, 2013
ISBN 978-3-11-027074-7, e-ISBN 978-3-11-027078-5

Topology of Algebraic Curves


Degtyarev, 2012
ISBN 978-3-11-025591-1, e-ISBN 978-3-11-025842-4

Lectures on the Topology of 3-Manifolds


Saveliev, 2011
ISBN 978-3-11-025035-0, e-ISBN 978-3-11-025036-7

Abstract Algebra
Carstensen/Fine/Rosenberger, 2011
ISBN 978-3-11-025008-4, e-ISBN 978-3-11-025009-1

Advances in Geometry
Grundhfer/Strambach (Managing Editors)
ISSN: 1615-715X, e-ISSN 1615-7168

Elena Rubei

Algebraic
Geometry

A Concise Dictionary

Mathematics Subject Classification 2010


14-00, 14-01
Author
Prof. Dr. Elena Rubei
Universit degli Studi di Firenze
Dipartimento di Matematica e Informatica U. Dini
Viale Morgagni 67/A
50134 Firenze
Italy
rubei@math.unifi.it

ISBN 978-3-11-031622-3
e-ISBN 978-3-11-031623-0
Set-ISBN 978-3-11-031600-1
Library of Congress Cataloging-in-Publication Data
A CIP catalog record for this book has been applied for at the Library of Congress.
Bibliographic information published by the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.dnb.de.
2014 Walter de Gruyter GmbH, Berlin/Boston
Typesetting: PTP Protago TEX-Production GmbH, Berlin (www.ptp-berlin.de)
Printing and binding: CPI buch bcher.de GmbH, Birkach
Printed on acid-free paper
Printed in Germany
www.degruyter.com

Notation

We denote by the affine space of dimension over .

()

For any manifold , () denotes the vector space of the -forms on


.

(, )

For any topological space and ring , (, ) (Betti number) denotes the
rank of (, ); it is also denoted by (, ); see Singular homology and
cohomology.

, ()

b.p.f.

For any almost complex manifold (see Almost complex manifolds, holomorphic maps, holomorphic tangent bundles), , () denotes the vector space
of the (, )-forms on .

The abbreviation b.p.f. stands for base point free, see Bundles, fibre -.

()

For any algebraic variety , () denotes the -th Chow group of ; see
Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard
groups.

()

For any algebraic variety , () denotes the divisor class group of ; see
Divisors.

, stands (as usual) for the Kronecker delta.

(, )

For any vector bundle on a manifold , (, ) denotes the set of


the sections of ; see Bundles, fibre -.

()

For any Cartier divisor , () denotes the line bundle associated to ; see Divisors.

||, L()

For any divisor , we denote by || the complete linear system associated to ,


see Linear systems; see also Linear systems for the definition of L().

, , 1

See Pull-back and push-forward of cycles, Direct and inverse image


sheaves, Singular homology and cohomology.

For any line bundle on a variety , denotes the map associated to ; see
Bundles, fibre -.

(, ), (, )

For any vector space , (, ) denotes the Grassmannian of -planes in , see


Grassmannians; analogously, for any projective space , (, ) denotes the
Grassmannian of projective -planes in .

(, ), (, )
(, ), (, )

denotes a linear system on a Riemann surface of degree and projective


dimension ; see Linear systems.

For any topological space and any ring , (, ) denotes the -th homology
module of with coefficients in (see Singular homology and cohomology);
(, ) denotes its rank; (, ) is also denoted by (, ) (Betti number)
For any topological space and any ring , (, ) denotes the -th cohomology module of with coefficients in (see Singular homology and cohomology); (, ) denotes its rank.

viii | Notation
(, E)

()

(, E)
,
F , F

( , )

nef

For any sheaf of Abelian groups E on a topological space , (, E) denotes


the -th cohomology group of E, see Sheaves; (, E) denotes its rank; if
is a holomorphic vector bundle on a complex manifold or an algebraic vector bundle on an algebraic variety , we sometimes write (, ) instead of
(, O()).
For any topological space , () denotes the EulerPoincar characteristic of
, i.e., the sum =1,..., (1) (, ), when it is defined.

For any sheaf of Abelian groups E on a topological space , (, E) denotes


the EulerPoincar characteristic of E, i.e., =1,..., (1) (, E), when it is defined.

For any complex manifold or smooth algebraic variety , denotes the


canonical bundle and the canonical sheaf, i.e., O( ); see Canonical bundle, canonical sheaf.
See Serre correspondence.

We denote by ( , ) the space of the matrices with entries in .

The abbreviation nef stands for numerically effective; see Bundles, fibre -.

()

For any algebraic variety , () denotes the -th NeronSeveri group of ;


see Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard
groups.

O , O ()

If is a complex manifold, O (or simply O) denotes the sheaf of holomorphic functions; if is an algebraic variety, it denotes the sheaf of the regular
functions; more generally, it denotes the structure sheaf of a ringed space; see
Space, ringed -; for the definition of O (), see Hyperplane bundles, twisting sheaves.

O(), ()

(), (), ()

()

Let be an algebraic vector bundle on an algebraic variety or a holomorphic


vector bundle on a complex manifold; then O() denotes the sheaf of the
(regular, resp. holomorphic) sections of ; see Bundles, fibre -. We denote
O() by ().

For any complex manifold , denotes the sheaf of the holomorphic -forms;
for any algebraic variety, denotes the sheaf of the regular -forms; see
Zariski tangent space, differential forms, tangent bundle, normal bundle.
We denote by the projective space of dimension over .

The symbols (), () and () denote respectively the arithmetic genus,


the geometric genus and the -th plurigenus of (for variety or manifold);
see Genus, arithmetic, geometric, real, virtual -, Plurigenera.

For any algebraic variety , () denotes the Picard group of ; see Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard groups.

PID

PID stands for principal ideal domain, i.e., for an integral domain such that
every ideal is principal.

(, )

For any topological space and any , (, ) denotes the -th fundamental group of at the basepoint ; see Fundamental group.

Notation

()

,
()

ix

For any complex manifold or algebraic variety , () denotes the irregularity


of ; see Irregularity.
See Direct and inverse image sheaves

For any ideal , () denotes the saturation of ; see Saturation.

For any , we denote by the group of the permutations on elements.

(), ()

For any ring , () denotes its spectrum, see Schemes. See Schemes
also for the definition of () for graded ring.

The symbol denotes the intersection of cycles; see Intersection of cycles.


Sometimes it is omitted.

The symbol denotes the fibred product of and over ; see Fibred
product.

For any ideal in a ring , we denote by the radical of , i.e.,

, + , ( : )

Let be a commutative ring and and be two ideals in . We define to


be the ideal {=1,..., | , , } . Moreover, we define + =
{ + | }, and ( : ) := { | }.

: (, ) (, )

= { | {0} s.t. }.

For any vector space , we denote by its dual space.

denotes the disjoint union.

If and , the notation : (, ) (, ) stands for a map :


such that () .

Note. The end of the definitions, theorems, and propositions is indicated by the symbol

A
Abelian varieties.

See Tori, complex - and Abelian varieties.

Adjunction formula. ([72], [93], [107], [129], [140]). Let be a complex manifold
or a smooth algebraic variety and let be a submanifold, respectively a smooth closed
subvariety. We have
= | , ,
where , is the normal bundle and and are the canonical bundles respectively of and (see Canonical bundle, canonical sheaf, Zariski tangent space,
differential forms, tangent bundle, normal bundle).
If, in addition, is a hypersurface, the formula becomes
= | ()| ,

where () is the bundle associated to the divisor , since, in this case the bundle ,
is the bundle given by (see Bundles, fibre - for the definition of bundles associated
to divisors).

Albanese varieties.

([93], [163], [166]). The Albanese variety is a generalization


of the Jacobian of a compact Riemann surface (see Jacobians of compact Riemann
surfaces) for manifolds of higher dimension.
Let be a compact Khler manifold of dimension (see Hermitian and Khlerian
metrics and Hodge theory). The Albanese variety of is the complex torus (see
Tori, complex - and Abelian varieties)
() :=

0 (, 1 )
,
(1 (, ))

where : 1 (, ) 0 (, 1 ) is defined by () = for any 1 (, ).


The Albanese map
: ()

is defined in the following way: we fix a point 0 of (base point) and we define

() =
0

for any , where is the integral along a path joining 0 and (thus, obviously, it
0

defines a linear function on 0 (, 1 ) only up to elements of (1 (, ))). If we choose


a basis 1 , . . . , of 0 (, 1 ) we can describe in the following way:

() = ( 1 , . . . , ).

2 | Algebras
For any compact Khler manifold of dimension , the Albanese variety of is isomorphic to the -th intermediate Griffiths Jacobian of . The Albanese variety of a
smooth complex projective algebraic variety is an Abelian variety, that is, it can be
embedded in a projective space. See Jacobians, Weil and Griffiths intermediate -,
Tori, complex - and Abelian varieties.

Algebras. We say that is an algebra over a ring if it is an -module and a ring


with unity (with the same sum) and, for all , and , we have
() = () = ().

Algebraic groups. ([27], [126], [228], [235], and references in Tori, complex - and
Abelian varieties). An algebraic group is a set that is both an algebraic variety (see
Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms
and rational maps) and a group and the two structures are compatible, that is, the
map
,

(, ) 1

is a morphism between algebraic varieties.

Structure theorem for algebraic groups (Chevalleys theorem). Let be an algebraic


group over an algebraically closed field. Then there exists a (unique) normal affine
subgroup such that / is an Abelian variety.

Definition. We say that an algebraic group is reductive if all its representations are
completely reducible (see Representations).

Almost complex manifolds, holomorphic maps, holomorphic tangent bundles. ([121], [147], [192], [251]). An almost complex manifold is the

data of
a manifold ;
a section of the bundle (where is the real tangent bundle)
such that, if we see as a map
: (, ) (, )

(where (, ) is the vector space of the sections of ), we have that


2 = ,

where is the identity map; in other words, for every , the linear map :
induced on the real tangent space of at is such that
2 = .

Almost complex manifold

We can extend to := by -linearity. We define the holomorphic tangent space of at to be the -eigenspace of ; we denote it by 1,0
and we denote the holomorphic tangent bundle, i.e., the bundle whose fibre in is
1,0 , by 1,0 . We define the antiholomorphic tangent space of at to be the
-eigenspace of ; we denote it by 0,1 and we denote the antiholomorphic tangent
bundle, i.e., the bundle whose fibre in is 0,1 , by 0,1 . Thus
= 1,0 0,1 .

Obviously a complex manifold (see Manifolds) is an almost complex manifold: if


{ = + } are the coordinates on a coordinate open subset, we can define by
(

)=
,

)=

for any . Thus, in the case of a complex manifold,


where

1,0 = (

:=

We define { , } to be the dual basis of

0,1 = (

) ,

:=
+
.

, .

Let

:= +

and

:= .

Observe that ( ) = 2, , where , is the Kronecker delta.

Remark. There is an isomorphism

1,0

given by 12 ( ()) (observe that, through such isomorphism,


Analogously, there is an isomorphism
given by 12 ( + ()).

goes to

1
).
2

0,1

For any almost complex manifold , let , () be the space of the (, )-forms
on .

NewlanderNirenberg theorem. Let (, ) be an almost complex manifold of (real)


dimension 2. The almost complex structure is induced by a complex structure if and
only if one of the following conditions holds:

4 | Almost complex manifold


(1) for all , (, 1,0 ), we have [, ] (, 1,0 ) (where [, ] stands
for );
(2) for all , (, 0,1 ), we have [, ] (, 0,1 );
(3) for all 0,1 (), we have 1,1 () 0,2 () and for all 1,0 (), we
have 1,1 () 2,0 ();
(4) for all , (), we have +1, () ,+1 () for every , {0, . . . , };
(5) for all , (, ), we have
[, ] + [, ] + [, ] [, ] = 0

(the first member of the equality is called Nijenhuis tensor).


Definition. Let be a complex manifold. We can decompose

: , () +1, () ,+1 ()

as

= + ,

where

: , () +1, (),
: , () ,+1 ()

are the compositions of respectively with the projections

+1, () ,+1 () +1, (),

+1, () ,+1 () ,+1 () .

Let : be a map between two complex manifolds. By extending the


differential
:
by -linearity, we get a map

: .

We say that is holomorphic if one of the following equivalent conditions holds:

(i) for every component = 1 + 2 of in local coordinates of and , we have

(ii)

and

= 1 ;

0 for every = 1, . . . , (), where { } are local coordinates of ;

(iii) = (where and denote the operators on and respectively), i.e., the differential of is -linear for the complex structures given
by ;

Bertinis theorem

(iv) (1,0 ) 1,0 :

(v) (0,1 ) 0,1 .

Ample and very ample.


Anticanonical.

See Bundles, fibre - (or Divisors).

See Fano manifolds.

Arithmetically CohenMacaulay or arithmetically Gorenstein.

CohenMacaulay, Gorenstein, (arithmetically -,-).

Artinian.

See

See Noetherian (and Artinian).

B
Base point free (b.p.f.)
Beilinsons complex.

See Bundles, fibre -.

([5], [23], [63], [207], [209], [137]).

Beilinsons theorem I. Let F be a coherent sheaf on (see Coherent sheaves). Then


there exists a complex of sheaves

+1

0 +1 1 0

with = , s.t. = () (F()) such that



F
={
0
1

if
if

= 0,
= 0.

Every morphism () () induced by one of the morphisms is zero.

Beilinsons theorem II. Let F be a coherent sheaf on . Then there exists a complex
of sheaves

+1

0 +1 1 0

with = , s.t. = O() (F ()) such that



F
={
0
1

if
if

= 0,
= 0.

Every morphism O() O() induced by one of the morphisms is zero.

Bertinis theorem. ([93], [104], [107], [129], [136], [228]). On a smooth quasiprojective algebraic variety over an algebraically closed field of characteristic 0,

6 | Bezouts theorem
the general element of a finite-dimensional linear system (see Linear systems) is
smooth away from the base locus of the system.

Bezouts theorem.

field.

([72], [93], [104], [107], [196]). Let be an algebraic closed

Bezouts theorem. Let and be two projective algebraic varieties of respective dimension and in . Suppose that + and that, for every irreducible
component of and for general point of , and are smooth at and
( ) = () + ( ), where denotes the tangent space at . Then
deg( ) = deg() deg( ) .

(See Degree of an algebraic subset for the definition of degree).


By using intersection multiplicities (see Intersection of cycles), we can state a
stronger result: suppose that and are two projective algebraic varieties of dimension and in with + and suppose they intersect properly, i.e.,
the codimension of every irreducible component of is the sum of the codimension of and the codimension of ; then, by using an appropriate definition
of intersection multiplicity of and along , which we denote by (, ), we have
that
() ( ) = (, ) (),

where the sum runs over all irreducible components of (see, e.g., [72]).

Bielliptic surfaces.
Big.

See Surfaces, algebraic -.

See Bundles, fibre - or Divisors.

Birational.

See Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps.

Blowing-up (or -process). ([22], [93], [104], [107], [196], [228]). We follow
mainly [93] and [104].
Roughly speaking, the blow-up of a manifold along a subvariety is a geometric transformation replacing the subvariety with all the directions pointing out from it. For
instance the blow-up of a manifold in a point replaces the point with all the directions
pointing out from it.
We define the blow-up of in a point as follows. By changing coordinates we
can suppose = 0; we define the blow-up of in 0 as the set
0 ( ) := {(, ) 1
| }

Blowing-up

(recall that 1
= ( ) is the set of lines of passing through 0), with the projection

: 0 ( ) ,
(, ) .

Observe that
1 () = {

a point
{0} 1

if
if

= 0,
= 0.

Precisely, if = 0, the set 1 () is {(, )} where is the unique line through and
1
0. Thus we can say that 0 ( ) is obtained from by replacing 0 with
, which

represents the set of the lines of passing through 0.


We call 1 (0) the exceptional divisor of the blow-up, and we denote it by .

Fig. 1. Blowing up.

By using the definition of blow-up of in a point, we can define the blow-up of any
manifold in a point. Let be a complex manifold and let . We define the blowup of in as follows: let be a neighborhood of ; the blow-up of in is

8 | Blowing-up
the set

:= () := (( {}) ())/ ,

where is the relation that identifies the points in {} in {} with the points
of {} in (), with the obvious projection
: () .

Observe that restricted to () 1 () is an isomorphism onto {}. Again,


:= 1 () is called the exceptional divisor of the blow-up in . We say that is
obtained from by blowing down .

deLet be an analytic subvariety of . The proper or strict transform of in ,


noted by , is defined by
:= 1 ( {}) = 1 () ,

that is, it is the closure of 1 () minus the exceptional divisor.

We can define also the blow-up of a manifold along a submanifold. Let < ; the
blow-up of along = { | +1 = = = 0} is defined to be
( ) = {(, ) 1 | [+1 : : ] }

with the projection map

: ( ) ,
(, ) .

Observe that the map is an isomorphism from ( ) 1 ( ) onto .


Let be a complex manifold of dimension and let be a submanifold of dimension . Let { } be a family of open subsets such that for every , is contained in := and, for every , there exist local coordinates 1 , . . . , on such
that = {+1 = = = 0}. Let
:

be the blow-up of along . We have that 1 ( ) and 1 ( ) are


isomorphic, so we can glue the blow-ups : to get a manifold and a map
: . Let
:= () := ( ( ))/

where is the equivalence relation given by the identification of the points of


in and in . We define the blow-up of along to be () with the map
: () equal to on and equal to the identity on .

Let be an algebraically closed field. The blow-up can be defined for algebraic varieties over in the following way. Let be an affine algebraic variety over and
be a closed subvariety. Let {0 , . . . , } be a set of generators of the ideal of in . Let
:

Blowing-up

be the rational map


[0 () : . . . : ()]

We define the blow-up of along as the graph of with the obvious projection
: (the graph of a rational map is defined to be the closure of the graph
of | , where is any open subset on which is defined). The divisor 1 () is called
the exceptional divisor of the blow-up. If is a projective variety, we take as 0 , . . . ,
homogeneous polynomials of the same degree generating an ideal whose saturation
(see Saturation) is the ideal of in , and we define the blow-up analogously.
We can prove that the definition of blow-up doesnt depend on the choice of 0 , . . . ,
and that, for any open affine subset in , we have that the blow-up of along
is equal to the inverse image of in the blow-up of along . The map :
is birational.
For example, the blow-up of in 0 is
with the projection

1
| }
0 ( ) = {(, )

: 0 ( ) ,

0 ( )

(, ) .

of an affine algebraic variety containing 0,


The proper transform in
i.e., 1 ( {0}), is isomorphic to the blow-up of in 0.

Proposition. Let be a nonsingular algebraic variety of dimension over and let


; let : be the blow-up in and call the exceptional divisor. Then we
have
= + ( 1)(),
= () ,
()
where () is the line bundle associated to (see Canonical bundle, canonical
sheaf and Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard
groups for the definitions of and ; see Bundles, fibre - for the definition of
the line bundle associated to a divisor; see Pull-back and push-forward of cycles for
the definition of ),
the bundle induced by on is the dual of the hyperplane bundle on (see
Hyperplane bundles, twisting sheaves).
Suppose now that is a projective surface; then we have
2 = 1 (which is an obvious consequence of the fact that the bundle induced by
on is the dual of the hyperplane bundle on );
() ( ) = for any , divisors of ;
= 0 for any divisor of .
Blowing-ups are the fundamental building blocks in birational geometry (see Hironakas decomposition of birational maps, Varieties, algebraic -, Zariski topology, reg-

10 | Buchbergers algorithm
ular and rational functions, morphisms and rational maps). Furthermore, they are
useful in the study of singularities (see Regular rings, smooth points, singular points
and Genus, arithmetic, geometric, real, virtual -).

Buchbergers algorithm.
Bundles, fibre -.

See Groebner bases.

([93], [110], [128], [135], [169], [228], [230], [237]).

Definition. A fibre bundle (bundle for short in the following) is a quadruple (, , , ),


where , , are topological spaces,
:

is a continuous surjective map, and there exists an open covering { } of and


homeomorphisms
: 1
such that the following diagram commutes:
1 ( )
FF
FF
FF
|1 ( ) FFF

/ ,
yy
yy
y
yy 1
y
|y

where 1 : is the projection onto the first component. We call the


trivializing homeomorphisms.

Observe that, for all , the fibre 1 () is homeomorphic to . It is generally denoted by .


We say that is the fibre of the bundle, is the base and is the total space. For the
sake of brevity, we sometimes say that is the bundle (on ) or that : is the
bundle.

We say that two bundles on , (, , , ) and ( , , , ), are isomorphic if and


only if there is a homeomorphism : such that = .
More generally, a morphism from a bundle on , (, , , ), to another bundle on
, ( , , , ), is a continuous map : such that = .
We say that a bundle (, , , ) is trivial if it is isomorphic to ( , 1 , , ) where
1 : is the projection onto the first component.
If (, , , ) is a bundle, we say that an open subset of is trivializing if 1 () is
trivial, i.e., there is a homeomorphism : 1 () such that 1 = .
Example. One of the simplest examples of a nontrivial bundle is the Mbius strip,
which is a bundle on the circle with a segment as fibre.

Let (, , , ) be a bundle and let { } be a trivializing open covering of and


: 1 for be trivializing homeomorphisms. For any , , let

Bundles, fibre -

11

us consider the composition


|1
(

|1 (

( ) 1 ( ) ( ) ;

it sends (, ) to (, ) for some ; we define

, : (),

(where () is the set of the homeomorphisms of ) to be the map such that


(, ) = (, , ()())

for any , . The , are called the transition functions of the bundle . They satisfy
()

, () , () = , () , , , .

Conversely, given topological spaces , , an open covering { } of and functions , : () satisfying () and such that the maps (, )
(, , ()()) are homeomorphisms on ( ) , it is easy to construct a bundle
on with fibre and with the , as transition functions: define
= ( )/

where, if (, ) and (, ) , we say that (, ) (, ) if and only if =


and = , ()(). The transition functions determine the bundle up to isomorphism.
Let (, , , ) be a bundle. Let and and suppose there exist a trivializing
open covering { } for and trivializing homeomorphisms : 1 ( )
such that
(1 ( ) ) = ;
in this case we say that (, | , , ) is a subbundle of (, , , ).

A section of a bundle (, , , ) is a continuous map


:

such that () for all .


If { } is a trivializing open covering for the bundle and : 1 is a
trivializing homeomorphism, we can consider
( ) | : .

It sends a point to a point whose first coordinate is , let us say (, ()). The
maps
:

12 | Bundles, fibre which we have just defined have the property that
() = , ()( ())

for any . Giving is equivalent to giving the maps .

If (, , , ) is a bundle, is a topological space and : a continuous map,


then the pull-back of through is ( , , , ) where
:= := {( , ) | ( ) = ()}

and : is the projection onto the first factor. The pull-back bundle is
denoted by . Observe that
( ) = {( , )| () = ( )} = ( ) .

We say that a bundle (, , , ) is a differentiable, respectively holomorphic, if ,


, and are differentiable manifolds, respectively complex manifolds, and the trivializing functions are diffeomorphisms, respectively biholomorphisms. In this case, we
have obviously that the functions (, ) , ()() are also differentiable, respectively holomorphic.
By a section of a differentiable, respectively holomorphic, bundle (, , , ), we
mean (unless otherwise specified) a differentiable, respectively holomorphic, map
: such that () for any .

We say that a bundle (, , , ) is a vector bundle over a field if is a topological


-vector space, the fibres have a structure of -vector space and there exist { }
trivializing open covering and trivializing homeomorphisms such that the maps
| : {} are vector spaces isomorphisms.
Thus the images of the transition functions , are in ().
In this case, the dimension of is called the rank of the bundle. A vector bundle of
rank 1 is called a line bundle. A complex vector bundle is a vector bundle over the
field .

Let (, , , ) and ( , , , ) be two -vector bundles on and let { } be an open

covering of trivializing for both bundles. Let the transition functions be , and ,
respectively.
The direct sum is the bundle whose fibre is and whose transition functions
are

0
( ,
).

0
,,
The tensor product is the bundle whose fibre is and whose transition

functions are , ,,
. The tensor product ( repeated times) is denoted
by , or, if is a line bundle, also by or .

Bundles, fibre -

13

The wedge product is the bundle whose fibre is and whose transition functions are , . If = , where is the rank of , then is called determinant
bundle and denoted by () (and the transition functions are obviously (, )).
The dual bundle is the bundle with fibre on is and whose transition
functions are
, ()1 .
If has rank 1, the bundle is denoted also by 1 , since is trivial.

Let (, , , ) be a vector bundle. Let and be a vector subspace of and


suppose there exist a trivializing open covering { } for and trivializing homeomorphisms : 1 ( ) such that
(1 ( ) ) = .

We say that (, | , , ) is a (vector) subbundle of (, , , ).


Let , be the transition functions of . Then the transition functions of are
(

,
0

,
)
,

for some functions , , , . The quotient bundle / is the bundle whose fibre on
is / and whose transition functions are , . We have that = if and only if
, = 0 for all , .
A morphism between vector bundles on , precisely from (, , , ) to ( , ,
, ), is a continuous map : such that = and such that :
is linear.
Let () := ( ) and () := ( ). They determine subbundles if and
only if the rank of does not depend on .

A frame of a vector bundle (, , , ) on an open subset of is a set of sections


{1 , . . . , } of on such that for all the set {1 (), . . . , ()} is a basis of .
Obviously, giving a frame of on is the same as giving a trivializing of on .
The projectivized of a vector bundle on , which we denote by (), is the bundle
on whose fibre on is ( ) with the obvious trivializing maps.

Let be a topological group. A bundle (, , , ) is said to be a principal -bundle


if there is an action of on that preserves the fibres and acts freely and transitively
on every fibre.
Let (, , , ) be a principal bundle with transition functions , . Let be a topological space. A homomorphism : () determines a bundle on with fibre
: the bundle
:= / ,

14 | Bundles, fibre where

(, ) (, (1 ))

for any .
Obviously, if is a topological vector space and : (), then the determined
bundle is a vector bundle on with fibre .

Definition. An (algebraic) vector bundle of rank on an algebraic variety over a


field is a scheme with a surjective morphism : such that every fibre has
a structure of -vector space and there is an open cover { } and trivializing maps
: 1 ( ) such that the are isomorphisms, we have 1 = , where
1 : is the projection onto the first factor, and the maps induced by the
from 1 () to {} are vector spaces isomorphisms.

By a section of an algebraic bundle : on an algebraic variety , we mean


(unless otherwise specified) a morphism : such that () for any ,
where is the fibre 1 (). In an obvious way we can define the morphisms and the
isomorphisms between algebraic vector bundles.
When we speak of a vector bundle on an algebraic variety, we always mean an algebraic vector bundle.

Let be an algebraic variety. We can associate a sheaf to any vector bundle on


in the following way: let O() be the sheaf associating to any open subset of the
set of the sections of on . Analogously for holomorphic vector bundles on complex
manifolds. We call O() the sheaf of sections of .
The map sending a vector bundle to the sheaf O() gives a bijection between the
set of vector bundles on a smooth algebraic variety up to isomorphisms and the
set of locally free sheaves of O -modules of finite rank up to isomorphisms (see [228,
Chapter 6] or [107, Chapter 2, Exercise 5.18]).
We recall that the group of Cartier divisors of an algebraic variety up to linear equivalence is called Picard group of and denoted by () (see Divisors and Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard groups).

Definition. We can associate to any Cartier divisor of an algebraic variety a


line bundle we denote by () in the following way: if the Cartier divisor is given by
local data ( , ), we define () to be the line bundle whose transition functions ,
from to are / .
Theorem. The map associating to a Cartier divisor of an algebraic variety the line
bundle () induces an isomorphism from the Picard group () to the group of the
isomorphism classes of line bundles on , where the multiplication is the tensor product and the inverse is the dual. The group of the isomorphism classes of line bundles

Bundles, fibre -

15

on is isomorphic also to 1 (, O ), via the isomorphism induced by the map sending a bundle to its transition functions.

If is a smooth projective algebraic variety over , one can prove that, for any divisor
, the first Chern class (see Chern classes) of the line bundle associated to is the
Poincar dual of the class of .

Definition. Let be a projective algebraic variety over an algebraically closed field


(or a compact complex manifold) of dimension . Let be an algebraic line bundle
on (resp. a holomorphic line bundle on the complex manifold ); in this case we
have that 0 (, O()) is finite dimensional, see for instance [107, Chapter 2, 5] and
[93, Chapter 0, 6]. Let
be the rational map

: (0 (, O()) ),

{ 0 (, O())| () = 0};

defined only on the set of the points of such that there exists 0 (, O()) with
() = 0.

We say that is very ample if the associated map is well defined on the whole and
is an embedding. In this case we have that (O(1)) = , where O(1) is the hyperplane
bundle on (0 (, O()) ) (see Hyperplane bundles, twisting sheaves).
We say that is ample if there exists such that is very ample.
We say that is numerically effective (nef) if (| ) 0 for any (irreducible)
curve in , that is the intersection number is nonnegative for any divisor such
that the associated line bundle is and any (irreducible) curve (see Intersection of
cycles).
Let
() = { | 0 (, O( )) = 0};
we define the Iitaka dimension of to be if () = {0}, to be
max { ()}

()

if () = {0} and is normal, to be the Iitaka dimension of on if is not normal


and : is the normalization of (see Normal).
The line bundle is said to be big if its Iitaka dimension is equal to .

We say that a vector bundle on is ample if O() (1) is ample on (), where O() (1)
is the line bundle that restricted to any fibre of () is O(1), i.e., the dual of the
universal bundle (see Tautological (or universal) bundle).
Let be a vector bundle on . We say that 0 (, O()) is base point free (b.p.f.)
if there does not exist such that () = 0 for all . We say that is base
point free if there does not exist such that () = 0 for all 0 (, O()).

16 | Bundles, fibre Let be a vector bundle on . We say that is globally generated if for all we
have that is generated by 0 (,O()) ().
Obviously, for line bundle, being globally generated is equivalent to being base point
free.
We say that a Cartier divisor is ample, very ample, big, nef, b.p.f., if and only if the
corresponding line bundle () is.

We say that a holomorphic line bundle on a complex manifold is positive if in the


first Chern class 1 () 2 (, ) there is a positive (1, 1)-form (see Positive).

If is a nef Cartier divisor on a complex projective algebraic variety of dimension ,


we can prove that is big if and only if > 0 (see Intersection of cycles for the
definition of the intersection number ); see, e.g., [169].
NakaiMoishezon theorem. ([107], [142], [190], [203]). Let be a projective algebraic
variety of dimension over an algebraically closed field. Let be a Cartier divisor
on . Then is ample if and only if > 0 for all irreducible subvariety in
with = and for all (see Intersection of cycles for the definition of the
intersection number ).

Thus ample implies nef. We have also that, under nice assumptions, for instance if
is a smooth projective algebraic variety over an algebraically closed field, the interior
part of the cone generated by the nef divisors in the space of the divisors up to numerical equivalence is the cone generated by the ample divisors; see [142] (we say that two
divisors are numerically equivalent if their intersection number with any curve is the
same).

By Kodaira embedding theorem (see Kodaira embedding theorem) a holomorphic


line bundle on a compact complex manifold is positive if and only if it is ample.
Observe that, if is a line bundle associated to a divisor , then the positivity of is
not equivalent to the effectivity of (and no implication is true: to show that the implication is false one can take a noneffective divisor of positive degree on a Riemann
surface; to show that that the implication is false one can take the exceptional divisor of the blow-up of a surface in a point: this is effective, but the associated line
bundle () is not positive, because, by NakaiMoishezon criterion, it is not ample,
since 2 is equal to 1.)
Finally one can easily prove that b.p.f. implies nef.

Canonical bundle, canonical sheaf

17

Summarizing, for a line bundle on a projective algebraic variety over an algebraically


closed field, the following implications hold:
very ample
b.p.f.

nef

ample

big

See also Sheaves, Chern classes, Equivalence, algebraic, rational, linear -, Chow,
NeronSeveri and Picard groups, CartanSerre theorem.

C
CalabiYau manifolds.

([94], [252]). There are several possible (also nonequivalent) definitions. One of the most common definitions is the following (see Canonical
bundle, canonical sheaf and Hermitian and Khlerian metrics for the definitions
of canonical bundle and Khler manifold).
Definition. A CalabiYau manifold is a compact Khler manifold with trivial canonical
bundle.
Another (nonequivalent, precisely weaker) definition is: a CalabiYau manifold is a
compact Khler manifold such that there exists a finite (holomorphically) covering
space (see Covering projections) of with trivial canonical bundle.

Examples. Elliptic Riemann surfaces and K3 surfaces satisfy both definitions, while
Enriques surfaces satisfy the second definition, but not the first (see Riemann surfaces (compact -) and algebraic curves, Surfaces, algebraic - for the definitions).

Canonical bundle, canonical sheaf.

([93], [107], [228]). Let be a complex


manifold. The canonical bundle is the determinant bundle of the dual of the holomorphic tangent bundle 1,0 (see Almost complex manifolds, holomorphic maps,
holomorphic tangent bundles). The canonical sheaf, denoted by , is the sheaf of
the holomorphic sections of the canonical bundle, i.e., O( ).

If is a smooth algebraic variety of dimension over an algebraically closed field, we


define the canonical bundle to be the determinant of dual of the tangent bundle ,
i.e., to be the bundle determined by the sheaf (see Zariski tangent space, differential forms, tangent bundle, normal bundle).
The associated sheaf of sections, i.e., O( ), coincides obviously with and is called
canonical sheaf; it is denoted also by .

18 | Cap product
Example.
= ( 1),

where is the hyperplane bundle on (see Hyperplane bundles, twisting


sheaves).
See also Dualizing sheaf.

Cap product.

See Singular homology and cohomology.

CartanSerre theorems.

([42], [93], [103], [107], [223]).

Serres theorem. Let F be a coherent sheaf (see Coherent sheaves) on a projective


algebraic variety over a field . For any , let F() be F O () (see Hyperplane
bundles, twisting sheaves for the definition of O ()). Then
(i) the -vector space (, F) has finite dimension for any and has dimension
0 for every > ();
(ii) there exists such that, for any and for any , the stalk F() is
spanned, as O, -module, by the elements of 0 (, F());
(iii) there exists such that (, F()) = 0 for all , > 0.
Statements (ii) and (iii) are often called Serres theorem A and B, respectively.

Furthermore, in [42], Cartan and Serre proved that, for any coherent sheaf F on a compact complex manifold , the complex vector space (, F) is finite dimensional for
any .

CastelnuovoEnriques Criterion.

See Surfaces, algebraic -.

CastelnuovoEnriques theorem.

See Surfaces, algebraic -.

CastelnuovoDe Franchis theorem.

See Surfaces, algebraic -.

Categories.

([26], [79], [116], [168], [179]). We follow mainly the exposition in [79].
A category C consists of
(i) a set of objects;
(ii) for every ordered pair of objects, , , a set denoted C (, ) (or simply
(, )), whose elements are called morphisms or arrows;
(iii) for each triple of objects , , , a map
(, ) (, ) (, )

(the image of (, ) (, ) (, ) by this map is called the composition of and and is denoted by or by ) such that

Categories

19

(a) (1 , 1 ) and (2 , 2 ) are disjoint unless 1 = 2 and 1 = 2 ;


(b) the associativity of the composition holds;
(c) for each object , there exists an arrow 1 (, ) such that 1 =
and 1 = for any , arrows.

We write an arrow (, ) as : .

Definition. Let C, D be two categories; a (covariant) functor from C to D (we will


write : C D) is given by a map, which we denote again by , from the set of the
objects of C to the set of objects of D and, for each ordered pair of objects , of C, a
map, which we denote again by , from C (, ) to D ((), ()) such that
(i) () = ()(),
(ii) (1 ) = 1() .
We can define the composition of two functors in the obvious way.
Definition. Let be a functor from a category C to a category D.
We say that is faithful if, for any , objects of C, we have that
: C (, ) D ((), ())

is injective.
We say that is full if, for any , objects of C, we have that
is surjective.

: C (, ) D ((), ())

Definition. Let C be a category. The dual category C is the category defined in the
following way: the objects are the objects of C (an object in C will be denoted by
as an object of C ); we define C ( , ) := C (, ) (an arrow : in
C (, ) will be denoted : as an element of C ( , )) and we
define = ( ) and = () .
A controvariant functor from C D is a covariant functor C D, thus it is given
by a map from the set of the objects of C to the set of objects of D and, for each ordered
pair of objects , of C, a map
C (, ) D ((), ()) .

Definition. We say that a category C is a subcategory of a category C if


the set of the objects of C is a subset of the set of the objects of C;
for any , objects of C , we have that C (, ) is a subset of C (, ):
the composition of the morphisms in C coincides with the composition of the morphisms in C and, for any object of C , the identity morphism in C coincides
with the identity morphism in C.

20 | Categories
Definition. We say that two objects , in a category C are isomorphic if and only if
there are two arrows : and : such that = 1 and = 1 .

Definition. Let C, D be two categories and and be two functors from C to D. A


morphism of functors (called also a natural transformation) from to
:

is a family of arrows in D, () : () (), one for each object of C, such that,


for any arrow : in C, the diagram
()

()


()

()
()

/ ()

()


/ ()

is commutative, i.e., () () = () ().

Taking the natural transformations as morphisms, we get that the functors from C to
D form a new category, which we denote by (C, D).
In particular two functors , from C to D are isomorphic if there exist a natural transformation from to and a natural transformation from to such that =
and = . Equivalently, and are isomorphic if there exists a natural transformation from to such that () : () () is an isomorphism for any
object of C.
Definition. We say that two categories C and D are isomorphic if and only if there are
functors : C D and : D C such that = D and = C .

Definition. We say that two categories C and D are equivalent if and only if there are
functors : C D and : D C such that is isomorphic to D and is
isomorphic to C .
Proposition. Two categories C and D are equivalent if and only if there exists a functor
: C D such that
: C (, ) D ((), ())

is a bijection for any , objects of C (that is, is full and faithful) and, for any object
of D, there exists an object of C such that and () are isomorphic.

Let C be a category and S be the category of the sets (that is, the category whose objects
are the sets and, for any sets 1 , 2 , (1 , 2 ) is the set of the functions from 1 to 2 ).
Let be an object of C. Let : C S be the covariant functor defined by
() = C (, )

Categories

21

for any object of C and sending an arrow : to the arrow


C (, ) C (, )

given by the composition with . Let : C S be the controvariant functor defined


by
() = C (, )

for any object of C and sending an arrow : to the arrow


C (, ) C (, )

given by the composition with . Sometimes and are denoted respectively by


(, ) and ( , ).

Definition. We say that a covariant functor : C S is representable if it is isomorphic to for some object of C (in this case, we say that represents ).
We say that a controvariant functor : C S is representable if it is isomorphic to
for some object of C (in this case, we say that represents ).
We can prove that an object representing a functor is unique up to isomorphism.

Definition. We say that a category C is additive if it satisfies the following three axioms:
C (, ) is an Abelian group for any objects and of C and the composition
of arrows is bi-additive, i.e.,

( + ) = + ,

( + ) = +

for all morphisms , , , .


Zero object. There exists an object 0 of C such that C (0, 0) is the zero group.
Direct sum. For any two objects 1 , 2 in C there is an object in C and arrows
: ,

: for = 1, 2

such that = for = 1, 2, 2 1 = 1 2 = 0 and


1 1 + 2 2 = .

Definition. Let C and D be two additive categories. We say that a functor from C to D
is additive if, for any , , objects of C, the map
: C (, ) D ((), ())

is a homomorphism of Abelian groups.

Let be the category of Abelian groups, that is, the category whose objects are the
Abelian groups and, for any Abelian groups 1 , 2 , (1 , 2 ) is the set of the homomorphisms from 1 to 2 . Let C be an additive category and let : be an
arrow in C. We define a controvariant functor
: C ,

22 | Chern classes
in the following way: for any object of C, we define

( )() = (C (, ) C (, )),

where C (, ) C (, ) is the map given by the composition with , and,


for any : arrow in C, let ( )() be the morphism ( )() ( )(),
i.e.,
(C (, ) C (, )) (C (, ) C (, )),

given by the composition with . If the functor is represented by an object ,


one can prove easily that there is a morphism : such that = 0. The
morphism , or the pair (, ), or , is called the kernel of .
The cokernel of a morphism : is a morphism : such that, for all
object in C, the following sequence (where the maps are induced by and ) is an
exact sequence of groups:
0 (, ) (, ) (, ).

Again, sometimes we call a cokernel or .

Definition. We say that a category C is Abelian if it is additive and it satisfies the following axiom:
Ker and Coker. For any arrow : in C, there exists a sequence

such that
(i) = ;
(ii) is the kernel of , is the cokernel of ;
(iii) is the kernel of and the cokernel of .

Chern classes. ([45], [93], [96], [107], [135], [146], [147], [181],[188]). Let be a holomorphic vector bundle of rank on a complex compact manifold . The Chern classes
() 2 (, )

for = 1, . . . , and the total Chern class

() = 1 + 1 () + + () =0,..., 2 (, )

(where 1 0 (, ) ) are defined by the following three axioms:

Axiom 1: Normalization. if = 1, 1 is the map 1 (, O ) 2 (, ) induced by the


exponential sequence on ,
0 O O 0

Chern classes

23

(the first map is given by the inclusion and the second by 2 ; see Exponential
sequence).

Axiom 2: Multiplicativity. if 0 0 is an exact sequence of vector


bundles on , then
() = ()(),

where the product at the second member is the cup product, see Singular homology
and cohomology.
Axiom 3: Functoriality. if : is a continuous map and a vector bundle on ,
then
( ) = ().
We define the Chern polynomial of to be

()() = 1 + 1 () + + () .

LerayHirsch theorem. Let : () be the projectivized bundle of . Let be


the tautological bundle on (), i.e., the subbundle of whose fibre on a point of
() is the line represented by this point (see Tautological (or universal) bundle)
()


()


/

We have that ((), ) is a free (, )-module generated by 1, , . . . , 1 , where


= 1 (1 ).

Theorem. If : () is the projectivized bundle of and = 1 (1 ), where is


the tautological bundle on (), we have
+ ( ()) = 0 .

The theorem above is sometimes used (once 0 and 1 are defined) to define the Chern
classes (to be the unique elements satisfying the equation above).
Some useful formulas about Chern classes are:
(i) For any
( ) = (1) ().

(ii) Let and be holomorphic vector bundles on . Let


()() = (1 + ),

()() = (1 + )

24 | Chern classes
be the formal factorizations of the Chern polynomials of and . Then
From (ii) we easily get that

( )() = (1 + ( + )).
,

1 ( ) = ()1 () + ()1 ().

Moreover, if () = and () = 1,

( ) = (
=0,...,

) ()1 () .

In particular, if is a bundle on the projective space and () = , then


1 (()) = 1 () + ,

where () = , where is the hyperplane bundle (see Hyperplane bundles,


twisting sheaves).

The definition of Chern classes can be given, more generally, for complex bundles on
compact manifolds in a way analogous to the definition above (the only difference
being in the normalization axiom). We skip it, but we mention other ways to define
Chern classes.
If is a complex vector bundle on a compact -manifold and is a connection on
(see Connections), we can define for any
() = ( (

)) ,
2

where is the curvature of , is the trace, and is the imaginary unit.

Another way is the following: if the rank of is , we define


() to be the Poincar dual of the zero locus of a general section of ;
1 () to be the Poincar dual of the zero locus of 1 2 where 1 , 2 are general
section of ;
more generally, () to be the Poincar dual of the zero locus of 1 +1 ,
where 1 , . . . , +1 are general sections of .
Finally we want to mention the definition of Chern character.
Let
()() = (1 + )

be the formal factorization of the Chern polynomial ()() of a complex bundle ;


the are called the Chern roots. Observe that, if splits, i.e.,
= 1

CohenMacaulay, Gorenstein, (arithmetically -,-)

25

with the line bundles, the are the 1 ( ). We define the Chern character () =
() of by
() = exp( ) = (1 + +
=1,...,

One can easily see that


() = + 1 () +

We have that

=1,...,

2 3
+
+ ).
2!
3!

1
1
[ ()2 22 ()] + [1 ()3 31 ()2 () + 33 ()] + .
2 1
6
( ) = () + (),
( ) = ()().

Chows group. See Equivalence, algebraic, rational, linear -, Chow, NeronSeveri

and Picard groups.

Chows theorem.

([93], [103], [196]). Any analytic subvariety in is algebraic,


i.e., it is the zero locus of a finite number of homogeneous polynomials (see Varieties
and subvarieties, analytic -, Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps).
See G.A.G.A.

Class group, divisor -. See Equivalence, algebraic, rational, linear -, Chow,


NeronSeveri and Picard groups.
Cliffords Index and Cliffords theorem.

and algebraic curves.

See Riemann surfaces (compact -)

CohenMacaulay, Gorenstein, (arithmetically -,-). ([20], [39], [62], [159],


[186], [187]). Let (, ) be a local Noetherian ring. We say that a finitely generated module is CohenMacaulay if
() = ()

(see Local, Noetherian, Artinian, Depth and Dimension).


A ring is Cohen-Macaulay if it is Cohen-Macaulay as module on itself.

Let be an algebraically closed field.

(1) We say that an algebraic set over is CohenMacaulay if, for any , the
local ring O, (the stalk in of the sheaf of the regular functions on ) is Cohen
Macaulay.

26 | CohenMacaulay, Gorenstein, (arithmetically -,-)


One can prove that a CohenMacaulay algebraic set is equidimensional: for instance,
the union of a line and a plane meeting in a point is not CohenMacaulay.
Hartshornes connectedness theorem states that a CohenMacaulay algebraic set
must be locally connected in codimension 1, i.e., removing a subvariety of codimension 2 cannot disconnect it.
For instance two surfaces meeting in a point in a space of dimension 4 cannot be
CohenMacaulay.
Example of CohenMacaulay algebraic sets: (locally) complete intersections.

(2) An equidimensional projective algebraic set of codimension in is said arithmetically CohenMacaulay if the minimal free resolution (see Minimal free resolutions) of the sheaf O has length (which is the minimal possible length):
0 F F1 F1 O O 0.

Equivalently, if we define = [0 , . . . , ] and = () is the ideal associated to ,


the minimal free resolution of the projective coordinate ring /, as module over , has
length :
0 1 1 / 0.

This is equivalent to the condition

(/) = ,

where the stands for projective dimension and the projective dimension of an module is the minimal length of a projective resolution of (a projective resolution
of is an exact sequence
0 0 0

with projective -modules; see Injective and projective modules and Injective
and projective resolutions; is the length of the resolution).
The condition (/) = is obviously equivalent to the condition
that is,

(/) = () = () (/),
() (/) = (/),

which is equivalent, by the AuslanderBuchsbaum theorem (see Depth), to


(/) = (/).

Thus is arithmetically CohenMacaulay if and only if the coordinate ring /, is


CohenMacaulay as -module.

The rank of F is called the CohenMacaulay type of .

Coherent sheaves

27

One can show that, for of dimension 1, being arithmetically CohenMacaulay


is also equivalent to the condition
(I()) = 0

for 1 and for all , where I is the ideal sheaf of .

Proposition. Arithmetically CohenMacaulay CohenMacaulay.

(3) An equidimensional projective algebraic set of codimension in is said to be


arithmetically Gorenstein if it is arithmetically CohenMacaulay of type 1, i.e., the
minimal free resolution of the sheaf O has length
0 F F1 F1 O O 0

and the rank of F is 1.

Remark. (See [186].) If is arithmetically CohenMacaulay of codimension in


and
0 1 1 / 0
is a minimal free resolution of /, then

0 1 (/, ) 0

is a minimal free resolution of (/, ).


The module := (/, )( 1) is called the canonical module.

Proposition. Let be an a.C.M. algebraic subset of . We have: is arithmetically


Gorenstein / = () for some the minimal free resolution of / is selfdual
up to twist by + 1.

(4) We say that an equidimensional projective algebraic subset of of codimension is Gorestein if it is CohenMacaulay and its dualizing sheaf, that is
EXT (O , O )( 1) (see Dualizing sheaf, , EXT ) is locally free of rank 1.

Remark. Strict complete intersection arithmetically Gorenstein arithmetically


CohenMacaulay (the first implication is due to the fact that, if is a strict complete
intersection, then its minimal resolution is the Koszul complex).

Coherent sheaves. ([93], [107], [129], [146], [223]).


The most common definition of coherent sheaf is the following.

Definition. Let (, O ) be a ringed space (see Space , ringed -).


We say that a sheaf F of O -modules (see Sheaves) is of finite type if for every
point there is an open neighborhood of and a surjective morphism of
sheaves of O -modules O | F| for some .

28 | Coherent sheaves

We say that a sheaf F of O -modules is coherent if the following two properties


hold:
(i) F is of finite type,
(ii) for any open subset of , any , and any morphism of sheaves of O modules : O | F| , we have that () is of finite type.
We say that a sheaf of O -modules F is quasi-coherent if, for every point ,
there is an open neighborhood of and an exact sequence

O | O | F| 0,

where O and O are direct sum of (possibly infinite) copies of O .

Theorem. Let (, O ) be a ringed space.


(i) Let 0 F1 F2 F3 0 be an exact sequence of sheaves of O -modules. If
two of the F are coherent, so is the third.
(ii) The kernel, the image and the cokernel of a morphism of sheaves of O -modules
between two coherent sheaves are coherent.
(iii) Let F and G be two coherent sheaves on . Then F O G and HOMO (F, G) are
coherent.
In some texts (for instance [107]), the definition of coherent sheaf on a scheme (see
Schemes) is the following.
Definition. Let be a commutative ring with unity and an -module. We define F
to be the following sheaf on (): for any open subset of (), let F () be
the set of the functions
:

(where is the localization of in , see Localization, quotient ring, quotient


field) such that () for every and is locally a fraction, that is, for
all , there exists a neighborhood of in and , such that for all
, we have and ( ) = . Consider the obvious restriction maps.
Let (, O ) be a scheme and F a sheaf of O -modules. We say F is a coherent (respectively quasi-coherent) sheaf if there is an open affine covering of , { = ( )} ,
and finitely generated (respectively not necessarily finitely generated) -modules
such that F| = F .
The two definitions of coherent sheaf agree on Noetherian schemes (see, e.g., [129,
Theorem 1.13] for a proof).
By Serres G.A.G.A. theorem (see G.A.G.A.) there is a functor from the category
of schemes of finite type over to the category of analytic spaces (see Spaces
analytic -) and, given a projective scheme over , the functor induces an equivalence of categories from the category of coherent sheaves on to the category of co-

Completion

29

herent sheaves on the analytic space associated to and this equivalence maintains
the cohomology.
See also CartanSerre theorems.

Cohomology of a complex.
Cohomology, singular -.

See Complexes.

See Singular homology and cohomology.

Complete intersections. ([104], [107], [228]). We say that a projective algebraic


variety of dimension in is a strict complete intersection if there exist elements of the ideal of that generate the ideal of ; we say that is set-theoretically
complete intersection if it is the intersections of hypersurfaces.
Complete varieties.

([107], [129], [140], [197], [201], [202], [228]).

Definition. An algebraic variety is said to be complete if, for all algebraic varieties
, the projection morphism : is a closed map, i.e., the image through
of any closed subset (closed in the Zariski topology) is a closed subset.

Proposition. An algebraic variety over is complete if and only if it is compact with


the usual topology.

Proposition. A projective algebraic variety over an algebraically closed field is complete.


There exist complete nonprojective algebraic varieties; see [201].

Chows lemma. For any complete variety over an algebraically closed field, there
exists a projective algebraic variety and a surjective birational morphism from
to .

Completion. ([12], [62], [107], [185], [256]). Let (, +) be a topological Abelian group
with a sequence of subgroups, { } , such that
= 0 1

and { } is a fundamental system of neighborhoods of 0.


We say that a sequence ( ) in is a Cauchy sequence if, for any neighborhood
of 0, there exists such that
, .

We say that two Cauchy sequences, ( ) and ( ) , are equivalent if


lim ( ) = 0.

30 | Complexes
The completion of , denoted by , is defined to be the set of the equivalence classes
of the Cauchy sequences in .

The map sending to the constant sequence = for all is injective


if and only if is Hausdorff. If the map is an isomorphism, we say that is
complete. We can prove that a completion is complete.
The completion of can be also defined by using the inverse limit (see Limits, Direct
and inverse): we can define
:= lim / .

Example. Let = and = ( ) for every , where is a prime number. Take


{ } as fundamental system of neighborhoods of 0; the topology we get is called
-adic and the ring is called ring of the -adic integers.
More generally, let be a ring, be an ideal and define = for any (define
0 = ) . The topology we get on by taking { } as a fundamental system of
neighborhoods of 0 is called -adic.

In algebraic geometry one often meets completions of local rings (, ) (see Local);
in this case we consider { } as fundamental system of neighborhoods of 0. Precisely, for any algebraic variety and any , the completion of the local ring
(O, , ) gives information about the local behavior of around and the study of
the completions of the local rings (O, , ) for is linked to the study of singularities of .
If and are two algebraic varieties over a field and and , we say that
in and in are analytically isomorphic if and only if the completions of O,
and O, are isomorphic as -algebras.

Cohens theorem. Let (, ) be a complete regular local Noetherian ring of dimension


(see Noetherian, Artinian and Dimension) containing some field and let be its
residue field, i.e., /. Then is isomorphic to the ring of the formal powers series in
variables over , usually denoted by [[1 , . . . , ]].
More generally, if (, ) is a complete local Noetherian ring containing some field and
is its residue field, then is isomorphic to [[1 , . . . , ]]/ for some and some
ideal .

In particular, if is a smooth point of an algebraic variety of dimension over a


field , then the completion of O, is isomorphic to [[1 , . . . , ]].
See Regular rings, smooth points, singular points.

Complexes.

Let be a ring. A complex of -modules, which is usually written

/ 1

/ +1

+1

/ ,

is the datum of a sequence of -modules and -homomorphisms : +1


such that +1 = 0 for any .

Cone, tangent -

31

The cohomology of the complex above in degree , usually denoted by ( ),


is defined as
( : +1 )
( ) :=
.
(1 : 1 )
We call it homology, instead of cohomology, if the indices of the -modules are
decreasing (instead of increasing).
A morphism from a complex of -modules

/ 1

to another complex of -modules

/ 1

/
/

/ +1

/ +1

+1

+1

/
/

is the datum of a sequence of -homomorphism : such that the following


diagram commutes

/ 1


/ 1


/

/ +1


+1

/ +1

+1

+1

/
/ .

We can see easily that it induces homomorphisms from ( ) to ( ) for any .


We say that two morphisms = ( ) and = ( ) from to are homotopically
equivalent if there are -homomorphisms : 1 such that

= +1 + 1 .
1

/
/ +1
|
|
|
|
|
|
|
|
||
||

||
|| +1
|
|
|
|
 ~||

 ~||
/ 1
/
/ +1

/ 1

+1

+1

/
/ .

Homotopically equivalent morphisms induce the same homomorphisms in cohomology.


In an analogous way we can define complexes of sheaves, etc.
See also Exact sequences.

Cone, tangent -.

([104], [228]). Let be an affine algebraic variety over an algebraically closed field and let = () be the ideal of (see Varieties, algebraic -,
Zariski topology, regular and rational functions, morphisms and rational maps). Let

32 | Connections
. Choose coordinates such that = 0. For every , let () be the sum of the
monomials of of the lowest degree. Let
() := {()| }.

The tangent cone to at = 0 is defined to be the zero locus of ().

Example. Let be generated by 2 2 + 3 . The ideal () is generated by 2 2 ; see


Figure 2.

Fig. 2. The tangent cone.

Connections.

([46], [93], [135], [146], [188], [251]). Let be a vector bundle on


a manifold . A connection on is a map
: () ( )

(where, for any bundle , () denotes the set of sections of and denotes
the dual of the tangent bundle) such that
(1 + 2 ) = 1 + 2
() = +

1 , 2 (),
(), : .

Remark. Let 1 , . . . , be a frame for on an open subset of , i.e., let 1 , . . . , be


sections of | such that, for all , {1 (), . . . , ()} is a basis of the fibre . We
define the connection matrix of with respect to 1 , . . . , in the following way:
= ,

(the entries of are 1-forms). Then

=1,...,

( ) = +

= + , = ( + , ),

which can be written for short as

+ ,

Connections

where

33

1
.
= ( .. ) .

Let () := ( ). Given a connection on , : 0 () 1 (), we can define


a map, which is usually called again ,
: () +1 ()

in the following way:

( ) = +

for any (), ( ). By composing : 0 () 1 () with : 1 ()


2 () we get a map
: 0 () 2 (),

called curvature of the connection. It satisfies

() = ()

for all () and for all maps : . If is the connection matrix of with
respect to a frame on an open subset of , we have that is given by the matrix of
2-forms
= + ,

where denotes the usual matrix product where the entries are multiplied by the
wedge product.

Remark. Suppose {1 , . . . , } and {1 , . . . , } are two frames on an open subset of


for the vector bundle and, if = , then
1
( ... ) = (

1
.. ) .
.

Then, if and are the connection matrices of with respect the two frames and
and are the matrices of the curvatures, we have that
= 1 + 1 ,

= 1 .

Proposition. Let be a connection on a vector bundle on and be its curvature.


Let , . Then
, = [,] ,

34 | Connections
where the subscript in an operator means that, after applying the operator, we evaluate the result in the vector indicated in the subscript and [, ] means .
Connections on tensor products and dual bundles

Remark. If and are two vector bundles and and are two connections respectively on and , we can define a connection on the tensor product in
the following way:
( )( ) = +
for any (), ( ).

Remark. If is a connection on a vector bundle , there exists a connection on the


dual bundle such that
(, ) = (, ) + (, )

for any (), ( ) (where (, ) means that we are applying the element of
( ) to the element of ()).

Bianchis identity. Let be a connection on a vector bundle . Let be its curvature;


it can be seen as an element of 2 ( ). Then
= 0

(by the remarks above, defines a connection on and thus also on = ()


and thus we have a map, we call again , from 2 ( ) to 3 ( )).
Compatibility with holomorphic structures and metrics

Definition. Let (, (, )) be a complex vector bundle with a Hermitian metric. A connection on is said to be compatible with the metric if
for all 1 , 2 ().

(1 , 2 ) = (1 , 2 ) + (1 , 2 )

Definition. Let be a complex manifold and be a holomorphic vector bundle. A


connection on is said to be compatible with the holomorphic structure if =
where is the projection
( 1,0 ) ( 0,1 ) ( (0,1 ) )

(i.e., the entries of the connection matrix are of type (1, 0)).

Proposition. Let be a complex manifold and be a holomorphic vector bundle with


a Hermitian metric. There is a unique connection on compatible both with the holomorphic structure and with the metric.

Covering projections

35

Correspondences. ([72], [93], [196]). Let and be two algebraic varieties; a


correspondence from to is an algebraic cycle in (see Cycles).
Let , , be three smooth algebraic varieties and let , be the projection from
onto and analogously , and , . If 1 is a correspondence from to
and 2 is a correspondence from to , we define

1 ,
2 ),
2 1 = , (,

where is the intersection of cycles (see Pull-back and push-forward of cycles,


Intersection of cycles).
One can prove that, if is smooth, then the set of the correspondences from to
with the product and the usual sum of cycles is an associative ring.

Covering projections.

([33], [91], [112], [158], [184], [215], [234], [247]).

Definition. Let and be two topological spaces. We say that a map


:

is a topological covering projection (covering projection for short) of if, for all
, there exists an open subset of such that 1 () is a disjoint union of open
subsets of such that | : is a homeomorphism for all . The space is
said to be covering space.

Definitions.
We say that a map between two topological spaces, : , is a local homeomorphism if, for all , there exists an open subset of containing
such that () is an open subset of and : () is a homeomorphism.
Let , , be topological spaces. Let : and : be continuous
maps. A lifting of (for ) is a continuous map : such that = .

?
~~
~

~~
~~

/

We say that a continuous map : is complete with respect to a topological space , if, for every continuous map : [0, 1] , every lifting of
|{0} can be extended to a lifting of .
We say that a continuous map : has unique path lifting if, given paths

and in (that is two continuous maps from [0, 1] to ) such that (0) = (0)
and = , then = .
We say that a topological space is locally path-connected if the path-connected
components of open subsets are open.

36 | Covering projections
We say that a topological space is semi-locally simply connected if every
has a neighborhood such that all loops in are homotopically trivial
in .

Fig. 3. A covering projection.

Theorem. Let and be topological spaces. A covering projection : is a


local homeomorphism and is complete with respect to any topological space.

Theorem. Let be a locally path-connected and semi-local simply connected topological space and let be a locally path-connected topological space. Let :
be a continuous map complete with respect to any topological space and with unique
path lifting. Then : is a covering projection.
Definition. Let and be two topological spaces and let and be points respectively of and . A pointed covering projection : ( , ) (, ) is a covering
projection : such that ( ) = .

Proposition . Let : ( , ) (, ) be a pointed covering projection. Let be a


connected topological space, and : (, ) (, ) be a continuous map.
(a) If there exists a lifting : (, ) ( , ) of , it is unique.
(b) If is locally path-connected, there exists a lifting : (, ) ( , ) of if and
only if
(1 (, )) (1 ( , )),

where we denote by 1 the first fundamental group (see Fundamental group,


also for the definition of and ).

Proposition. Let : ( , ) (, ) be a pointed covering projection. The maps


: ( , ) (, )

are injective for every and are isomorphisms for 2.

Cremona transformations

37

Theorem. Let be a path-connected, locally path-connected, semi-locally simply connected topological space. Let . There is a bijection between the following sets:
and

{pointed covering maps on (, )}/ pointed covering homeomorphisms


{subgroups of 1 (, )},

where a pointed covering homeomorphism between two pointed covering projections


on (, ), : ( , ) (, ) and : ( , ) (, ), is a homeomorphism
: such that ( ) = and = . The bijection is given by associating to a pointed covering projection on (, ), : ( , ) (, ), the subgroup
(1 ( , )) of 1 (, ).

Definition. We say that a covering projection : with path-connected, is


universal if the first fundamental group of is trivial.
Let : ( , ) (, ) be a pointed covering projection with path-connected. Let
= (1 ( , )).

Obviously the fibre 1 () is in bijection with the set of lateral classes of in 1 (, ):


{| 1 (, )}.

In fact, if 1 (), we can send to , where is the image through of a path


from to .
We have an action, called monodromy action, of 1 (, ) on 1 (): if is a point of
= ; we define to be
1 () and 1 (, ), let be the lifting of such that (0)

(1).
Observe that, if is a point of such that ( ) = and is the image through
of a path from to , then (1 ( , )) = 1 . Thus, in the assumptions of the
theorem above, there exists a pointed covering homeomorphism between ( , ) and
( , ) if and only if 1 = .
In particular, if is normal, the group of covering homeomorphisms of on is
transitive on 1 () and is isomorphic to 1 (, )/.
Note. In the case of two complex manifolds, (ramified) covering projections sometimes stands for holomorphic surjective maps between complex manifolds of the same
dimension. When it is a true topological covering projection, i.e., it is not ramified, we
say that it is an tale covering projection.

See also Riemanns existence theorem

Cremona transformations. See Quadratic transformations, Cremona transfor-

mations.

38 | Cross ratio

Cross ratio.

Let be a field and let 1 = 1 . The cross ratio of an ordered set of

four points of , (1 , 2 , 3 , 4 ), with 1 , 2 , 3 distinct, is the element [ 1 ] 1 such


1

that there exists an automorphism : 1 1 such that

1
1
0
((1 ), (2 ), (3 ), (4 )) = ([ ] , [ ] , [ ] , [ 1 ]) ,
2
1
0
1

or, using not homogeneous coordinates,

((1 ), (2 ), (3 ), (4 )) = (, 0, 1, ),

where = 2 / 1 . It is equal to

simple ratio(1 , 2 , 3 )
,
simple ratio(1 , 2 , 4 )

where, if = [ ] and := / , we define simple ratio(1 , 2 , 3 ) =

precisely, the cross ratio is equal to


det (
det (

1
1
1
1

) det ( 2
3
2
4

) det ( 2
4
2

4
)
4
3
)
3

3 1
.
3 2

More

Since the cross ratio of (1 , 2 , 3 , 4 ) and the cross ratio of ((1) , (2) , (3) , (4) ) are
the same for any composition of two disjoint transpositions, we have that the possible values of the cross ratio of ((1) , (2) , (3) , (4) ) for 4 are at most 6, precisely,
if is the cross ratio of (1 , 2 , 3 , 4 ), the 6 possible values are

Cup product.
Curves.
Cusps.

1
,

1 ,

1
,
1

1
,

1
.
1

See Singular homology and cohomology.

See Riemann surfaces (compact -) and algebraic curves.


See Regular rings, smooth points, singular points.

Cycles. Let be an algebraic variety. An algebraic cycle of codimension in is an


element of the free Abelian group generated by the closed (for the Zariski topology)
irreducible subsets of of codimension (see Varieties, algebraic -, Zariski topology,
regular and rational functions, morphisms and rational maps); in other words an
algebraic cycle of codimension in is an element of the form

Deformations

39

with irreducible algebraic subset of of codimension and .


Analogously for analytic cycles.
See Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard
groups, Pull-back and push-forward of cycles.

D
Deformations.

([43], [151], [153], [154], [160], [161], [218], [221]).


Deformation theory is strictly connected with the theory of moduli spaces (see Moduli spaces), i.e., varieties parametrizing geometric objects of a certain kind. It was
created to parametrize the possible complex structures on a fixed differentiable manifold. We recall that if : is a holomorphic surjective map between complex
manifolds such that the differential of at every point has maximal rank and the fibres of are compact complex manifolds, then the fibres are diffeomorphic by Ehresmanns theorem (see [58] or [149, Theorem. 2.3]), and so parametrizes some complex
structures on the same differentiable manifold.

Definition. Let be a compact complex manifold. A deformation of is the datum


of a proper flat morphism : of complex analytic spaces, a point and an
isomorphism 1 () . We denote this by : (, ).

(See Proper, Flat (module, morphism), Spaces, analytic - for the definitions of
these terms.)

Let : (, ) be a deformation and let : ( , ) (, ) a morphism of analytic


spaces. The pull-back : ( , ) of the deformation : (, ) through
is defined in the following way: let
= = {(, ) | () = ( )}

and let : be the projection onto the second factor.

We say that two deformations of , : (, ) and : ( , ), are isomorphic if there exist isomorphisms of analytic spaces : and : ( , ) (, )
such that = and the composition of from 1 ( ) to 1 () with the isomorphisms with is the identity.

Definition. We say that a deformation : (, ) of is complete if any other


deformation of , : ( , ), is locally the pull-back of : (, ), i.e.,
there exist neighborhood of in and : such that the deformation |
(in the obvious sense) is isomorphic to the pull-back of : (, ) through .
We say that : (, ) is universal if it is complete and the map is locally unique.
We say that : (, ) is semiuniversal if it is complete and the map (the
differential of in ) is unique.

40 | Deformations
Observe that all the universal deformations of a complex manifold are locally canonically isomorphic, and all the semiuniversal deformations of are locally isomorphic.
Obviously, from the moduli problem viewpoint, the best situation is the one of universal deformations.
Definition. Let : (, ) be a deformation of . Let 1,0 be the holomorphic
tangent space of at and let = O(1,0 ), where 1,0 is the holomorphic tangent
bundle of . The KodairaSpencer map associated to : (, ) is the map
: 1,0 1 ( )

defined as follows: let be a neighborhood of and let { } be a finite open covering


of 1 such that, defined = , there are isomorphisms : such
that the composition of with the projection onto the second factor is ; let 1,0 ;
for all there exists a unique () 1,0 orthogonal (through ) to 1,0
and such that ( ()) = . The family
, :=

defines an element of 1 ( ), which we call ().

Observe that the trivial deformation (i.e., ) has zero KodairaSpencer map.

Remark. If (, ) and ( , ) are two deformations of with Kodaira


Spencer maps respectively and and the second deformation is the pull-back of
the first through a map : ( , ) (, ), then
= .

Theorem. For every complex compact manifold , there exists a deformation


(, ), called Kuranishi family, with the following properties:
(1) its KodairaSpencer map is bijective;
(2) it is a semiuniversal deformation of ; furthermore, and it is complete in every
point of , if we consider it as a deformation of the fibre on ;
(3) if 0 ( ) = 0, then it is a universal deformation of ;
(4) is the zero locus of a holomorphic map from a neighborhood of 0 in 1 ( ) to
2 ( ); in particular () 1 ( ) 2 ( ) and, if 2 ( ) = 0, then is
smooth.
From the remark and from statement (1) of the theorem we have that a complete deformation must have surjective KodairaSpencer map. Furthermore, if a complete deformation has bijective KodairaSpencer map, then it is semiuniversal.
We considered deformations of compact complex manifolds, but in an analogous way
we can consider deformations of other objects: schemes, bundles . . . . In [218], Schlessinger developed a theory to which many deformation theories can be related.

Degree of an algebraic subset

41

Let be a local Noetherian complete ring with residue field and let be the category of the local Artinian -algebras with residue field . Schlessingers theory studies
the functors
:

(see Categories for the definition of functor) such that () is a set with only one
element.
The link between this theory and deformation theory is the following: consider for
instance deformations of schemes; consider the functor : such that
() is the set of isomorphisms classes of deformations on () of a scheme
and, if : is a homomorphism, () is the map from the set of isomorphisms
classes of deformations on () of to the set of isomorphisms classes of deformations on () of associating to a deformation on () its pull-back through
: () (). If we fix and (), the pull-back gives a morphism
of functors (, ) ; obviously its surjectivity, that is the surjectivity of the maps
(, ) () for , corresponds to the completeness of the deformation
() and the bijectivity to the universality; so the universality is related with
representability of the functor .

Degeneracy locus of a morphism of vector bundles.


varieties.

See Determinantal

Degree of an algebraic subset. ([93], [104], [107], [196]). Let be an algebraically closed field.
Let be an algebraic subset of dimension in
. The degree of is defined to be !
times the leading coefficient of the Hilbert polynomial (see Hilbert function and
Hilbert polynomial).
Equivalently, one can define the degree of an algebraic variety of dimension in

to be the number of intersection points of with a generic ( )-dimensional subspace (i.e., a subspace of complementary dimension) and the degree of an algebraic
subset of dimension in
to be the sum of the degrees of its irreducible components
of dimension .
If is the zero locus of a homogeneous polynomial , we have that the degree of is
the degree of .

If = , the degree of a -dimensional smooth algebraic variety in


is its fundamental class in 2 (
,
)

(see
Singular
homology
and
cohomology).
More

,
)
is
,
where

is
the
fundamental
class
of
a
-subspace
of

precisely 2 (

; so
the fundamental class of is for some ; is the degree of .
The degree of is also , where is the Fubini-Study form on
(see FubiniStudy metric).
See also Bezouts Theorem and Minimal degree.

42 | Depth

Depth. ([62], [159], [185]). Let be a Noetherian ring and let be a finitely generated
-module. If is an ideal of such that = , we define the -depth (sometimes
called -grade) of to be the number of the elements of a maximal -regular sequence in (see Regular sequences). If is also local and is its maximal ideal, we
call depth of the -depth of .
AuslanderBuchsbaum theorem. Let be a Noetherian local ring. For any finitely generated -module with finite projective dimension, we have that
() + () = (),

where is the projective dimension (see Dimension).

See also CohenMacaulay, Gorenstein, (arithmetically -,-).

Del Pezzo surfaces.


De Rhams theorem.

See Surfaces, algebraic -.


([85], [93], [251]).

De Rhams theorem. Let be a real manifold. Let


() denote the so-called
de Rham cohomology, which is defined to be the quotient

{closed -forms on }
,
{exact -forms on }

where a -form is said to be closed if = 0 and is said to be exact if there exists


a ( 1)-form such that = . Let (, ) denote the singular cohomology (see
Singular homology and cohomology). Then we have

() .
(, )

For de Rhams abstract theorem, see Sheaves.

Derived categories and derived functors.

([5], [79], [108], [242]).

Definition. Let A be an Abelian category (see Categories). The homotopy category


of A, denoted by (A), is the following category:
the objects are the complexes of objects of A;
the morphisms are homotopy equivalence classes of morphisms of complexes
(See Complexes for the definition of homotopically equivalent morphisms of
complexes).
We denote by (A) the subcategory of (A) whose objects are the bounded
complexes, by + (A) the subcategory whose objects are the complexes bounded
below, and by (A) the subcategory whose objects are the complexes bounded
above.

Derived categories and derived functors

43

We say that a morphism in (A), where is one among 0, , +, , is a quasi-isomorphism if it induces an isomorphism in cohomology.
Let be a complex
1 +1 ;

we denote by () the complex such that () = +1 for any and () = .


We call shift operator.
Let : be a morphism in (A), where is one among 0, , +, . The cone of
, denoted by (), is the complex () with the differential
(

) ( ()

0
).
)(

Proposition. Let A be an Abelian category. If : and : are two


morphisms in (A) and is a quasi-isomorphism, then there exist a complex in A
and two morphisms : and : with quasi-isomorphism such that the
following diagram commutes:

? `A
 AAAA

AA

A

_?
> .
??
}}
}
??
}
}}
??
}}

Definition. Let A be an Abelian category. The derived category of A, denoted by (A)


is the following category:
the objects of (A) are the complexes of objects of A;
a morphism : in (A) is a triplet (, , ), where is a third complex,
: and : are homotopy equivalence classes of morphisms of
complexes and is a quasi-isomorphism. We write it in the following way:
_@
@@
@@
@@

?.





By the proposition above one can define the composition of two morphisms
_@
@@
@@
@@

?





44 | Derived categories and derived functors


and
_?
??
??
??

}>
}
}
}}
}}

in (A) in the following way: by the proposition above there exist , , with
quasi-isomorphism such that the following diagram is commutative:
_?
??
??
??

? _?
 ???

?

??


_?
??
??
??

?





;
|>
|
||
||
|
|

let the composition be given by the triplet , , . We denote by (A) the


subcategory of (A) whose objects are the bounded complexes, and analogously
+ (A) and (A).

Let : be the following morphism in (A) (where is one among 0, , +,


and so throughout the item):
_?
??
??
??

.
~>
~~
~
~~
~~

We define the cone () to be the cone of .

Definition. We define the localizing functor from the homotopy category to the
derived category
A : (A) (A)

in the following way: it is the identity on the set of the objects and, for any morphism
in (A), we define A () to be
_@
@@
@@
@@

?


.



Observe that A () is an isomorphism in (A) for every quasi-isomorphism in


(A).

In fact, the idea of derived category is to identify an object of an Abelian category A


with all its resolutions; to do this we consider a category, (A), whose objects are

Derived categories and derived functors

45

all the complexes of objects in A and the morphisms are defined in such way that
two quasi-isomorphic complexes are isomorphic in (A). In such a way, any object
of A, considered as an element of (A) (that is, 0 0) is isomorphic to all its
resolutions.
Definition. Let C be an additive category and let be an additive automorphism of C
(we call shift operator). A triangle in C is a sextuple (, , , , , ) of objects , ,
in C and morphisms : , : , : (). It is often denoted

().

A morphism of triangles is a commutative diagram


/


/

/ ()

()


/ ( ) .

Definition. We say that an additive category C equipped with an additive automorphism and with a family of triangles, called distinguished triangles, is a triangulated category if the following axioms hold:
(1) Every triangle isomorphic to a distinguished triangle is a distinguished triangle.
For every morphism : , there is a distinguished triangle (, , , , , ).
The triangle (, , 0, , 0, 0) is a distinguished triangle.
(2) A triangle (, , , , , ) is distinguished if and only if the triangle
(, , (), , , ())

is distinguished.
(3) Given two distinguished triangles (, , , , , ) and ( , , , , , ) and
morphism : and : commuting with and , there exists
a morphism : such that (, , ) is a morphism from the first triangle to
the second.
(4) Given distinguished triangles
(, , , , , ),
(, , , , , ),

(, , , , , ),

there exist morphisms : , : , such that


( , , , , , ())

is a distinguished triangle and ( , , ) and (, , ) are morphisms of triangles


(i.e., = , = , = , = ).

46 | Derived categories and derived functors


Definition. We say that a functor between two triangulated categories is a -functor
if it is additive, if it commutes with the shift operators, and if it takes distinguished
triangles to distinguished triangles.
A distinguished triangle in (A) is defined to be a triangle isomorphic to a triangle
of the form

() (),

where and are the natural maps () and () ().


Analogously we define the distinguished triangles in (A). With these families of
distinguished triangles, (A) and (A) are triangulated categories.
Definition. Let A and B be Abelian categories and let
: (A) (B)

be a -functor. A right derived functor of is a -functor


: (A) (B)

together with a morphism of functors from (A) to (B)


: B A

with the following universal property: if

is a -functor and

: (A) (B)

: B A

is a morphism of functors, then there exists a unique morphism : such


that
= ( A ) .
Analogously, a left derived functor of is a -functor
: (A) (B)

together with a morphism of functors from (A) to (B)


: A B

with an analogous universal property.

If exists, it is unique up to isomorphism of functors.

Theorem. Let A and B be two Abelian categories and : (A) (B) a -functor.
Suppose there exists a triangulated subcategory of (A) such that

Determinantal varieties

47

(i) for every object of (A), there exists a quasi-isomorphism from it to an object
of ;
(ii) for every exact object of (i.e., () = 0 for any ), we have that () is also
exact.
Then there exists a right derived functor of and, if is an object of (A) and
an object of and they are quasi-isomorphic, then () and ( ) are isomorphic
in (B).

Let be an additive functor between two Abelian categories. If it is a left exact functor
(that is, it takes any exact sequence 0 to an exact sequence 0
() () ()), and exists, then we define
:= ()

and we call it the classical -th right derived functor for . Analogously, if is a right
exact functor, we can define the classical -th left derived functor for , .
For instance, let A be an Abelian category such that every bounded complex of objects
in A admits a quasi-isomorphism to a bounded complex of injective objects (we say
that an object in an Abelian category is injective if, for any morphism :
and any monomorhism : , there exists a morphism : such that
= ), e.g., we can take A equal to the category of -modules for some commutative
ring with unity . Let be an object of A. Let be the subcategory of (A) given by
the complexes of of injective objects. Consider the left exact functor = (, )
from (A) to () (where is the category of Abelian groups). Then there exists
the derived functor (, ) and
(, ) (, ).

Let A be the category of -modules for some commutative ring with unity and let
be an -module. The classical -th left derived functor of the right exact functor
is (, ).
See Ext, EXT and Tor, TOR.

Determinantal varieties. ([15], [77], [104], [106], [209]). Let be an algebraic


variety (or a manifold) and and be two vector bundles on and let : be
a morphism of vector bundles. For any , the set
() = { | ( : ) }

is said to be a determinantal variety (or the -degeneracy locus of ).

Example. Take =
and = =1,..., O( ), = =1,..., O( ). Then is given by a
matrix whose entry , is a polynomial of degree if and 0 if <
and () is the zero locus in of the minors ( + 1) ( + 1) of .

48 | Dimension
For instance, take = 1 for some and . Let = O and = O(1) . Call the
coordinates in , for = 1, . . . , , = 1, . . . , . Let be the matrix such that , = , .
We have that 1 () is the image of the Segre embedding (see Segre embedding)
1 1 ,
(. . . , , . . .), (. . . , , . . .) (. . . , , . . .).

Furthermore, () is the -secant variety to (see [104]).

Dimension.

([12], [104], [107], [159], [164], [185], [228]). We define the dimension of
a topological space to be the supremum of the set of all integers such that there
exists a chain 0 1 of distinct irreducible closed subsets of .
The (Krull) dimension of a ring is defined to be the supremum of the set of all
integers such that there exists a chain 0 1 of distinct prime ideals
of .
The dimension of an algebraic variety is its dimension as topological space (with
the Zariski topology).
Let be an algebraically closed field. By using Hilberts Nullstellensatz (see Hilberts
Nullstellensatz), we can prove easily that the dimension of an affine algebraic variety
over is the dimension of the affine coordinate ring and the dimension of a projective
algebraic variety over is the dimension of the homogeneous coordinate ring minus 1.

Theorem. The dimension of an integral domain that is a finitely generated algebra


over a field is the transcendence degree over of the quotient field of (see Transcendence degree and Localization, quotient ring, quotient field).

In particular, the dimension of the affine coordinate ring of an irreducible affine algebraic variety over a field is equal to the transcendence degree over of its quotient
field.
Proposition. The degree of the Hilbert polynomial (see Hilbert function and Hilbert
polynomial) of a projective algebraic variety is the dimension of the projective
variety.
The (Krull) dimension of a nonzero -module is defined to be the Krull dimension of
/(),
where () = { | = 0}.

We say that an -module has finite projective dimension if there exists a projective resolution of (see Injective and projective modules,Injective and projective
resolutions) of the form
0 0 0

Discrete valuation rings

49

(the number is said to be the length of the resolution). In this case, the projective
dimension of the -module , denoted by (), is the minimum of the lengths of
such resolutions.

If = [0 , . . . , ] for some field , then () is the minimum of the length of a


free resolution (which is the length of a minimal free resolution), since any finitely
generated projective [0 , . . . , ]-module is free (see [159, Chapter IV, Theorem 3.15]).

See also Length of a module.

Direct and inverse image sheaves.

([93], [107]). Let : be a continuous map between two topological spaces. Let F be a sheaf on (see Sheaves).
The direct image sheaf (push-forward) F is the sheaf on
F(1 ()),

for any open subset of .

The -direct image sheaf (F) (or (F)) is the sheaf on associated to the
presheaf
(1 (), F|1 () ),

for any open subset of .


We can prove that it is the -th right derived functor of the left exact functor (see
Derived categories and derived functors).
The sheaf 1 G of a sheaf G on is the sheaf on associated to the presheaf

lim

open, ()

G()

for any open subset of (see Limits, direct and inverse -).
Moreover, if (, O ), (, O ) are ringed spaces (see Spaces, ringed -) and G is a sheaf
of O -modules on , then we define
G = 1 G 1 O O ;

this is called inverse image sheaf by . One can prove easily that, if F is a sheaf of
O -modules, then
O ( G, F) O (G, F).

Discrete valuation rings. ([12], [73], [107]). Let be an integral domain that is
not a field. We say that it is a discrete valuation ring if it is Noetherian (see Noetherian,
Artinian), local (see Local) and the maximal ideal is principal.
We can prove that these three conditions together are equivalent to the following condition: there exists an irreducible element such that every {0} can be
written uniquely in the form
= ,
(1)

50 | Divisors
where is a unit in and . The element is a generator of and it is called
uniformizing parameter.
If is a discrete valuation ring, we can define a map
: {0}

by sending to , with as in (1); we can prove that the map does not depend on the
choice of the uniformizing parameter ; the map is called valuation map of . If we
denote by the quotient field of (see Localization, quotient ring, quotient field),
we can extend to a map
: {0}

by sending {0} to the unique such that = , where is a unit in ;


we have that
= {0} { {0}| () 0},

= {0} { {0}| () > 0}.

We can prove that a point in an algebraic curve over an algebraically closed field
is smooth (see Regular rings, smooth points, singular points) if and only if O, is a
discrete valuation ring.

Divisors.

([72], [74], [93], [107], [129], [228], [241]). Let be an algebraically closed
field and let be an algebraic variety over .

A Cartier divisor on is given by an open covering { } of and not identically


zero rational functions on such that, for any and , the function / on
is a nowhere vanishing regular function. Other data ( , ) give the same Cartier
divisor if / is a nowhere vanishing regular function on for any , . In other
words, a Cartier divisor is a global section of the sheaf that is the quotient of the sheaf
of the not identically vanishing rational functions and the sheaf of the nowhere vanishing regular functions.
(Analogously a Cartier divisor on a complex manifold is a global section of the sheaf
M /O , where M is the sheaf of the not identically vanishing meromorphic functions
on and O is the sheaf of nowhere vanishing holomorphic functions on ).
In some books Cartier divisors are called locally principal divisors. The support of a
Cartier divisor on given by {( , )} is the set of the points such that, if ,
then in is either zero or not regular.
We can define the sum of two Cartier divisors on by multiplying their local equations; so we get an Abelian group structure on the set of Cartier divisors on .
A Weil divisor on a finite formal sum =1,..., where is a subvariety of of
codimension 1 and for any . Its support is the set s.t. =0 .
Also the set of the Weil divisors on forms an Abelian group in an obvious way.

Divisors

51

For any subvariety of of codimension 1, let O, be the set of the pairs (, ), where
is an open subset of such that = 0 and is a regular function on , up to
the following equivalence relation: (, ) is equivalent to ( , ) if = on .

Definition. For any subvariety on of codimension 1 and any element of O, we


define
() = O, (O, /()),
where O, (O, /()) denotes the length of the O, -module O, /(), (see
Length of a module).
If is smooth along , we can write as for some unit of O, , a generator
of the maximal ideal of O, , an integer and, equivalently, we can define ()
to be .
For any not identically zero rational function on , we define
() = () (),

where , O, and = /.

Theorem. There is a group homomorphism


: {Cartier divisors on } {Weil divisors on }

such that
(1) if is normal, then is injective;
(2) if is smooth, then is bijective.
The map can be described as follows. For any Cartier divisor on and any subvariety on of codimension 1, we define () to be ( ), where is a local
equation of on any affine open subset such that = 0. For any Cartier divisor
, we define () to be the Weil divisor
() ,

where the sum is over all the subvarieties of of codimension 1 with () = 0


(we can prove that only a finite number of subvarieties of codimension 1 are such
that () = 0).

Definition. A Weil divisor =1,..., is said to be effective if and only if 0 for


any .
A Cartier divisor is said to be effective if it is defined by regular local equations
.

Definition. Let be a not identically vanishing rational function on . Obviously it defines a Cartier divisor, we denote by () or simply by (), by taking any open covering
and the restriction of to any open set of the covering.

52 | Dolbeaults theorem
The function defines also a Weil divisor, we denote by () or simply by (): we
define
() = (),

where the sum is over all the codimension 1 subvarieties with () = 0.


A divisor of the form () for some rational function is called principal.
Obviously (() ) = () .

Definition. Two Cartier divisors, respectively Weil divisors, 1 and 2 on are said
linearly equivalent if there exists a rational function on such that 1 2 = () ,
respectively 1 2 = () .
The group of Weil divisors of up to linear equivalence is called divisor class group
of and denoted by ().
The group of Cartier divisors of up to linear equivalence is called Picard group of
and denoted by ().
Theorem. The map induces a homomorphism : () () and is injective
if is normal, it is bijective if is smooth.

Definition. We can associate to any Cartier divisor a line bundle we denote by ()


in the following way: if the Cartier divisor is given by local data ( , ), we define
() to be the line bundle whose transition functions , from to are / , see
Bundles, fibre -.
Theorem. The map () induces an isomorphism from () to the group of the
isomorphism classes of line bundles on .

Definition. We say that a Cartier divisor is ample, or very ample, or big, or nef, or
b.p.f. if the corresponding line bundle () is. See Bundles, fibre -.

See Linear systems and Equivalence, algebraic, rational, linear -, Chow, Neron
Severi and Picard groups.

Dolbeaults theorem.

([93], [251]). Let be a complex manifold. Let be the


,
sheaf of the holomorphic -forms and let () be

{-closed (, )-forms on }
{-exact (, )-forms on }

where a (, )-form is said to be -closed if = 0 and is said to be -exact if there


exists a (, 1)-form such that = . Then
(, )

(),

Dualizing sheaf

53

Dominant.

See Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps.

Dual variety and biduality theorem. ([59], [60], [78], [143], [162]). Let be a
projective algebraic variety in := . Let ( ) be the closure of the set of all
hyperplanes tangent to (we say that a hyperplane is tangent to if there exists a
smooth point in such that contains the tangent space of at ). It is called the
dual variety of . In most cases it is a hypersurface.

Theorem. (Biduality theorem) Let be a projective algebraic variety in . Then


( ) = .
See [143], [60] for the case of fields different from .

Dualizing sheaf.

([107], [224]). We follow strictly [107].


The dualizing sheaf on a singular algebraic variety is a sheaf that has, in some sense,
the same role of the canonical sheaf on a nonsingular algebraic variety (see Canonical
bundle, canonical sheaf).
Let be an algebraically closed field. For the canonical sheaf on the projective space
we have the following theorem.

Theorem. (Serre duality for ). Let be the canonical sheaf on := .


We have that ( , ) and, for any coherent sheaf F on (see Coherent
sheaves), the natural bilinear form
(F, ) ( , F) ( , )

is a perfect pairing from finite-dimensional vector spaces over to (let , be vector


spaces over a field ; a bilinear form : is said to be a perfect pairing
if the induced maps and are isomorphisms). Furthermore, for all
0, there is a natural isomorphism
(F, ) = ( , F) ,

which, for = 0, is the one induced by the bilinear form above (see Ext, EXT for the
definition of ).

Definition. Let be a proper scheme over of dimension . A dualizing sheaf for


0
0
is defined to be a coherent sheaf
on and a linear map : (,
) such
that the composition

0
0
(F,
) (, F) (,
) ,

where the first map is the natural bilinear form, induces an isomorphism
0
(F,
) (, F) .

54 | Dynkin diagrams
One can prove that, if a dualizing sheaf exists, then it is unique.
If is projective scheme, then a dualizing sheaf exists: if and its codimension
is , we can prove that the dualizing sheaf is the sheaf

EXT (O , ).

Theorem. If is a CohenMacaulay and equidimensional projective scheme of dimension over , we have that the natural maps
0
) (, F)
(F,

are isomorphisms for any coherent sheaf F and any 0.

Corollary. (Serre duality). If is a CohenMacaulay and equidimensional projective


scheme of dimension over , we have a natural isomorphism
0
)
(, F) (, F

for any locally free sheaf F of finite rank on .

In fact,
for any sheaf G of O -modules, (O , G) (, G) for any 0;
for any sheaves of O -modules, G, G , L with L locally free of finite rank, we have
that (G L, G ) (G, L G );
(see Ext, EXT ), so

0
0
0
) (O , F
) (F,
) (, F).
(, F

Corollary. (Serre duality). If is a smooth projective algebraic variety of dimension


over , we have a natural isomorphism
(, F) (, F )

for any locally free sheaf F of finite rank on (see Zariski tangent space, differential
forms, tangent bundle, normal bundle for the definition of ).

(It follows from the previous corollary and from the fact that ( ) ). See
Serre duality.

If is a locally complete intersection of codimension in a projective space :=


we can prove that
0
(I/I2 )) ,

where I is the ideal sheaf of . Finally, if is a smooth projective algebraic variety,


0
then the dualizing sheaf
coincides with the canonical sheaf .

Dynkin diagrams.

See Lie algebras.

Elliptic Riemann surfaces, elliptic curves

55

E
Effective.

See Divisors.

Elliptic Riemann surfaces, elliptic curves. ([93], [107], [127], [144], [148],
[165]). A compact Riemann surface (see Riemann surfaces (compact -) and algebraic curves) is said to be elliptic if its genus is 1. One can easily prove that it is
isomorphic to its Jacobian (see Jacobians of compact Riemann surfaces), thus it is
isomorphic to a torus
/,
where is a lattice of rank 2 in , that is, there exists a basis { 1 , 2 } of over such
that = 1 + 2 (see Figure 4).

Fig. 4. Elliptic Riemann surface.

Up to isomorphisms, we can suppose that


= 1,

with the imaginary part of , (), greater than 0 , where 1, means 1 + (see
Figure 5).
Observe that a holomorphic map : / 1 / 2 is, up to translation, equal to a map
for some such that 1 2 , in fact: can be lifted to a map :
(see Covering projections) and
is 1 -invariant, thus it defines a holomorphic

function / 1 , which must be constant (since / 1 is compact); therefore is


given by the multiplication by a constant.
Hence two tori /1, 1 and /1, 2 with (1 ), (2 ) > 0 are isomorphic if and
only if
+
1 = 2
2 +

56 | Elliptic Riemann surfaces, elliptic curves

1
Fig. 5. Generators of the lattice.

with ( ) (2, ). In fact, let : /1, 1 /1, 2 be an isomorphism; then


is the multiplication by a constant with 1, 1 1, 2 ; so 1 = 2 + ,

1 = 2 + for some , , , ; since is an isomorphism, the matrix ( ) must
be invertible over and, since preserves the orientation, the determinant must be
1.
Therefore the set of the elliptic Riemann surfaces up to isomorphisms is parametrized
by
H/(2, ),
where H = { | () > 0} (Siegel upper half-plane) and (
on H by
+

.
+

) (2, ) acts

Let be an algebraic closed field. A complete (see Complete varieties) smooth algebraic curve is said to be elliptic if its genus is 1.
Observe that, if , then the line bundle (3) embeds into 2 and the image
is a smooth cubic curve (see again Riemann surfaces (compact -) and algebraic
curves). Moreover, any smooth cubic curve in 2 is an elliptic curve (see Genus,
arithmetic, geometric, real, virtual -). If the characteristic of is 0, a smooth cubic
curve can be written in the Weierstrass form
2 = ( )( )

for some {0, 1} (see Weierstrass form of cubic curves) and two curves,
2 = ( )( ),

2 = ( )( ),

are isomorphic if and only if there is an automorphism of 1 such that


({0, 1, , }) = {0, 1, , },

1
, 1 , 1
} (see Cross ratio). We can prove that
thus if and only if { 1 , 1 , 1

this is true if and only if () = ( ), where

() = 28

(2 + 1)3
.
2 ( 1)2

Elliptic Riemann surfaces, elliptic curves

57

+
Fig. 6. The group law on an elliptic curve.

Thus, through the function , the set of the elliptic curves over , up to isomorphisms,
is parametrized by .
Let be an elliptic complete smooth curve over and let 0 . Observe that the
map
0 ()

sending to the class of the divisor 0 is a bijection (see Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard groups for the definition
of 0 ()); in fact, by using RiemannRoch theorem, we can prove that, for any divisor on of degree 0, there exists such that is linearly equivalent to 0
(apply RiemannRoch theorem to +0 ). So inherits a group structure from 0 ().
In particular it is an Abelian variety (of dimension 1), see Tori, complex - and Abelian
varieties.
As we have already said, the line bundle (30 ) embeds into 2 and the image is a
smooth cubic curve, we call again . The smooth cubic curve inherits a group structure from 0 (). We want to illustrate how we can read the group structure inherited from 0 () on the cubic curve. Observe that
=

=1,...,

=1,...,

(where is the sum on inherited from 0 ()) if and only if there is the following
linear equivalence of divisors:
0 0 ,

=1,...,

=1,...,

(where denotes linear equivalence), and obviously this holds if and only if
+ ( )0 .

=1,...,

=1,...,

In particular 0 is the zero element of the group. Observe also that three points , ,
in embedded in 2 are collinear if and only if the divisor ++ is linearly equivalent

58 | Elliptic surfaces
to 30 . So, if , , are collinear, then for the sum we have defined on we have that
+ + = 0.
Now take coordinates on 2 such that 0 = [0 : 1 : 0]. Observe that any vertical line
= meets the cubic in 0 and in two points, call them , ; then, for the sum we
have defined on , we have that + = 0.
Thus, if , are two points on the cubic, then is the third point of intersection of
the cubic with the line and + is the third point of intersection of the cubic with
the line passing through and 0 (which will be a vertical line); see Figure 6.

Elliptic surfaces.

See Surfaces, algebraic -.

Embedded components. See Primary ideals, primary decompositions, embed-

ded ideals.

Embedding. An embedding (between groups, rings, varieties . . . ) is a injective map

that is an isomorphism onto its image.


We can prove (see [228, Chapter 2, 5]) that a morphism between two algebraic varieties : is an embedding if and only if it is injective and the differential in any
point is an isomorphic embedding of the Zariski tangent space of at into
the Zariski tangent space of at () (see Zariski tangent space, differential forms,
tangent bundle, normal bundle).

Enriques surfaces.

See Surfaces, algebraic -.

Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and


Picard groups. ([16], [72], [74], [93], [107], [166], [217]). Let be an algebraic variety

of dimension over an algebraically closed field . Throughout this item, with the
word cycle we will mean an algebraic cycle (see Cycles).

Notation. Let be a smooth algebraic curve. If is a cycle in and , we denote


by the push-forward through the projection from onto of the intersection
cycle ( ) when this is defined (see Pull-back and push-forward of cycles and
Intersection of cycles).
Definition. Two Weil divisors 1 and 2 on are said linearly equivalent if there
exists a rational map on such that 1 2 = (), where () is the Weil divisor on
associated to , and analogously for Cartier divisors (see Divisors).
Two cycles of codimension in , 1 and 2 , are said rational equivalent if there
are a finite number of ( 1)-codimensional subvarieties in , 1 , . . . , , and, for any
= 1, . . . , , a not identically vanishing rational function on such that
1 2 = ( ).
=1,...,

Equivalence, algebraic, rational, linear -

59

Another (equivalent) definition of rational equivalence is the following: two cycles


of codimension in , 1 and 2 , are said rational equivalent if there is a codimensional cycle of 1 such that
1 2 = 0 .

Two cycles of codimension in , 1 and 2 , are said algebraic equivalent if there


are a smooth algebraic curve , a cycle of codimension in and two points in
, 1 and 2 , such that
1 2 = 1 2 .

. .
. .
. .
.
.
1 2

Fig. 7. Algebraic equivalence.

Remark. Obviously rational equivalence implies algebraic equivalence.


For divisors, linear equivalence and rational equivalence coincide (it is clear from the
first definition of rational equivalence).
Definition. The -th Chow group () of is defined to be the group of the cycles
of codimension in up to rational equivalence.
The -th NeronSeveri group () of is defined to be the group of the cycles of
codimension in up to algebraic equivalence.
We write (), respectively (), to denote the group of the cycles of dimension
in up to rational equivalence, respectively algebraic equivalence.
The group of Weil divisors of up to linear equivalence is called divisor class group
of and denoted by ().
The group of Cartier divisors of up to linear equivalence is called Picard group of
and denoted by ().
Theorem. There is a homomorphism () () that is injective if is normal
and that is bijective if is smooth (see Divisors for the description of this homomorphism).

Definition. We can associate to any Cartier divisor a line bundle we denote by ()


in the following way: if the Cartier divisor is given by local data ( , ), we define
() to be the line bundle whose transition functions , from to are / , see
Bundles, fibre -.

60 | Equivalence, algebraic, rational, linear Theorem. The map associating to a Cartier divisor the line bundle () induces an
isomorphism from the Picard group () to the group of the isomorphism classes of
line bundles on . The group of the isomorphism classes of line bundles on is isomorphic also to 1 (, O ), via the isomorphism induced by the map sending a bundle
to its transition functions.

Suppose from now on that is a smooth projective algebraic variety over .


In this case, one can prove that, for any divisor , the first Chern class (see Chern
classes) of the line bundle associated to is the Poincar dual of the class of (see
Singular homology and cohomology for the definition of Poincar duality).

Let 0 () denote the group of the isomorphism classes of the holomorphic line bundles on with first Chern class 1 () equal to 0. Since the first Chern class of a
line bundle associated to a divisor is the Poincar dual of the class of , the group
0 () can be seen also as the group of divisors on that are homologous to 0, up to
linear equivalence.
Consider the exponential sequence

0 O O 0

(see Exponential sequence); by taking the cohomology we get


1

0 1 (, O)/1 (, ) 1 (, O )
2 (, )
2 (, O)

where 1 is the first Chern class. Thus

0 () =

(2)

1 (, O)
1 (, )

(since 0 () = (1 )). So the group 0 () is the first intermediate Jacobian 1 of


(see Jacobians, Weil and Griffiths intermediate -).
We can prove that 0 () is an Abelian variety and thus an algebraic variety, and, by
using this result, one can easily prove:

Proposition. Two divisors on a smooth projective algebraic variety over are algebraic
equivalent if and only if they are homologous.

Therefore, 1 (), which is defined to be the set of divisors up to algebraic equivalence, is also the set of divisors up to homological equivalence and thus it is equal to
1 (, O )/(1 ) (1 ) and the sequence (2) becomes
0 0 () () 1 () 0.

Observe that 1 () (1 ) = () and that () is 2 (, ) 1,1 () (in fact


2 (, O) 0,2 () and the (2, 0)-part of a real form, and in particular an integer form,
is conjugate of the (0, 2)-part); thus we get:

Exponential sequence

61

Lefschetz theorem on (1, 1)-classes. Let be a smooth projective algebraic variety


over . Every element of 2 (, ) 1,1 () is the class of a divisor on .

See Albanese varieties, Jacobians of compact Riemann surfaces, Jacobians, Weil


and Griffiths intermediate -.

Euler sequence. ([93], [207]). Let be a complex vector space of dimension + 1


and let = (). Consider the trivial bundle on whose total space is ; by a
slight abuse of notation (that is, by using for the bundle the notation of the associated
sheaf) we denote it O or simply O . Let be the subbundle of O whose
total space is
{(, ) | }

with the obvious map to ( is called universal bundle, see Tautological (or universal) bundle).
The Euler sequence on is the following exact sequence:

0 O 0,

where is the inclusion map, is the quotient bundle and the projection.
We can easily prove that is O(1) (see Hyperplane bundles, twisting sheaves) and
is 1,0 (1), where 1,0 is the holomorphic tangent bundle of . So the Euler
sequence is

0 O(1) O 1,0 (1) 0.

Exact sequences.

We say that a complex of modules (see Complexes)


2

+1

1 +1

is an exact sequence if +1 = for all .

We have that a complex of the kind 0 is exact if and only if is injective.

We have that a complex of the kind 0 is exact if and only if is surjective.


A resolution of a module is an exact sequence of modules of the kind
or of the kind

1 0 0

0 0 1

(see Sheaves for the definition of exact sequences of sheaves).

Exponential sequence. ([93], [107]). Let be a complex manifold. The exponential sequence is the exact sequence of sheaves on

0 O O 0,

62 | Ext, EXT
where
is the sheaf associating to any open subset of the set of the continuous
functions from to the group endowed with the discrete topology;
O is the sheaf associating to any open subset the group of the holomorphic functions on (with the sum as group operation) ;
O is the sheaf associating to any open subset the group of the holomorphic
functions on that vanish nowhere (with the product as group operation);
the map is given by the inclusion and the map is the map sending any
O() to 2 O () .

Ext, EXT . ([62], [79], [84], [93], [107], [116]). Let be a commutative ring with unity.
Let and be -modules and let
1 0 0

be a projective resolution of (see Injective and projective resolutions); we denote


by the complex
1

1 0 0.

We can consider the complex ( , ):


0

0 (0 , ) (1 , ) ( , ) ,

where, for every , the map is given by the composition with the map . For every
, we define
(, ) = ( ( , )),

i.e.,

One can prove that, if

(, ) = ( )/(1 ).

0 0 1

is an injective resolution of and is the complex

0 0 1 ,

then (, ) is equal to ( (, )). We can easily prove that (, )


does not depend on the choice of the resolutions and that
0 (, ) = (, ).

We will write (, ) instead of (, ) for simplicity.


Proposition. For any exact sequence of -modules

0 0,

Ext, EXT

63

there are exact sequences


0 (, ) (, ) (, )
1 (, ) 1 (, ) 1 (, )
2 (, ) 2 (, ) 2 (, )
,

0 (, ) (, ) (, )
1 (, ) 1 (, ) 1 (, )
2 (, ) 2 (, ) 2 (, )
.

More synthetically, (, ) can be defined as the -th classical right derived functor
of the left exact functor (, ) (see Derived categories and derived functors).

Let (, O ) be a ringed space (see Spaces, ringed -). Let M() be the category of
sheaves of O -modules (see Sheaves).
Let F and G be in M(). Define (F, G) to be the group of O -module homomorphisms and HOM(F, G) to be the sheaf associated to the presheaf
O| (F| , G| ).

For F M() we define (F, ) to be the -th right derived functor of the functor
(F, ) and we define EXT (F, ) to be the -th right derived functor of the functor
HOM(F, ).
Properties. Let F be in M(). We have

if = 0,
0 if > 0;
(b) (O , F) (, F) 0.
(c) If 0 H G F 0 is an exact sequence in M(), then, for any T in M(),
we have an exact sequence
(a) EXT (O , F) = {

0 (F, T ) (G, T ) (H, T )


1 (F, T ) 1 (G, T ) 1 (H, T )

and analogously for EXT .


(d) Let F, G and L be in M(). Let L be locally free of finite rank and let L =
HOM(L, O ). Then
(F L, G) (F, L G)
and

EXT (F L, G) EXT (F , L G) EXT (F, G) L .

(f) If is a Noetherian scheme and F, G M() with F coherent (see Schemes and
Coherent sheaves), then, for any and any 0, we have:

EXT (F, G) O (F , G ).

64 | Fano varieties
(g) Let be a projective scheme over a Noetherian ring and let L, F and G be in
M(); let L be a very ample locally free sheaf of rank 1 and F and G be coherent.
Then there is an integer 0 > 0 depending on F, G and such that for all 0 we
have
(F, G L ) 0 (, EXT (F, G L )) .

More generally the relation between and EXT is given by a spectral sequence,
see [84].
See also Tor, TOR.

F
Fano varieties. ([68], [131], [191]). A Fano (or anticanonical) variety is a smooth
complete algebraic variety such that 1 is ample (where is the canonical bundle,
see Complete varieties, Canonical bundle, canonical sheaf).
Examples of Fano varieties are the projective spaces and the Del Pezzo surfaces (see
Surfaces, algebraic -).
An instrument to classify Fano varieties is the so-called index: the index of a Fano
variety is defined to be
:= max{ | 1 = for some line bundle }.

Fibred product.

If : and : are two maps, then the fibred product


of the sets and with respect to and is defined to be the set

:= {(, ) | () = ( )}.

Five Lemma. ([62], [79], [91], [208], [234]). Let 1 , . . . , 5 , 1 , . . . , 5 be modules over
a ring and let


1

/ 2


/ 2

/ 3


/ 3

/ 4


/ 4

/ 5


/ 5

be a commutative diagram of morphisms of -modules. If the rows are exact sequences, 1 is surjective, 5 is injective and 2 and 4 are isomorphisms, then 3 is
an isomorphism.

Flag varieties. ([36], [126], [181]). Flag varieties parametrize the chains of projective
subspaces in a projective space. Precisely, for any field and 1 , . . . , , , we

Flat (module, morphism)

65

define the flag variety (1 , . . . , , , ) in the following way:

(1 , . . . , , , ) = {(1 , . . . , ) (1 , ) ( , )| 1 },

where (, ) denotes the Grassmannian of projective -planes in .

Theorem. (1 , . . . , , , ) is a rational, homogeneous (hence smooth) variety (see


Homogeneous varieties, Rational varieties) of dimension
(+1 )( + 1)

=1,...,

where we set +1 = .

Flat (module, morphism).

([12], [62], [107], [116], [185]). Let be a ring. We say


that an -module is flat if the functor

is exact, that is, if it changes every exact sequence into an exact sequence.
Proposition. We have that is flat if and only if every exact sequence of -modules
implies the exact sequence

0 1 2 3 0

(3)

0 1 2 3 0 .

Observe that, in general, the exact sequence (3) implies only the exact sequence

1 2 3 0,

thus the functor is only right-exact. For instance, let and be positive
natural numbers and consider the exact sequence of -modules

0 / 0,

where the map we have indicated by is the multiplication by the number ; by


tensoring by /, we get

0 /(, ) / / /(, ) 0.

We say that a morphism of varieties (or schemes) : is flat if, for all ,
such that () = , we have that O, is a flat O, -module via the map induced
by f, : O, O, , where O, is the stalk in of the sheaf of the regular functions
on and analogously O, .
See Tor, Grauerts semicontinuity theorem.

66 | Flexes

Flexes. ([93], [107], [228], [246]). Let be an algebraic closed field of characteristic 0.

Definition. A point of a plane algebraic curve over is said to be a flex if it is a


smooth point and (, ()) 3, where () is the tangent line of at and
(, ()) is defined as follows:
In an affine neighborhood of , write () = { + | } for some vector ; let
be a polynomial whose zero locus is ; consider the polynomial, in , | () ; define
(, ()) to be the multiplicity of 0 as zero of | () .

Fig. 8. Flex.

Proposition. Let be a projective algebraic curve in 2 and let be a homogeneous


polynomial in 0 , 1 , 2 such that is the zero locus of . A smooth point of is a flex
if and only if it is in the zero locus of the hessian of , where the hessian of is the
determinant of the matrix
2
(
)
,{0,1,2}

(observe that the partial derivatives of a polynomial can be defined over any field by
applying the usual rules for derivatives).
In particular, if is smooth and of degree 3, then has at least one flex (see Bezouts theorem).

Fubini-Study metric.

([93]). Let : +1 {0} := be the standard


projection. Let : {0} be a holomorphic map on an open subset of
such that is the identity; define
+1

:=

log(||2 ).
2

One can show that does not depend on the choice of the map and in this way
we define a form on all . We can prove that is closed, of type (1, 1), positive
and its class in cohomology is the Poincar dual of a hyperplane (see Positive and
Singular homology and cohomology). The form is called the Fubini-Study form.
We define the Fubini-Study metric to be the Klerian metric whose Khler form is
(see Hermitian and Khlerian metrics).

Functors.

See Categories.

Fundamental group

67

Fundamental group. ([37], [38], [91], [112], [158], [215], [234]). Let be a topological space.
A path in is a continuous map : [0, 1] .
A loop in is a path : [0, 1] such that (0) = (1).
Given two paths and in , we define their composition to be the path in such
that
(2)
if [0, 1/2],
()() = {
(2 1) if [1/2, 1].

Let 0 , 1 and let and be two paths in with (0) = (0) = 0 and (1) = (1) =
1 . We say that and are homotopic with end points held fixed if there exists a
continuous map : [0, 1] [0, 1] such that
a) (, 0) = () for any ;
b) (, 1) = () for any ;
c) (0, ) = 0 for any ;
d) (1, ) = 1 for any .
Obviously, for each , the map (, ) is a path from 0 to 1 .

Let 0 . Let ,0 be the set of loops in such that (0) = (1) = 0 . The first
fundamental group of , denoted by 1 (, 0 ), is defined as follows:
1 (, 0 ) := ,0 / homotopy with end points held fixed.

It is a group with the composition given by the composition of paths defined above.
Definition. We say that a path-connected topological space is simply connected if its
first fundamental group is trivial.
Definition. Let and be two topological spaces and let 0 and 0 . If :
is a continuous map such that (0 ) = 0 , then we define
: 1 (, 0 ) 1 (, 0 )

to be the homomorphism such that ([]) = [ ] for any ,0 .

Before stating the main theorems about the fundamental group, we need to recall the
definition of the free product of groups.
Let , for , be groups. Their free product is defined to be
( )/ ,

where ( ) is the set of words with letters in (i.e., the finite sequences of
elements of the union of the ) and is the equivalence relation satisfying
(1)

(1 . . . +1 . . . + ) (1 . . . +1 . . . + )

68 | Fundamental group
for any , , for any 1 , . . . , + , for any and where is the identity
element of , and
(1 . . . +1 . . . + ) (1 . . . ( ) +1 . . . + )

(2)

for any , , for any 1 , . . . , + , , , if and are elements of the


same group for some .
Let us denote by [1 . . . ] the equivalence class of the word (1 . . . ).
We define the product in to be the product given by the concatenation, i.e., let
[1 . . . ] [1 . . . ] = [1 . . . 1 . . . ].

With this product, the free product is a group; it contains all the as subgroups.

If R = { } is a subset of ( ), then the free product of the with the


relations R, denoted by /R, is defined to be
( )/ ,

where is the equivalence relation satisfying (1), (2), and

(3)

(1 . . . +1 . . . + ) (1 . . . +1 . . . + )

for any , , for any 1 , . . . , + and for any . If we endow it with the
product given by concatenation, it is a group.
The following theorem (due to Seifert and Van Kampen, but generally called Van Kampens theorem I) computes, under some assumptions, the fundamental group of a
space from the fundamental groups of two subspaces whose union is .

Van Kampens theorem I. Let = 1 2 be a topological space and let 0 = 1 2


with open, path-connected for = 0, 1, 2. Let 0 0 . We have
1 (, 0 ) = (1 (1 , 0 ) 1 (2 , 0 ))/1 () 2 ()1 1 (0 ,0 ) ,

where : are the inclusions and the second member can be interpreted both
as the free product of 1 (1 , 0 ) and 1 (2 , 0 ) with the relations 1 () 2 ()1 for
1 (0 , 0 ) and as the quotient of 1 (1 , 0 ) 1 (2 , 0 ) by the normal subgroup
generated by 1 () 2 ()1 for 1 (0 , 0 ).

Fig. 9. Van Kampens theorem I.

Fundamental group

69

For the following theorem, which is often called Van Kampens theorem II, see [37].
The main instrument for the proof is the so-called fundamental groupoid: instead
of considering the loops based at a fixed point up to homotopy, one can consider all
the paths up to homotopy fixing the initial and final point; this yields not a group but
a groupoid, a category where all arrows are isomorphisms; the groupoid obtained in
this way is called the fundamental groupoid of the space.
Van Kampens theorem II. Let = 1 2 be a topological space such that
is open and path-connected for = 1, 2;
1 2 = with and disjoint path-connected open subsets;
2 , and are simply connected.

Then

1 () = 1 (1 ) .

Fig. 10. Van Kampens theorem II.

Attaching cells. Let = { | || 1} and 1 = { | || = 1}.


Let be a path-connected Hausdorff topological space. Let be a topological space
obtained by attaching an -cell ( , 1 ) to by a continuous map
: 1 ,

i.e., let be the quotient of by the equivalence relation determined by the


identification of with () for every 1 . Let 0 .
If = 1, then 1 (, 0 ) = 1 (, 0 ) .
If > 1, then 1 (, 0 ) = 1 (, 0 )/ (1 ((1 ))), where : (1 ) is the
inclusion.

Theorem. Let be a topological space, and let 0 . Let 1 (, ) denotes the first
singular homology module of over (see Singular homology and cohomology).
There is a group homomorphism
: 1 (, 0 ) 1 (, )

which sends the homotopy class of to the homology class of for any loop with
(0) = 0 . If is path connected, then is surjective and its kernel is the commutator

70 | Fundamental group
subgroup

1 1 | , 1 (, 0 ) .

Definition. For any topological space , we denote by 0 () the set of the pathconnected components of .

Let be a topological space. Let [0,1] be the set of the paths in ; endow it with the
compact-open topology, i.e., the topology whose prebase is the set of the subsets
[, ] := { : [0, 1] | () }

for compact subset of [0, 1] and open subset of . For any 0 , the subset
,0 is closed in [0,1] if 0 is closed in . Endow the subset ,0 with the induced
topology.

Remark. Let be a topological space and 0 . Two elements of ,0 are in the


same path-connected component of ,0 if and only if they are homotopic with end
points held fixed, i.e., 1 (, 0 ) = 0 (,0 ).
Definition. We define

(, 0 ) = 1 (,0 , )

for all 2, where is the constant path in 0 .

Theorem. (, 0 ) is commutative for all 2.

Remark. (, (0 , 0 )) = (, 0 ) (, 0 ) for any , topological spaces, 0 ,


0 and for all 1.

Remark. If a topological space is path-connected, then, for every , the group


(, 0 ) does not depend on 0 ; thus we can speak of -fundamental group ()
without specifying the point 0 .

Remark. Let and be two topological spaces and let 0 and 0 . If :


is a continuous map such that (0 ) = 0 , then there is an obvious induced map from
,0 to ,0 , so, by induction, we can define a homomorphism
for every .

: (, 0 ) (, 0 )

Theorem. Let : and : be two continuous maps between two


topological spaces such that (0 ) = (0 ) = 0 . If and are homotopic, then,
for every , the maps : (, 0 ) (, 0 ) and : (, 0 ) (, 0 ) are
equal.
In particular the fundamental groups of two homotopic topological spaces are isomorphic.

G.A.G.A.

71

Hurewicz theorem. If 2 and () is trivial for any {0, . . . , 1}, then () = 0


for any {1, . . . , 1} and () = ().
Theorem. Let : be a topological bundle and let be path-connected. Let
, = () and let be the fibre on . Then there is an exact sequence
(, ) (, ) (, ) 1 (, ) 0 () 1.

In particular, if the fibre is discrete, i.e., the bundle is a covering space, we get that
(, ) and (, ) are isomorphic for any 2.

Examples.
1 (1 ) = , 1 ( ) = 0 if 2.
If is the bouquet of circles, then 1 ( ) = ( times).
1 (1 ) = .
1 ( ) = /2 if 2.
1 ( ) is trivial for all .
Let be the topological torus with holes, i.e., the topological space obtained
by attaching to a bouquet of 2 circles ( 1 ,. . . , 2 ) a 4-agon with the law
1
1 , +1 , 1
1 , +1 , . . . . We have

1
1 1
1 ( ) = 1 2 / 1 +1 1
1 +1 2 2 .

( ) = 0 if < , ( ) = .

G
G.A.G.A.

([93], [107], [222], [228], [241]). G.A.G.A. is the abbreviation of the title of
Serres paper [222] Gomtrie algbrique et gomtrie analytique. Roughly speaking, the G.A.G.A. principle can be formulated as follows. Global analytic things on a
projective algebraic variety over are algebraic; for instance an analytic subvariety
of is an algebraic subvariety (see Chows theorem), a meromorphic function on
is rational, a holomorphic vector bundle is an algebraic vector bundle. Precisely,
there is a functor from the category of schemes of finite type over to the category
of analytic spaces (see Spaces, analytic -) and Serre proved the following theorem
and obtained some of the results above (some of them were already known) as consequences of it.
Theorem. For any projective scheme over , the functor induces an equivalence
of categories from the category of coherent sheaves on (see Coherent sheaves)
to the category of coherent sheaves on the analytic space associated to and this
equivalence maintains the cohomology.

72 | GaussBonnetHopf theorem

GaussBonnetHopf theorem.

of dimension . We have

([93]). Let be a compact complex manifold

() = (),

where () is the Euler characteristic (1) (, ) and () is the -th Chern class
of the holomorphic tangent bundle 1,0 . More precisely, here, by (), we mean
() evaluated in the fundamental class of , i.e., in the element of 2 (, ) giving
the orientation of (see Chern classes, Singular homology and cohomology).

General type, of -.

See Kodaira dimension (or Kodaira number).

Genus, arithmetic, geometric, real, virtual -. ([93], [107], [118], [196], [241]).
Let be an algebraically closed field.
Let be a compact complex manifold or a projective algebraic variety over . The
arithmetic genus () is defined by
() = (1) ((, O ) 1).

where O is the sheaf of the holomorphic, respectively regular, functions on .


Let be a compact complex manifold or a smooth projective algebraic variety over .
The geometric genus () is defined by
() = 0 (, ),

where is the canonical sheaf (see Canonical bundle, canonical sheaf).

Theorem. If and are birationally equivalent smooth projective algebraic varieties over (see Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps for the definition of birational equivalent),
then () = ( ) and, if the characteristic of is 0, () = ( ) (see [118],
[107]).
If is a singular projective algebraic variety, its geometric genus is defined to be the
geometric genus of a desingularization of , that is, the geometric genus of a smooth
projective algebraic variety that is birational equivalent to , when it exists.

Proposition. If is a projective algebraic curve, then () = 1 (, O ).


Moreover, if is smooth, we have () = () = 1 (, O ) (by Serre duality; see
Serre duality).
If = and is smooth, and then, from the topological viewpoint, is a topological
torus with a certain number of holes, we have that the number of holes is equal to ()
(and ()) (Riemanns Theorem; see Riemann surfaces (compact -) and algebraic
curves).

Let be a smooth projective algebraic surface and be an algebraic curve in (we


recall that for us a curve is irreducible). By applying the RiemannRoch theorem for

Genus, arithmetic, geometric, real, virtual -

73

surfaces (see Surfaces, algebraic -) to calculate (O ()) and by using the exact
sequence 0 O () O O 0, we get that
and thus

1 (, O ) =

+
+ 1,
2

+
+1
2
(observe that, if is smooth, we can be prove that this number is also equal to the
geometric genus by adjunction formula).
Sometimes, the number
+
+1
2
is called the virtual genus of a curve on a smooth algebraic projective surface , or,
more generally, of a divisor on a smooth projective algebraic surface . If is a curve
in a smooth projective algebraic surface over , we have that the virtual genus is the
genus (the geometric one, the arithmetic one and the number of the holes) of every
smooth curve homologous to (in fact, since is homologous to we have that
+
+
+1=
+ 1, which is the genus
= and = , so
2
2
of ).
() =

Observe that, if = 2 and is an algebraic curve of degree , then the virtual genus
(and thus the arithmetic genus) is
( 1)( 2)
2

(in fact = 2 and = 3).

In some books, e.g., [93], the real genus of a projective algebraic curve is defined to
be the genus of , where is the desingularization (which is the normalization; see
Normal) of . So it is another name for the geometric genus.

Let be a smooth projective algebraic surface over , and let be an algebraic curve
in . It holds (see, e.g., [107, Chapter V, 3])
() = ()

( 1)
,
2

where is the multiplicity at the point (see Multiplicity of a curve in a surface at


a point) and the sum is over all points of , including the so-called infinitely near
points, i.e., all the points on the strict transforms obtained by blowing up the curve
in the singular points until we get a smooth curve (see Blowing-up (or -process)).
Example 1. Let = {[ : : ] 2 | 2 (2 2 ) + 4 + 4 = 0}; we have that
() =

(4 1)(4 2)
= 3.
2

74 | Genus, arithmetic, geometric, real, virtual -

, smooth curve homolgous to

Fig. 11. Real genus and virtual genus.

desingularization of

Let us calculate (). The unique singular point of is the point = [0 : 0 : 1]; its
multiplicity is 2. Let = {[ : : ] 2 | = 0} = 2, :
= {(, )| 2 2 + 4 + 4 = 0};

the blowing up of = 2, in = (0, 0) is

{((, ), [0 : 1 ])| 0 = 1 } = 2, ;

let = 1 /0 ; the strict transform 1 of is

{(, )| 1 2 + 2 + 4 2 = 0}

and 1 () 1 = { = 0} 1 is the union of the points = (0, 1) and = (0, 1),


both smooth (thus is a node; see Regular rings, smooth points, singular points).
Hence
1
() = 3 ( 1) = 2
2
(see Figure 11).
Example 2. Let = {[ : : ] 2 | 4 2 6 + 6 = 0}; we have
() =

(6 1)(6 2)
= 10.
2

Let us calculate (). The unique singular point of is the point = [0 : 1 : 0]; its
multiplicity is 4. Let = {[ : : ] 2 | = 0} = 2, :
= {(, )| 4 6 + 6 = 0};

the blowing up of = 2, in = (0, 0) is

{((, ), [0 : 1 ])| 0 = 1 } = 2, ;

Geometric invariant theory (G.I.T.)

75

let = 1 /0 ; the strict transform 1 of is

{(, )| 2 + 4 + 2 6 = 0}

and 11 () 1 = { = 0} 1 is the point = (0, 0), which has multiplicity 2; the


blowing up of 2, in = (0, 0) is
2

{((, ), [0 : 1 ])| 0 = 1 } 2, ;

let = 0 /1 ; the strict transform 2 of 1 is

{(, )| 2 + 2 + 2 6 = 0},

and 21 () 2 = { = 0} 2 is the point = (0, 0), which has multiplicity 2;


the blowing up of 2, in = (0, 0) is
3

{((, ), [0 : 1 ])| 0 = 1 } 2, ;

let = 1 /0 ; the strict transform 3 of 2 is

{(, )| 1 + 2 + 6 6 = 0}

and 31 () 3 = { = 0} 3 is the set of the smooth points = (1, 0) and = (1, 0).
Thus
1
() = 10 [ ( 1) + ( 1) + ( 1)] = 10 8 = 2.
2

Geometric invariant theory (G.I.T.). ([81], [82], [194], [199], [204]). Let be an
algebraic closed field. Let be a reductive algebraic group over (see Representations, Algebraic groups, Lie groups). Suppose acts on a scheme of finite type
over () (see Schemes). Geometric invariant theory is the theory investigating
whether the quotient / is a good quotient. More precisely:
Definition. A good quotient of by is a surjective affine -invariant morphism
: ,

with scheme of finite type over (), such that


(i) (O ) = O , where (O ) denotes the -invariant part of (O ) (see Direct
and inverse image sheaves for the definition of (O ));
(ii) the image of a -invariant closed subset is closed and the images of two disjoint
-invariant closed subsets are disjoint.

If, in addition, for all , the set 1 () is exactly one orbit, then we say that is a
geometric quotient.

76 | Globally generated
Remark. When a good quotient exists, it is unique (up to isomorphism).
From now on, suppose that and that the action of on is the restriction of
a representation : ( + 1, ).

Definition. We say that is semistable if there exists a -invariant homogeneous


nonconstant polynomial in + 1 variables such that () = 0.
We say that is stable if
(i) the dimension of the orbit of is equal to the dimension of ;
(ii) there exists a -invariant homogeneous nonconstant polynomial such that
() = 0, i.e., (), where () is the zero locus of in , and, for
every (), the orbit of is closed in ().

We denote the set of semistable points by and the set of stable points by .

Theorem. There exists a projective scheme and a morphism : which is


a good quotient of by . Moreover, there exists an open subset of such that
1 () = and | : is a geometric quotient.
Definition. An 1-parameter subgroup of an algebraic group is a nontrivial homomorphism of algebraic groups
: ,

where is the multiplicative group of the nonzero elements of the field .

Definition. Fix a 1-parameter subgroup : ; we can prove that there exists a


basis {0 , . . . , } of +1 and integers for = 0, . . . , such that
( )()( ) = ,

for any , = 0, . . . , . Let 0 , . . . , be the coordinates with respect to {0 , . . . , }.


Let = [0 : : ] . We say that is -semistable (respectively -stable) if
max{ | = 0} 0 (respectively > 0).
Theorem. A point in is semistable (respectively stable) if and only if it is -semistable
(respectively -stable) for every 1-parameter subgroup of .

Globally generated.
Gorenstein.

See Bundles, fibre -.

See Cohen-Macaulay, Gorenstein, (arithmetically -,-).

Grassmannians.

([75], [93], [104], [120], [181], [188]). Let be a vector space of


dimension over a field . For any = 1, . . . , , let (, ) denote the Grassmannian
of the -subspaces of .
If = or , the set (, ) can be endowed with a topology which makes it a compact manifold of dimension ( ) respectively over and over (for instance, if

Grassmannians

77

= ; see (, ) as ()/() ( ), where () is the group of the orthogonal


matrices ).
The map
: (, ) ( ) ,

1 , . . . , [1 ]

is injective and is called the Plcker embedding.


Let us fix a basis {1 , . . . , } of . It induces a basis on and thus homogeneous
coordinates on ( ):
1 ,..., ,

{1, . . . , },

1 < < .

For any (, ), the coordinates 1 ,..., of () are called Plucker coordinates


of . We can define 1 ,..., for all 1 , . . . , {1, . . . , }: let 1 ,..., = 0 if 1 , . . . , are
not distinct, and let
1 ,..., = ()(1) ,...,()
for every permutation on {1, . . . , } (where () is the sign of ).

Theorem. Let the characteristic of be 0. The image of the Grassmannian (, ) by


the Plcker embedding is a smooth algebraic variety defined by quadrics, precisely by
the following quadrics:
=0,..., (1) 0 ,..., ,..., ,1 ,...,1 = 0

for all 0 , . . . , , 1 , . . . , 1 .

Let = (but an analogous theory can be developed also over ; see [188]). Let
= 1 1 0

be a flag of subspaces in with ( ) = . Let (, ). For any = 1, . . . , we


define
() = min{ | = } .

Observe that 1 1 () < < () . We define the Schubert symbol of with


respect to the given flag by
() = (1 (), . . . , ()).

Now let 1 , . . . , be natural numbers with 1 1 < < . We define the Schubert cell with Schubert symbol = (1 , . . . , ), and we denote it by , as follows:
= { (, )| () = } .

78 | Grauerts semicontinuity theorem


Thus,
= { (, )| ( ) = and ( 1 ) < = 1, . . . , }.

Proposition. Let 1 , . . . , be natural numbers with 1 1 < < . We have that


=1,..., ( ) . Moreover, if we denote the closure of by , we have that
= = { (, )| ( ) = 1, . . . , }

(where means for any ) and is an algebraic subvariety of the Plcker


embedding of (, ). The Grassmannian (, ) is a CW-complex, the cells of which
are the Schubert cells.

The subset is called the Schubert cycle, with the Schubert symbol .

Please note that in some works (for instance [75], [93], [181]) another notation is used:
let
= +

for = 1, . . . , ; then

0 1 ;

the Schubert cells (1 ,..., ) are denoted by ( 1 ,..., ) (or something else instead of )
and analogously the Schubert cycles, so
(1 ,..., ) = ( 1 ,..., )

and

if + + +1

}
= { (, ) ( ) =
for 1

(1 ,..., ) = ( 1 ,..., )
.
= { (, )| ( + ) for 1 }

The dimension of ( 1 ,..., ) in terms of 1 , . . . , is ( ) =1,..., .

Theorem. The algebra ((, ), ) (with the intersection as product) is torsion free,
and it is freely generated by the Schubert cycles.

Finally, we mention that the algebra ((, ), ) can be seen as a quotient of the
algebra of symmetric polynomials in variables; see [75] and[181] (see Symmetric
polynomials for the definition of the symmetric polynomials).

Grauerts semicontinuity theorem. ([86], [107], [241]). Let be a topological


space. We recall that a map : is said to be upper semicontinuous if and only if
for every the subset { | () } is closed. This is equivalent to the following
condition: for all 0 , there exists an open subset 0 such that () (0 ) for
all .

Groebner bases

79

Grauerts semicontinuity theorem. Let : be a proper morphism of analytic


spaces (respectively a projective morphism of Noetherian schemes), and let F be a
coherent sheaf on that is flat over , i.e., F is a flat O,() -module for all ,
where O,() is the stalk in () of the structure sheaf on (see Spaces, analytic -,
Coherent sheaves, Schemes, Flat (module, morphism)).
(a) Then the map :
( , F| )

(where is the fibre of on ) is upper semicontinuous for all .


(b) If is connected and reduced, where reduced means that O() has no nilpotent
elements for any open subset (respectively is an integral scheme), and for
some the map is constant, then the sheaf (F) (see Direct and inverse
image sheaves) is locally free and the fibre on of the corresponding bundle is
( , F| ).

Groebner bases.

([50], [51], [62]). We follow mainly the exposition in [62].


Let be a field and let = [1 , . . . , ]. A monomial in is an element of of the

form 11 . . . for some 1 , . . . , .


Let be a free finitely generated -module, and let us fix a basis E = {1 , . . . , } of .

A monomial in with basis E is an element of equal to for some monomial


and some .
A term in with basis E is the product of an element of and of a monomial in
with basis E. Obviously any element of can be written as sum of terms.
A monomial order on with basis E is a total order on the monomials of such that
if is a monomial in {1} and 1 , 2 are monomials in , then
1 > 2 1 > 2 > 2 .

If 1 , 2 are monomials in , with 1 > 2 , and , {0}, then we say that 1 > 2 ,
and analogously replacing > with .
Let . We denote the greatest (with respect to >) term of by > ().
Let be a submodule of . Let > be the submodule of generated by the elements
> () for in . We say that 1 , . . . , form a Groebner basis of if they generate
and > (1 ), . . . , > ( ) generate > .

Let 1 , . . . , {0}. For every , {1, . . . , } with = , we define


, = 0 if > ( ) and > ( ) involve different elements of the basis of (i.e.,
> ( ) = and > ( ) = with = );
otherwise (i.e., if > ( ) and > ( ) involve the same element of the basis of ),
we define
> ( )
, =
,
(> ( ), > ( ))

80 | Grothendieck group
where stands for the greatest common divisor) in the obvious sense; we
define , in the following way: one can prove that we can write , , as
, , = , + ,

for some , and , , with

=1,...,

> (, , ) > (, )

for every , and there is no monomial of , in {> (1 ), . . . , > ( )}.

Buchbergers theorem. The elements 1 , . . . , form a Groebner basis for the submodule they generate if and only if , = 0 for all , .

Buchbergers algorithm. Buchbergers algorithm is useful for finding a Groebner


basis for a submodule of : let 1 , . . . , be a set of generators of ; if they are a
Groebner basis, we have finished; if not, add to them the , computed as above; if
1 , . . . , and the , are a Groebner basis, then we have finished; if not, repeat the
operation, and so on.
Groebner basis give us a method to compute syzygies (see Syzygies). Precisely the
following theorem tells us how to compute the syzygies of a Groebner basis.

Schreyers theorem. Let 1 , . . . , be a Groebner basis of the submodule of they


generate. Then, with the notation above,
, , ,

generate all the syzygies among the .


More precisely, consider on the free -module =1,..., the following monomial order: > if and only if
> ( ) > > ( )

Then the elements

or

> ( ) = > ( ) and < .

, , ,

form a Groebner basis of the submodule of the syzygies among the .

Obviously computing syzygies among a Groebner basis of a submodule of allows


us to compute syzygies among any set of generators of .

Grothendieck group. ([11], [12], [21] [72]). Let be a Noetherian ring. Let M be
the free Abelian group generated by the isomorphism classes of the finitely generated
-modules. For any finitely generated -module , denote the isomorphism class of

Group algebra

81

by []. Let S be the subgroup of M generated by the elements [1 ] [2 ] + [3 ]


for any exact sequence of finitely generated -modules 0 1 2 3 0.
We define the Grothendieck group of the ring to be the quotient
M/S .

Let be an algebraic variety. The Grothendieck group of vector bundles on , denoted by 0 (), is the quotient of the free Abelian group generated by the set of isomorphism classes of vector bundles on by the subgroup generated by the elements
[1 ] [2 ] + [3 ] for any exact sequence 0 1 2 3 0 (where [] is the
isomorphism class of ). The group 0 () with the tensor product is a ring.
The Grothendieck group of coherent sheaves on (see Coherent sheaves), denoted by 0 (), is the quotient of the free Abelian group generated by the set of isomorphism classes of coherent sheaves on by the subgroup generated by the elements [E1 ] [E2 ] + [E3 ] for any exact sequence 0 E1 E2 E3 0, where [E]
is the isomorphism class of E. We have that 0 () is a 0 ()-module with the tensor
product as product.
We say that a function from the set of isomorphism classes of coherent sheaves on
to an Abelian group is additive if (A) + (C) (B) = 0 for any exact sequence
0 A B C 0. Obviously, for any additive function from the set of isomorphism classes of coherent sheaves on to an Abelian group (for instance for the
EulerPoincar characteristic), we have that factorizes through a map from 0 () to
; analogously for bundles and 0 (). The Grothendieck groups of bundles and of coherent sheaves were introduced by Grothendieck to prove the HirzebruchRiemann
Roch theorem (see HirzebruchRiemannRoch theorem).

GrothendieckSegre theorem.

([80], [93, Chapter IV, 3], [97], [107, Chapter V], [207], [219]). Let be an algebraically closed field. Let E be a locally free sheaf
(see Sheaves) on 1 of rank . Then there exists a unique (up to permutation)
sequence of integer numbers 1 , . . . , such that
E O( )
=1,...,

(see Hyperplane bundles, twisting sheaves for the definition of O( )).

Grothendiecks vanishing theorem.

See Vanishing theorems.

Group algebra. Let be a group. The group algebra of over a field , generally
denoted by [], is the algebra whose underlying vector space is
endowed with the product given by

= .

82 | Hartogs theorem

H
Hartogs theorem.

([40], [93], [121]). Let > 1. Let be an open set in and


let be a compact subset of such that is connected. Then any holomorphic
function on extends to a holomorphic function on .

Hartshornes conjecture.

([19], [109], [209], [240]). Hartshornes conjecture


(formulated in 1974) states that if is a smooth algebraic variety of dimension in
(or, more generally, in , where is an algebraically closed field) and > 23 ,
then is a complete intersection.
In particular, the conjecture says that if 7, every smooth algebraic subvariety of
of codimension 2 is a complete intersection. One can prove that another formulation of the last statement is: if 7, every (algebraic) vector bundle of rank 2 on
is a direct sum of line bundles. This can also be conjectured for 5, while we know
that there exists an indecomposable algebraic bundle of rank 2 on 4 : the Horrocks
Mumford bundle (see HorrocksMumford bundle). Tango found an indecomposable algebraic vector bundle of rank 2 over 5 , with a field of characteristic 2.

HartshorneSerre theorem (correspondence).

([9], [87], [109], [209]).


Let be a smooth projective algebraic variety over and be a subscheme locally
complete intersection of codimension 2. Suppose that there exists a line bundle on
such that det(, ) = | (where , is the normal bundle, which in this case can
be defined to be the vector bundle associated to the sheaf (I O ) , where I is the ideal
sheaf of in ).
If 2 (O( )) = 0, then there exists a vector bundle on of rank 2 such that | =
, , and there exists 0 (O()) such that is the zero locus of .
Furthermore, if 1 (O( )) = 0 and is connected and reduced, then the vector bundle
is unique up to isomorphism.

Hermitian and Khlerian metrics. ([93], [147], [192]). We use the notation introduced in Almost complex manifolds, holomorphic maps, holomorphic tangent
bundles.
A Riemannian metric on a manifold is given by a positive definite symmetric
bilinear form
: ( ) ( )

for every , depending in a way on .


A Hermitian metric on a complex manifold is a R iemannian metric such that
( , ) = (, )

for all , ( ) and for all .

Hilbert Basis theorem

83

Let be a Hermitian metric on . The skew-symmetric form given by


: ( ) ( ) ,
(, ) := ( , ),

for any , , is called Khler form of the Hermitian metric .


Let be the -bilinear extension of to () . Observe that it is zero on 1,0
1,0 and on 0,1 0,1 . For any , define
: ( ) 1,0 to be the isomorphism given by () = ;
: ( ) 0,1 to be the isomorphism given by () = + .
Remark. Obviously (, ) = , where denotes extended by -linearity in the
first variable and by -linearity in the second variable. On ( ) ( ) we have
that
, ( , ) = , ( , ) = 2[ (, ) (, )] .

We recall that a Hermitian form on a complex vector space is a map : ,


-linear in the first variable and such that (, ) = (, ) for any , .
By the remark above, a Hermitian metric gives, for every , a Hermitian form
: 1,0 1,0 defined to be , (, ) restricted to 1,0 1,0 .
Vice versa, given, for any , a positive definite Hermitian form
: 1,0 1,0 ,

we can define = ( , ) on ( ) ( ) ; one can easily check that is


Riemannian and that (, ) = (, ).
In local coordinates, given a Hermitian metric , define , := ( ,

the Hermitian metric locally as


, ,

).

We write

and we have

= , .

We say that a Hermitian metric is a Khler metric if and only if the associated Khler
form is closed.

Hilbert Basis theorem.

([12], [164] [256]).

Theorem Hilbert basis theorem. If is a Noetherian ring, then [] is Noetherian.


Corollary. If is a Noetherian ring, then [1 , . . . , ] is Noetherian.

See Noetherian, Artinian for the definition of Noetherian ring. Since every field is
Noetherian, from the Hilbert basis theorem we get the following corollary:

84 | Hilberts Nullstellensatz
Corollary. If is a field, then [1 , . . . , ] is Noetherian.

Hilberts Nullstellensatz.

([12], [104], [256]).

Notation. For any ideal in a ring , let denote the radical ideal of , i.e.,
:= { | {0} such that } .

Notation. Let be a field.


For any [1 , . . . , ], let () denote the zero locus of in the affine space
, i.e.,
() := { | () = 0 }.

For any , let () denote the ideal of , i.e.,

() := { [1 , . . . , ]| () = 0 } .

Theorem Hilberts Nullstellensatz. Let be an algebraically closed field. We have


that
(()) =

for any ideal of [1 , . . . , ].


In particular, if () = 0, then = [1 , . . . , ] and then = [1 , . . . , ].

Corollary. Let be an algebraically closed field. There is a bijection between the set
of the algebraic subsets of and the set of the radical ideals of [1 , . . . , ].

Notation. Let be a field.


For any subset of homogeneous elements of [0 , . . . , ], we denote
() = { | () = 0 }.

For any homogeneous ideal (see Homogeneous ideals), we define () to be


the zero locus of the set of the homogeneous elements of .
For any , we denote by () the ideal generated by
{ [0 , . . . , ]| homogeneous and () = 0 };

obviously it is a homogeneous ideal.

Theorem Projective Hilberts Nullstellensatz. Let be an algebraically closed field.


Let be a homogeneous ideal of [0 , . . . , ].
We have that () = 0 if and only if there exists such that the part of degree of
[0 , . . . , ] is contained in for any .
If () = 0, then (()) = .
See also Weierstrass preparation theorem and Weierstrass division theorem.

Hilbert function and Hilbert polynomial

Hilbert function and Hilbert polynomial.

85

([12], [104], [107], [164]).

Definition. We say that a polynomial () [] is numerical if () for any


with >> 0.

Proposition. If () [] is a numerical polynomial of degree , then it is a linear


combination with integer coefficients of the polynomials () := (1)(+1)
for =
!
0, . . . , (where (0) = 1). In particular () for every .
Let be an algebraically closed field.

Definition. Let = be a graded module over [0 , . . . , ], where is its part


of degree . The Hilbert function of , we denote , is defined in the following way:
() :=

Theorem (HilbertSerre). Let be a graded finitely generated module on


[0 , . . . , ]. There exists a unique polynomial, called Hilbert polynomial of ,
() [] such that
() = ()
for >> 0, .

Let be a projective algebraic variety in :=


. Its Hilbert function and its Hilbert
polynomial are defined to be the Hilbert function and the Hilbert polynomial of its
homogeneous coordinates ring. Thus the Hilbert function is

and the Hilbert polynomial is

() =

0 (O ())
0 (I ())

() = (O ()),

where O is the sheaf of the regular functions on , I is the ideal sheaf of and
is the EulerPoincar characteristic. In fact, by Serres theorem (see CartanSerre
theorems, for >> 0, we have (I ()) = (O ()) = 0 for all > 0, thus, for >> 0,
we have
0 (O ())
0
= 0 (O ()) = (O ())
(I ())
by the sequence

0 I O O 0.

We can prove that the degree of is the dimension of , and we may define the
degree (see Degree of an algebraic subset) of to be ! times the leading coefficient
of .

86 | Hilbert syzygy theorem


Examples.
) 1, be the Veronese embedding (see
Let , : , where = (+

Veronese embedding). Let , be the Veronese variety , ( ). Its Hilbert function is


, () = 0 (O ())/0 (I, ()) = 0 (O, ()) = 0 (O )()) = (

+
),

where the second equality holds since the Veronese variety is projectively normal
(see Normal, projectively -, -normal, linearly normal). Since , () is a polynomial we have that
, () = , ().
In particular for = 1, i.e., for the rational normal curve, we get

1, () = + 1.

Let be a smooth projective curve of degree and genus . For >> 0


0 (O ()) = + 1,

by the RiemannRoch theorem (see Riemann surfaces (compact -) and algebraic


curves), since O () is a line bundle of degree on (the restriction of times
the hyperplane bundle to , i.e., times the line bundle given the embedding of
in the projective space). Thus
() = + 1.

Hilbert schemes. See Moduli spaces


Hilbert syzygy theorem.

([62], [159], [164]).

Hilbert syzygy theorem. Let be a field and = [0 , . . . , ]. Every finitely generated graded -module has a finite graded free resolution of length by finitely
generated free -modules, that is, there exists a graded exact sequence (i.e., an exact
sequence where the maps preserve the degree)
0 1 0 0,

with finitely generated free -modules.


See also Horrocks theorem.

Hironakas decomposition of birational maps. ([107], [117], [118]). Let


be a field of characteristic zero and let and be two smooth projective algebraic
varieties over . Let : be a birational map. Then there exists a morphism
: such that is the composition of a finite sequence of blow-ups along
smooth subvarieties and the birational map is a morphism.

Hirzebruch-Riemann-Roch theorem

87

In the case where and are surfaces, we have a stronger statement; see Structure
of birational maps on surfaces in Surfaces, algebraic -.
Finally, we want to mention that it is known that in dimension 3 not every birational
morphism is the composition of blow-ups along smooth subvarieties (see [229]).
See also [2].

Hirzebruch surfaces.

See Surfaces, algebraic -.

HirzebruchRiemannRoch theorem.

([13], [29], [107], [119], [146]). Let be


a compact complex manifold and let be a vector bundle on of rank . Write the
Chern polynomial (see Chern classes)
as

()() = 0 () + 1 () + + ()
()() = =1,..., (1 + ).

We define the Chern character to be

() :=
=1,...,

(where = 1 + + 12 2 + ). Furthermore, we define the Todd class by


() = =1,...,

,
1

1 2
1 4
where 1 is the series 1 + 12 + 12
720
... .
We recall also that the symbol denotes the cup product; see Singular homology and
cohomology for the definition.

HirzebruchRiemannRoch theorem. Let be a compact complex manifold of dimension and let be a holomorphic vector bundle on of rank . We have that
(O()) = () (1,0 ),

that is, (O()) (the EulerPoincar characteristic of O()) is equal to the component
in 2 (, ) of () (1,0 ) evaluated in the fundamental class of , i.e., in the
class of the 2-cycle determined by the natural orientation of .
In the case where is a compact Riemann surface of genus (see Riemann surfaces
(compact -) and algebraic curves) and is a line bundle, we get the usual Riemann
Roch theorem:
0 (, O()) 1 (, O()) = () + 1;

in fact, the Chern polynomial is ()() = 1 + 1 () and the Chern character is


() = 1 () = 1 + 1 (); in addition, 1,0 = , and thus (1,0 ) = 1 12 1 ( ).

88 | Hodge theory
The component of (1 + 1 ()) (1 12 1 ( )) in 2 () is 1 () 12 1 ( ); thus we get

1
(O()) = deg() ( ).
2
Taking to be trivial, we get that the degree of is 2 2; hence, by substituting,
we get the RiemannRoch formula for any holomorphic line bundle on a compact
Riemann surface.
Analogously, in the case where is a surface and is a holomorphic line bundle, we
get the usual RiemannRoch theorem for surfaces (see Surfaces, algebraic -).
The theorem also holds for nonsingular projective algebraic varieties over algebraically closed fields; we will not state it in this context, since it requires the generalization of some concepts, such as Chern classes and intersection theory, for such
varieties. The statement for complex nonsingular projective varieties is due to Hirzebruch, the one for compact complex manifolds to AtiyahSinger, the one for nonsingular projective varieties over algebraically closed fields to Grothendieck (see [29]).

Hodge theory. ([44], [90], [93], [245]). Let be a compact complex manifold of
(complex) dimension . Let be a Hermitian metric on and let be the associated
(1, 1)-form (see Hermitian and Khlerian metrics).

Let us denote by (,) the bundle 1,0 0,1 , where 1,0 and 0,1
are the holomorphic and the antiholomorphic tangent bundles (see Almost complex
manifolds, holomorphic maps, holomorphic tangent bundles).
For any , the metric induces a Hermitian metric on (,) , we call again ,
defined in the following way: let = 2 =1,..., in local coordinates 1 , . . . ,
around ; let be such that the for || = and || = form an orthogonal
basis in (,) and the norm of every is 2+ . For any , , let ( , ) be the
following positive definite product on the set , () of the (, )-forms on :

(, ) = ((), ())

()
.
!

Let be the adjoint operator of : , () ,+1 () with respect to ( , ), i.e., let

: , () ,1 () be the operator such that

( , ) = (, )

for all , (), ,1 ().


Let : , () , () be the so-called Laplacian operator:

:= + .

The forms s.t = 0 are called -harmonic. Let H () be the space of the

harmonic (, )-forms. Observe that = 0 if and only if = 0 and = 0, in fact


(, ) is positive definite and

( , ) = (( + ), ) = ( , ) + (, ).

Hodge theory

89

Theorem.
,
(i) The space H () is finite dimensional. Thus, we can define an orthogonal pro

jection : , () H ().

(ii) There exists an operator (called Greens operator) : , () , () such that

,
(H ()) = 0, commutes with and and, for all , (),

= + .

The above decomposition is called Hodge decomposition.


(iii) Let , (); there exists , () such that = if and only if is
,
orthogonal to H ().

Observe that the implication of (iii) is obvious and the other implication follows
,
from (ii); in fact, let H (); by (ii) we have = + = .

Corollary. Every -closed form is -homologous to a -harmonic form. Thus


H

()

() .

In fact, let be such that = 0; by Hodge decomposition, we have

= + + = + + = + ,

so we have found a harmonic form, , that is -homologous to .

Thus, by Dolbeaults theorem, H

() (, ) (see Dolbeaults theorem).

An analogous theory can be developed for a Riemannian manifold and the operator
instead of the operator (and := + instead of , where is the adjoint
operator of with respect to , where is the star operator; see Star operator).
If is a Khler manifold, we have that
,

2 = = 2 .

In particular, H () = H () = H () (where H () is the set of the (, )-forms

such that = 0).


Since = 2 , the operator preserves the bidegree; hence
moreover is real, thus

H () = += H ();
,

H () = H ().

By Hodge decomposition for , we have


() = H () and , () = H (),

where () is the set the -closed -forms (over ) modulo the set of the -exact
,
-forms (over ) and () is the set of the -closed (, )-forms modulo the -exact
(, )-forms. Thus we get the following theorem:

90 | Holomorphic
Theorem. If is a compact Khler manifold, then

(, ) = += , (),

Holomorphic.

tangent bundles.

, () = , () .

See Almost complex manifolds, holomorphic maps, holomorphic

Homogeneous bundles. ([30], [210]). Let be a -homogeneous complex algebraic variety (see Homogeneous varieties) and be a vector bundle on . We say
that is homogeneous if there is an action of on (that is, a homomorphism from
to the group of automorphisms of ) such that

for all and for all (where denotes the fibre on ).


If we write as / where is the isotropy subgroup of a point of , we can easily
prove that is homogeneous if and only if it comes from the principal bundle /
and a representation : (, ), where is the rank of (see Bundles, fibre -
and precisely principal bundles), i.e., is homogeneous if and only if
:= / ,

where is the equivalence relation such that

(, ) (, (1 ))

for any .
For a vector bundle on a homogeneous rational variety /, with simply connected semisimple group, parabolic subgroup (see Lie groups), we have that is
homogeneous if and only if for every , where : is the left
multiplication by .

Homogeneous ideals. ([73], [185], [256]). Let be a field and be an ideal


of [0 , . . . , ]. We say that is a homogeneous ideal if and only if the following
property holds: if we write an element of as sum of homogeneous polynomials,
= =1,..., , we have that for = 1, . . . , .

Proposition.
An ideal in [0 , . . . , ] is homogeneous if and only if it is generated by homogeneous polynomials.
The sum, product, intersection of homogeneous ideals are homogeneous, the radical of a homogeneous ideal is homogeneous.
A homogeneous ideal is prime if and only if, for any , [0 , . . . , ] with ,
homogeneous and such that , we have either or .

Horrocks theorem

91

Homogeneous varieties.

([28], [210]). Let be an algebraic group, respectively


a topological group. Let be an algebraic variety, respectively a manifold. We say that
acts on if there is a morphism , (, ) such that 1 = for any
, 1 (2 ) = (1 2 ) for any and for any 1 , 2 ; we say that the action
is transitive if, for any , , there exists such that = . We say that is
-homogeneous if acts transitively on it.
Remark. Every homogeneous variety is smooth.

Theorem (BorelRemmert). A homogeneous compact Khler manifold is isomorphic


to the product of a complex torus and a rational homogeneous projective algebraic
variety (see Tori, complex - and Abelian varieties, Rational varieties).
Furthermore, a rational homogeneous projective algebraic variety is isomorphic to
1 /1 /

for some simple Lie groups and parabolic subgroups (see Lie groups).

Homology, Singular -.

See Singular homology and cohomology.

Homology of a complex.

See Complexes.

Horrocks Criterion.

([17], [122], [207]). An algebraic vector bundle on is


direct sum of algebraic line bundles if and only if the intermediate cohomology of all
its twists is zero, i.e.,
( , O()()) = 0
for all and for all with 0 < < (see Hyperplane bundles, twisting sheaves
for the defintion of O()()).

HorrocksMumford bundle. ([122], [123], [124]). The HorrocksMumford bundle is an indecomposable algebraic vector bundle of rank 2 on 4 (indecomposable
means that is it not the direct sum of two algebraic line bundles). It is the only bundle
with these features we know so far.
The zero locus of a generic section of the HorrocksMumford bundle is an Abelian
surface (see Tori, complex - and Abelian varieties) of degree 10.
See also Hartshornes conjecture.
Horrocks theorem. ([122], [209]). Let be an algebraic vector bundle (or more
generally a torsion-free sheaf of O-modules) on . Then has a free resolution of
length 1, that is, there exists an exact sequence
0 1 1 0 0

with all the direct sum of algebraic line bundles on , thus direct sum of hyperplanes bundles (see Hyperplane bundles, twisting sheaves).

92 | Horseshoe lemma

Horseshoe lemma.

([208]).

Horseshoe lemma. Let be a ring and let

/ 1

be an exact sequence of -modules. Let


...

and

...

/ 2

/ ......
/ ......

1
1

/ 3

/0
0

/ 0

/ 0

/ 1

/0

/ 3

/0

be two projective resolutions of 1 and 3 respectively (see Injective and projective


resolutions). Then there is a projective resolution of 2
...

such that the diagram

...
...
...


/


/


/

0

/ ......

/ ......
/ ......
/ ......

1
1
1

/ 0 0
0


/ 0


/ 0 0
0


/ 0

0

0
0
0

/ 2

/0


/ 1


/ 2


/ 3

0

/0
/0
/0

is commutative with exact rows and columns (where, for any , the map is the obvious injection and the map is the obvious projection).
The lemma takes its name from the shape of the given part of the diagram (the short
exact sequence, and the given projective resolutions).

An analogous statement holds in case we have an injective resolution of 1 and an


injective resolution of 3 .

Hurwitzs theorem.

See Riemann surfaces (compact -) and algebraic curves.

Hypercohomology of a complex of sheaves.

ly [93].

([79], [93]). We follow strict-

Hyperplane bundles, twisting sheaves

Let

93

F F+1

be a complex of sheaves of Abelian groups on a topological space (see Sheaves).


We denote it by F .
The cohomology sheaf H (F ) is the sheaf on associated to the presheaf

(F () F+1 ())

(F1 () F ())

for any open subset of . Let U = { } be an open covering of . Consider the


bigraded complex (U, F ) (see Sheaves and in particular the cohomology of the
sheaves for the notation) with the differentials
: (U, F ) +1 (U, F ),

: (U, F ) (U, F+1 ),

where is induced by the differential of F . Let (U) be the associated single complex
given by (U) = += (U, F ) with differential = + .
The hypercohomology of the complex of sheaves F is defined to be the limit on U of
the cohomology of the complex (U), where the set of the open coverings is partially
ordered by refinement (see Limits, direct and inverse -):
IHI (, F ) = lim ( (U))
U

for any .
By considering the two filtrations of U and by passing to the limit for U, we get two
,
spectral sequences (see Spectral sequences) and with 2 (, H (F ))

and 2

( (, F )).

Proposition. If a map between complexes of sheaves induces an isomorphism on cohomology sheaves, then it also induces an isomorphism on hypercohomology.

Hyperelliptic Riemann surfaces.

gebraic curves.

See Riemann surfaces (compact -) and al-

Hyperelliptic or bielliptic surfaces.

See Surfaces, algebraic -.

Hyperplane bundles, twisting sheaves. ([93], [107], [129]). Let be a field


and . The line bundle on is defined in the following way: let the total space
be
((+1 {0}) )/ ,
where is the following equivalence relation:

(0 , . . . , , ) (0 , . . . , , ) {0};

94 | Hyperplane bundles, twisting sheaves


the projection

is given by

([(0 , . . . , , )]) = [0 : : ].

We can take as trivializing subsets for the sets

= {[0 : : ] | = 0}

for = 0, . . . , , and the trivializing function : | is


([0 : : ], ) = [(0 , . . . , , )].

Thus the transition functions are

, ([0 : : ]) = .

The bundle 1 is called a hyperplane bundle (since it is associated to the divisor given
by a hyperplane) and is the dual of the universal bundle (see Tautological (or universal) bundle). Obviously = . The sheaf O() (or, more precisely, O ()) is defined to be the sheaf of the sections of . Sometimes O() denotes also the line bundle
itself. The sheaves O() are called Serres twisting sheaves.
For any sheaf E on , we denote E O() by E().

A more general way to define the hyperplane bundles and the twisting sheaves is the
following one.
Let be a graded commutative ring with unity. For any graded -module , let F
be the following sheaf on the scheme () (see Schemes): for every (),
let () be the group of the elements of degree 0 in the localization of with respect
to the multiplicative system of the homogeneous elements in (see Localization,
quotient ring, quotient field); for every open subset of (), let F () be the set
of the functions
: ()

such that () () for every and is locally a fraction, i.e., for all ,
there exists a neighborhood of in and homogeneous elements , of
the same degree such that, for all , we have and ( ) = .
For any graded -module and , let () be the graded module whose part of
degree is + (the part of degree + of ). For any , we define the twisting
sheaf O () on := () in the following way:
O () = F() ,

i.e., O () is defined to be the sheaf associated to the -module (). We can prove that
O () is locally free (see Sheaves) of rank 1 and that, for any , ,
O () O () = O ( + ).

Injective and projective modules

95

If we take = () where = [0 , . . . , ] for some algebraically closed field


(thus is the scheme corresponding to the projective space of dimension over ),
we get the sheaves we have described before.
Theorem.
Let be a Noetherian graded commutative ring with unity and let
[0 , . . . , ]. Let = (). Denote O () by O() for the sake of brevity.
For any , we have

0 (, O()) = [0 , . . . , ] ,

where [0 , . . . , ] is the part of degree of [0 , . . . , ], i.e., the one given by


homogeneous polynomials of degree . Therefore
and

0 (, O()) = (

+
)

for 0

0 (, O()) = 0

Furthermore,

By Serre duality (see Serre duality)

(, O()) = 0

for < 0.

, for > 0, = .

0 (, O()) = (, O( 1)) .

Theorem. Let be an algebraically closed field. Then any locally free sheaf of rank 1
on the scheme is isomorphic to O() for some .

Injective and projective modules. ([41], [62], [116], [159], [208]). Let be a
ring. An -module is said to be injective if for every injective -morphism of modules : and every -morphism : there exists an -morphism
: such that = :





/
|
|
|
.

||
 }|||

An -module is said to be projective if for every surjective -morphism of -modules


: and every -morphism : there exists an -morphism :

96 | Injective and projective resolutions


such that = :


//

}}
}
}
.
}}

~ }
}
//

Proposition. Let be a ring with unity and consider only unital modules.
Free -modules (i.e., direct sums of copies of ) are projective.
Any projective module is a direct summand of a free module.
Let be a field. Any finitely generated projective [0 , . . . , ]-module is free (see [159,
Chapter IV, Theorem 3.15]).

Injective and projective resolutions.

([41], [62], [116], [208]). Let be a ring.


An injective resolution of an -module is an exact sequence of -modules
0 0

with injective -modules.


A projective resolution of an -module is an exact sequence of -modules
1 0 0

with projective -modules.

Integrally closed. ([12], [62], [164], [256]). Let be a commutative ring with unity.
Let be a subring of (so the unity belongs to ). We say that is integral over
if there exists a monic polynomial with coefficients in such that () = 0.
The integral closure of in is defined to be the set
:= { | integral over }.

If = , then we say that is integrally closed in . If = , then we say that is


integral over .
We say that a domain is integrally closed if it is integrally closed in its quotient
field (see Localization, quotient ring, quotient field).

Proposition.
(i) Let be a commutative ring with unity. Let be a subring of .
An element is integral over if and only if [] is a finitely generated
-module.
The integral closure of in is a subring of .
If is integral over and is an ideal of , then / is integral over /( ).
Let be the integral closure of in and let be a multiplicative part of ,
then 1 is the integral closure of 1 in 1 (see Localization, quotient
ring, quotient field for the notation 1 ).

Intersection of cycles

97

(ii) Let be a commutative ring with unity and let , be two subrings of with
. If is integral over and is integral over , then is integral
over .
(iii) Let be a domain. Then is integrally closed if and only if is integrally closed
for every prime ideal of and this is true if and only if is integrally closed for
every maximal ideal of (see Localization, quotient ring, quotient field for
the notation and ).
(iv) A unique factorization domain is integrally closed, in particular, for any field ,
the ring [1 , . . . , ] is integrally closed.

Intersection of cycles.

([72], [74], [93], [107], [214], [227], [228]).

Intersection in topology. We follow mainly [93].


Let be an oriented manifold of (real) dimension . Let 1 and 2 be two singular
homology classes of cycles of complementary dimension, precisely,
1 (, ),

2 (, )

(see Singular homology and cohomology for the notation). Let 1 and 2 be two
piecewise smooth cycles such that [1 ] = 1 , [2 ] = 2 , and 1 and 2 intersect transversely (i.e., every intersection point is smooth in 1 and 2 and = 1 2 ,
where denotes the tangent space at ). We define
1 2 =

(1 , 2 ),

1 2

where (1 , 2 ) is defined as follows: take a positively oriented basis B1 of 1 and a


positively oriented basis B2 of 2 (observe that 1 and 2 have natural orientations
given respectively by 1 and 2 ); if B1 , B2 is a positively oriented basis of , then we
define (1 , 2 ) to be 1, otherwise we set (1 , 2 ) = 1. One can show that we can find
1 and 2 as above and that 1 2 does not depend on the choice of 1 and 2 .
We can define also the intersection of two singular homology classes of cycles of noncomplementary dimension. Precisely, let
1 1 (, ),

2 2 (, ),

with 1 + 2 < . Let 1 and 2 be two piecewise smooth cycles such that [1 ] = 1 ,
[2 ] = 2 and 1 and 2 intersect transversely almost everywhere (we say that they
intersect transversely in a point if is smooth in 1 and 2 and 1 + 2 = ).
We define 1 2 to be the element of 1 2 (, ) given by 1 2 with the following
orientation: if B is a positively oriented basis of (1 2 ), where is a smooth point of
1 2 , and B1 , B is a positively oriented basis of 1 and B, B2 is a positively oriented
basis of 2 , then B1 , B, B2 is a positively oriented basis of .

98 | Inverse image sheaf


One can prove that the intersection of cycles corresponds through the Poincar duality
to the cup product; see Singular homology and cohomology.
Intersection in algebraic geometry. Let be a smooth algebraic variety of dimension
over an algebraic closed field and let 1 and 2 be two subvarieties of of codimension 1 and 2 , respectively. Suppose 1 + 2 . We say that 1 and 2 intersect
properly if the codimension of every irreducible component of 1 2 is equal to 1 +2 .
Suppose that 1 and 2 intersect properly. We define the intersection cycle of 1 and
2 , which we denote by 1 2 or simply by 1 2 , to be the algebraic cycle (see Cycles)
(1 , 2 ) ,

where the sum runs over all the irreducible components of 1 2 and the number
(1 , 2 ) is the so-called intersection multiplicity of 1 and 2 along . There are
several definitions of intersection multiplicity; we report Serres definition (see [227]):
if is an irreducible component of 1 2 ,
(1 , 2 ) := (1) ( (/1 , /2 )) ,

where is the local ring O, of at a generic point and is the ideal of in


for = 1, 2 (see Length of a module, Tor, TOR for the definitions of and
). Serre showed that the numbers (1 , 2 ) are nonnegative.
By linearity we can define the intersection cycle of any two cycles 1 and 2 when
they intersect properly, i.e., when the subvarieties of which 1 is a linear combination
intersect properly the subvarieties of which 2 is a linear combination.
Chows moving lemma states that if is a smooth quasi-projective algebraic variety
and 1 and 2 are cycles of of codimension 1 and 2 , respectively, we can find a
cycle 1 rationally equivalent to 1 and such that 1 and 2 intersect properly (see
Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard groups
for the definition of rational equivalence). Moreover, one can show that if 1 and 1
are rationally equivalent and intersect properly 2 , then 1 2 and 1 2 are rationally
equivalent. This allows us to define the intersection of algebraic cycles (also intersecting not properly) up to rational equivalence.

Inverse image sheaf.

See Direct and inverse image sheaves.

Irreducible topological space.

We say that a topological space is irreducible


if it is not the union of two proper closed subsets.

Irregularity. ([93], [107]). Let be a complex manifold or an algebraic variety over


a field . The irregularity of is
1 (, O ),

where O is the sheaf of the holomorphic functions, respectively of the regular functions, on . The irregularity of is generally denoted by ().

Jacobians of compact Riemann surfaces

99

If is a compact Khler manifold (see Hermitian and Khlerian metrics), then, by


Dolbeaults theorem and by Hodges theorem (see Dolbeaults theorem, Hodge the1
ory), we have that 1 (, O ) = 0,1 () = 1,0 () = 0 (,
) (and somewhere the
1,0
irregularity is defined to be ()).
If is a smooth projective algebraic surface over an algebraic closed field, then the
irregularity is equal to () () (where () and () are respectively the arithmetic and the geometric genus of ; see Genus, arithmetic, geometric, real, virtual -
and Dualizing sheaf).

J
Jacobians of compact Riemann surfaces. ([93], [101], [102], [163], [165],
[195]). To every compact Riemann surface we can associate a principally polarized
Abelian variety, called its Jacobian. Jacobians were the first Abelian varieties to be
studied.
Let be a compact Riemann surface of genus (see Riemann surfaces (compact -) and algebraic curves). By Riemanns theorem, the complex vector space
0 (, O( )), where is the canonical bundle of (see Canonical bundle, canonical sheaf), has dimension . Consider the map
: 1 (, ) 0 (, O( ))

defined by

It is injective, and the quotient

() = .

0 (, O( ))
(1 (, ))

is a complex torus of dimension (see Tori, complex - and Abelian varieties). The
Jacobian of is the complex torus above endowed with the following polarization:
let be the alternating form on 0 (, O( )) obtained extending on the form on
1 (, ) given by the intersection of 1-cycles; let
: 0 (, O( )) 0 (, O( )) ,

defined by
0

(, ) = (, ) + (, ),

for any , (, O( )) . We endow the complex torus above with the polarization
given by . One can prove that it is a principal polarization. Thus the Jacobian of is
a principally polarized Abelian variety.
The Jacobian of coincides with the Albanese variety of (see Albanese varieties).

100 | Jacobians of compact Riemann surfaces


Definition. Let () be the set of divisors of degree on . The map
defined by

0 () ()

( )

=1,...,

=1,...,

for any , , , is called the AbelJacobi map of and denoted by .


Obviously, if we fix a point on , we can also define a map
() ()

(again called the AbelJacobi map and denoted by ) by composing the map
() 0 (),

with the AbelJacobi map 0 () ().

AbelJacobi theorem. The AbelJacobi map defines an isomorphism (again called the
AbelJacobi map)
0 () () .

(See Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard


groups for the definition of the Picard group 0 (); in case is a compact Riemann
surface, 0 () is the set of the divisors of degree 0 up to linear equivalence.)

Notation. Define := (() ), where () is the symmetric -product of , i.e., the


set of effective divisors of degree on .

Proposition. The dimension of is if , while it is if . If () , then


1 (()) = || = (0 (, O()),

where || denotes the linear system of (see Linear systems) Let 1; then the
map : () is injective and the map : () () is generically injective
for .

Theorem. Let be the divisor in () associated to a section of a holomorphic line


bundle on () defining the polarization. We have that the intersection number of 1
and is :
1 = .

In addition, is a translate of the image through the AbelJacobi map of the set of the

Jacobians, Weil and Griffiths intermediate -

101

effective divisors of degree 1 on , i.e.,

= 1 + ,

for some ().

Torellis theorem. The map from the set of compact Riemann surfaces up to isomorphisms to the set of the principally polarized Abelian varieties up to isomorphisms,
associating to every Riemann surface its Jacobian () is injective.

Poincars formula. Let [ ] be the class of in the singular homology group


22 ((), ). We have that
[]
[ ] =
,
( )!
where is the divisor associated to a section of a holomorphic line bundle defining
the polarization on ().

Jacobians, Weil and Griffiths intermediate -.

([92], [93], [166], [250]). We

follow mainly [166].


Let be a compact Khler manifold of dimension (see Hermitian and Khlerian
metrics). Let 1 . The Griffiths -th intermediate Jacobian of is the complex
torus (see Tori, complex - and Abelian varieties)
() =

=,...,21 21, (, )
(21 (, ))

where : 21 (, ) =,...,21 21, (, ) is given by the composition of the


canonical map : 21 (, ) 21 (, ) with the projection 21 (, )
=,...,21 21, (, ) induced by the Hodge decomposition (see Hodge theory and
Singular homology and cohomology, in particular the Universal Coefficient Theorem). If = 1 we have the Picard variety of (see Equivalence, algebraic, rational,
linear -, Chow, NeronSeveri and Picard groups):
1 () =

1 (, O)
= 0 ().
(1 (, ))

If = we have the Albanese variety of (see Albanese varieties):


() =

1, (, )
1,0 (, )
0 (, 1 )

= (),
21
(
(, )) (1 (, )) (1 (, ))

where is the map (see Dolbeaults theorem, Hodge theory, Serre duality
to understand the isomorphisms above).
We have
(1,0 (, ), )
0,1 ((), )
0 (()) =

(1 (, ))
(1 ((), ))

1 (, O)
= 0 (),
(1 (, ))

102 | Jacobians, Weil and Griffiths intermediate and analogously we also have an isomorphism (0 ()) (). So 0 () and
() are dual complex tori (see Tori, complex - and Abelian varieties). More generally, we can prove that () and +1 () are dual complex tori.
Now let be a smooth complex projective algebraic variety, and let be the FubiniStudy form restricted to (thus is a closed positive integer (1, 1)-form).
If 2 1 , we can consider on () the polarization with index defined by the
following Hermitian form (if 2 1 > , we define the polarization with index as the
dual polarization with index of the one on +1 ()):
(, ) = 2(1) 2+1 ;

the form is Hermitian; in fact it is -linear in the first variable and


(, ) = 2(1) 2+1

= 2(1) 2+1 = 2(1) 2+1 = (, ).

In general it is not positive definite, so in general the Griffiths intermediate Jacobian


is not an Abelian variety; the Albanese variety and the Picard variety (of a smooth
complex projective algebraic variety) are Abelian varieties.
Let 1 . Let : (, ) (, ) be the linear operator defined to be the
multiplication by on , (); it takes (, ) to (, ) and, if is odd, we
have 2 = 1. The operator is called Weils operator. The -th Weil intermediate
Jacobian of is the following torus:
() =

2 1 (, )
,
(2 1 (, ))

with the complex structure given by ; here : 2 1 (, ) 2 1 (, ) is the


canonical map; its image is isomorphic to the free part of 2 1 (, ). We can prove
that () and +1 () are dual complex tori.
If 2 1 , we endow () with the polarization defined by the Hermitian form
: 2 1 (, ) 2 1 (, ) ,

(, ) := (, ) + (, ),

where, if = and = are the Lefschetz decomposition of and (see


Lefschetz decomposition and Hard Lefschetz theorem),
(, ) := (1)+ 2+1+2 .

Jumping lines and splitting type of a vector bundle on

103

For 2 1 > , we define the polarization on () to be the dual polarization of the


one on +1 ().
One can prove that the Weil intermediate Jacobian is an Abelian variety (with the polarization we have just defined).
As we have already said, Griffiths intermediate Jacobian is not an Abelian variety in
general. The advantage of Griffiths Jacobian with respect to Weil Jacobian is that Griffiths Jacobian varies holomorphically in a family of smooth projective algebraic varieties (while Weil Jacobian does not).
See also Jacobians of compact Riemann surfaces.

Jumping lines and splitting type of a vector bundle on .

([61], [207]).
Let be a holomorphic vector bundle of rank on = . Let be a line in . By
the GrothendieckSegre theorem (see GrothendieckSegre theorem), we have that
| = O (1 ) O ( )

for some 1 , . . . , (see Hyperplane bundles, twisting sheaves for the definition
of O ( )). Suppose 1 . We say that
(1 , . . . , )

is the splitting type of on . We denote it by ().


We order by the lexicographic order. By using Grauerts semicontinuity theorem
(see Grauerts semicontinuity theorem), one can prove the following lemma:

Lemma. Let (1 , . . . , ) . The set { (1, )| () > (1 , . . . , )} is a Zarisky


closed subset of (1, ) (the Grassmannian of the lines in ).

Let

:= { ()| (1, )}.

By Lemma above, there is a Zariski open subset of (1, ) such that () = for
. We say that is the generic splitting type of . If (1, ) , we say that
is a jumping line for .
About the generic splitting type we can recall the following theorem (see Stable
sheaves for the definition of semistable bundle):
GrauertMlichSpindler theorem. Let be a semistable holomorphic -vector bundle on with generic splitting type (1 , . . . , ), 1 . We have +1 1 for
1 1.

104 | Khler

K
Khler.

See Hermitian and Khlerian metrics.

Kodaira dimension (or Kodaira number).

([22], [93], [107], [241]). Let be a


compact complex manifold or a smooth projective algebraic variety over a field .

Let be the set of positive natural numbers such that 0 (,


) = 0, where is the
canonical sheaf (see Canonical bundle, canonical sheaf).
If = 0, we say that the Kodaira dimension of is 1 (or ).
If = 0, the Kodaira dimension of is the maximum of the following set:
{( ())| },

where is the canonical bundle and is the map associated to the line bundle
(see Bundles, fibre -).

For instance, the Kodaira dimension of a compact Riemann surface of genus is 1 if


and only if = 0; it is 0 if and only if = 1; and it is 1 if and only if 2.

) by () (plurigenera). If is a complex smooth algebraic


Let us denote 0 (,
surface, one can prove the following statement: if ( ()) is a bounded nonzero sequence, the Kodaira dimension of is 0; if ( ()) is an unbounded sequence, but
there exists such that () for all , the Kodaira dimension of is 1; if ( ()/)
is an unbounded sequence, the Kodaira dimension of is 2.

When the Kodaira dimension of is equal to dimension of , we say that is of the


general type.

Kodaira embedding theorem.

([93], [107], [150]).

Kodaira embedding theorem. Let be a compact complex manifold and be a positive


holomorphic line bundle on (see Positive).
Then there exists a natural number such that, for all , the map associated to ,
: (0 (, O( )) ), is well defined, and an embedding (see Bundles, fibre -
for the definition of the map associated to ).
In other words, a positive holomorphic line bundle is ample.

Observe that the converse is also true: an ample holomorphic line bundle is positive.
In fact, if embeds into and is the hyperplane bundle of , then () =
and so 1 ( ) = (1 ()) and, since the hyperplane bundle is positive (see FubiniStudy metric), we conclude.
Another statement of the Kodaira embedding theorem is the following.

Lefschetz decomposition and Hard Lefschetz theorem

105

Kodaira embedding theorem. Let be a compact complex manifold. It is embedable


in a projective space if and only if it has a closed positive (1, 1)-form such that the
cohomology class of is rational, that is, such that [] 2 (, ).

KodairaNakano vanishing theorem.

See Vanishing theorems.

Koszul complex. ([62], [89], [93], [107], [164], [185], [209]). Let be a commutative ring with unity and 1 , . . . , ; let be a free -module of rank with basis
{1 , . . . , }; the Koszul complex is the following complex:

1 /(1 , . . . , ) 0,

(4)

where 0 is the projection and, for every = 1, . . . , , : 1 is the module


morphism such that
(1 ) (1)+1 ( )(1
)
=1,...,

for any and for any distinct 1 , . . . , {1, . . . , }. If is an -module, by tensoring (4) by we get the complex

1 /(1 , . . . , ) 0,

(5)

Proposition. If 1 , . . . , is an -regular sequence (see Regular sequences), the


Koszul complex (5) is exact.

K3 surfaces.

See Surfaces, algebraic -.

Kummer surfaces.

See Surfaces, algebraic -.

L
Lefschetz decomposition and hard Lefschetz theorem.

([44], [90], [93],


[245], [250], [251]). Let (, ) be a compact Hermitian manifold of complex dimension
(see Hermitian and Khlerian metrics). Let be the (1, 1)-form associated to . Let
be the operator on the set of forms on defined by
() =

and let be the adjoint operator of with respect to the product (, ) = , where
the star operator is defined as usual on real forms and extended by -linearity on
complex forms (see Star operator). We say that a -form is primitive if
() = 0.

106 | Lefschetz theorem on (1, 1)-classes

We can prove that, if , then a -form is primitive if and only if +1 () =


0 and that any primitive form of degree greater than must be zero. Let us denote

by (, ) the -th De Rham cohomology of , i.e., the vector space of the closed
complex -forms on modulo the vector space of the exact complex -forms on . If
(, ) is Khler, i.e., is closed, then the operators and induce operators, called

(, ) and (, ) is said to
again and , on the De Rham cohomology
be primitive if () = 0.
Let

(, ) = { (, )| () = 0}.
Lefschetz decomposition.

(, ) = 2 (, ) .

Hard Lefschetz. For any with 1 , the map


is an isomorphism.

+
:
(, )
(, )

Lefschetz theorem on (1, 1)-classes.

([16], [93], [245]). Let be a smooth


complex projective algebraic variety. Every element of the intersection 2 (, )
1,1 () is the class of a divisor on (see Equivalence, algebraic, rational, linear -,
Chow, NeronSeveri and Picard groups).

Lefschetz hyperplane theorem.

([6], [31], [93], [172], [245]). Let be an dimensional compact complex manifold and a smooth hypersurface such that
the associated line bundle () (see Bundles, fibre -) is positive. The map
(, ) (, )

induced by the inclusion of into is an isomorphism for 2 and injective for


= 1 (see Singular homology and cohomology for the definition of (, )).

Length of a module. ([12], [62], [256]). Let be a commutative ring with unity. Let
be an -module. A chain of length of submodules of is a sequence of submodules , = 0, . . . , , such that = 0 1 = 0 and all the inclusions are
strict.
We say that a chain is maximal if we cannot insert other submodules. One can prove
that two maximal chains of a module have the same length.
If in a module there exists a maximal chain of submodules, we say that is of
finite length and we call length of the length of any maximal chain; we denote it
by (). If in there is not a maximal chain of submodules, we say that the length of
is +.

Liaison or linkage

107

Proposition.
(i) A module is of finite length if and only if it is Artinian and Noetherian (see
Noetherian, Artinian).
(ii) If 0 1 2 3 0 is an exact sequence of -modules of finite length,
then (2 ) = (1 ) + (3 ).
(iii) Let be a field. A -vector space has finite dimension if and only if has finite
length and, in this case, the dimension is equal to the length.

Leray spectral sequence.

([34], [53], [84, II 4.17], [93], [129], [175], [176]). Let


: be a continuous map between two topological spaces. Let F be a sheaf on
. The -direct image sheaf R (F) is the sheaf on associated to the presheaf
(1 (), F|1 () ),

where is any open subset of (see Sheaves, Direct and inverse image sheaves).
Under certain assumptions, there exists a spectral sequence { } (see Spectral sequences), called Leray spectral sequence, such that
,

2 = (, R (F)),
(, F).

Let and be compact Khler manifolds (see Hermitian and Khlerian metrics)
and : be a surjective holomorphic of maximal rank. One can prove (see [53]
or [93]) that in this case Leray spectral sequence degenerates at 2 , i.e.,

Liaison or linkage.

2 = .

([186], [187], [212], [213]). Let be an algebraically closed field


and let be the scheme ([0 , . . . , ]). Let 1 and 2 be two subschemes of
such that no component of 1 is contained in any component of 2 and conversely. We
say that 1 and 2 are geometrically G (respectively CI) linked if 1 2 is arithmetically Gorenstein (respectively complete intersection).

Let 1 and 2 be two subschemes of . We say that 1 and 2 are algebraically G(resp. CI-) linked if there exists arithmetically Gorenstein (resp. complete intersection) such that 1 2 , ( : 1 ) = 2 and ( : 2 ) = 1 , where denotes the
saturated ideal of (see Notation for the definition of ( : )).
One can prove that geometric linkage implies algebraic linkage.

Let be a subscheme of . The G-linkage (respectively CI-linkage) class of is the


set of the subschemes of that can be obtained from by a finite number of G (respectively CI) algebraic linkages.

108 | Lie algebras

Lie algebras.

with a map,

([76], [115], [125], [133]). A Lie algebra is a vector space over a field
[ , ] : ,

called a bracket, such that


(i) [ , ] is bilinear;
(ii) [, ] = 0 for all ;
(iii) [, [, ]] + [, [, ]] + [, [, ]] = 0 for all , , (Jacobi identity).

Examples. Let (, ) be the set of the matrices with entries in . It is a Lie algebra
with the bracket:
[, ] = .
Let (, ) be the set of the matrices with entries in and with null trace. It is a
Lie algebra with the bracket:
[, ] = .

The Lie algebras (, ), (, ) take their names from the fact that they can be identified respectively with the tangent space in the identity of (, ) (the set of invertible
matrices with entries in ) and the tangent space in the identity of (, ) (the
set of matrices with entries in and with determinant equal to 1).
Definition. A Lie algebras morphism from a Lie algebra (, [ , ]) to a Lie algebra
( , [ , ] ) is a linear map
:

such that [(), ()] = ([, ]) for any , .

Definition. We say that a Lie algebra (, [ , ]) is Abelian if [, ] = 0 for any


, .

Definition.
(i) A subspace of a Lie algebra (, [ , ]) is called a Lie subalgebra of if [, ] .
(ii) A subspace of a Lie algebra (, [ , ]) is called an ideal of if [, ] .
(iii) Let be a Lie subalgebra of a Lie algebra (, [ , ]). The normalizer of is the set
{ | [, ] } .

From now on, we will consider only Lie algebras of finite dimension.
Definition.
We say that a Lie subalgebra of a Lie algebra (, [ , ]) is solvable if the sequence
of subalgebras
1 := [, ],

terminates to zero.

2 := [1 , 1 ],

:= [1 , 1 ], . . .

Lie algebras

109

We say that a Lie subalgebra of a Lie algebra (, [ , ]) is nilpotent if the sequence


of subalgebras
1 := [, ],

terminates to zero.

2 := [, 1 ],

:= [, 1 ], . . .

From now on, we will consider only Lie algebras over an algebraically closed field
of characteristic 0.

Notation. Let (, [ , ]) be a Lie algebra and let ; we denote by


the map
for any .

() :
[, ]

Notation. The Killing form on a Lie algebra (, [ , ]) is the symmetric bilinear form
defined in the following way:
(, ) := (() ())

for any , (where denotes the trace).

Definition.
(a) We say that a nonzero Lie algebra (, [ , ]) is simple if there is no nontrivial ideals
(in some works, the condition > 1 is also required; in other works, the
(stronger) condition that is not Abelian is also required).
(b) We say that a nonzero Lie algebra (, [ , ]) is semisimple if there are no nonzero
solvable ideals (equivalently if the Killing form is nondegenerate).
Theorem. A semisimple Lie algebra is the direct sum of all its simple ideals.
Definition. We say that a Lie subalgebra of a Lie algebra (, [ , ]) is a Cartan subalgebra if it is nilpotent and is equal to its normalizer.
The Cartan subalgebras are useful for classifying semisimple Lie algebras.

Definition. Let (, [ , ]) be a semisimple Lie algebra and let be a Cartan subalgebra (we can prove that it exists and that, in the case of semisimple Lie algebras, it is
Abelian and its elements are semisimple, i.e., diagonalizable).
We say that a linear map : is a root if it is nonzero, and if we define
:= { | [, ] = () }

110 | Lie algebras


we have that = {0}. We call the subspace the root space of . Let be the set of
the roots.
Definition. Let (, ( , )) be a finite dimensional euclidean space, i.e., a finite dimensional vector space over endowed with a positive definite symmetric bilinear form
( , ). Define
2(, )
, =
(, )
for any , with = 0. We say that a subset of is a root system if
(1) is finite, it generates and does not contain 0;
(2) if , the unique multiples of in are and ;
(3) if , () = , where is defined by () = , ;
4) if , , , .

We say that a root system is reducible if there exist 1 , 2 such that 1 and 2 are
perpendicular for ( , ) and = 1 2 .

Definition. Let (, ( , )) be a finite dimensional euclidean space, and let be a root


system. Let = {1 , . . . , } be a subset of . We say that is a basis of if
(1) is a basis of ;
(2) we can write every as = =1,..., with , and the all
nonnegative or all nonpositive.
Definition. Let (, ( , )) be a finite dimensional euclidean space and let be a root system. Let = {1 , . . . , } be a basis of . The associated Cartan matrix is defined by
, = , =

2( , )
( , )

for any , {1, . . . , }.


The associated Dynkin diagram is defined in the following way: we consider a vertex
for every element of the basis and we join the two vertices corresponding to and
by , , lines (we can prove that , , is a number between 0 and 3);
if between the two vertices corresponding to and there are 2 or 3 lines, we also
put an arrow toward the vertex corresponding to if | , | < | , |.

Theorem. The Dynkin diagram associated to a roots system does not depend on the
choice of a basis.
The Dynkin diagram of an irreducible root system is one of the diagrams in Figure 12.
The map
{irreducible roots systems}/isomorphisms {Dynkin diagrams of kind , . . . , },

associating to every class of an irreducible root system its Dynkin diagram, is a bijection.

Lie algebras

( > 0)

( > 1)

( > 2)

( > 3)
6

l-1

l-2

l-1

l-2

l-1

l-3

l-2

111

(2 + 1)

(2 + 1)

l
l-1
l

(2)

(2)

7
8

Fig. 12. Dynkin diagrams and Lie algebras.

Theorem. Let be a semisimple Lie algebra and one of its Cartan subalgebras. On
(the dual space of ) we define the following form ( , ):
For any there exists a unique such that ( , ) = () for all , where
is the Killing form; we set (, ) = ( , ) for any , .
Let be the -subspace of spanned by the set of the roots; the form just defined
gives a positive definite bilinear form on it, so it is an euclidean space.
(1) The set of roots is a root system of .
(2) We have
=
(Cartan decomposition).
(3) For any , the subalgebra has dimension 1. For any , the subalgebra
[ , ] has dimension 1 and it is generated by .
(4) If , , = 0, then there exists such that, if we define
= [ , ], we have , , generate a subalgebra of isomorphic to (2, )
via
0 1
0 0
1 0
),
),
).
(
(
(
0 0
1 0
0 1

Theorem. The root system associated to a semisimple Lie algebra, as described in the
theorem above, does not depend on the choice of the Cartan subalgebra. The map
{semisimple Lie algebras over }/isomorphisms {root systems}/isomorphisms

is a bijection.

112 | Lie groups


Therefore the classification of the Dynkin diagrams associated to the irreducible root
systems allows us to classify simple Lie algebras and then semisimple Lie algebras.
See Lie groups and Representations.

Lie groups.

([3], [76], [115], [126], [210], [226]). A Lie group is a group endowed
with a structure of manifold such that the product and the map associating to any
element of its inverse are maps.
A morphism of Lie groups is a homomorphism of groups.
Let be a Lie group, and let () be the group of the automorphisms of . Let
be defined in the following way:

: ()
()() = 1

for any , . Observe that, if is the identity element, ()() = for all ; thus
the differential of () in is a map from the tangent space of at to itself:
(()) : .

We define the adjoint representation (see Representations) of to be


: ( ),
(()) .

Its differential in is denoted by :

: ( ),

where ( ()) is the vector space of the endomorphisms of (). We can define a
bracket on by
[, ] = ()().

With this bracket, the vector space is a Lie algebra (see Lie algebras), which is
denoted (), and is a representation of the Lie algebra () on ().

Remark. If ( , ), then the bracket in () is


[, ] = .

In fact, for any , the map () is the map 1 , then () is the map
1 and, for any , , the map (+) is the map (+)(+)1 .
So () () is the map .
Definition. We say that a subset of a Lie group is a Lie subgroup if it is a subgroup
and a closed submanifold.

Lie groups

113

Remark. Let and be two Lie groups and let : be a morphism of Lie
groups. The map : is a Lie algebra morphism.

Proposition. Let and be two Lie groups with connected. Then a morphism of
Lie groups : is uniquely determined by .
Furthermore, if is simply connected, the map sending a morphism of Lie groups
: to its differential at , : , is a bijection between the set of
the morphisms of Lie groups from to and the set of the morphisms of Lie algebras
from () to ( ).
Proposition. Let be a Lie group. Let . There exists a unique Lie group morphism : such that ( )0 (1) = .
Definition. We define the exponential map

by
for any .

exp :

exp () = (),

Proposition. For every Lie group , the map is smooth.


Furthermore, if : is a morphism of Lie groups, we have exp =
exp .
Theorem.
(a) Every finite dimensional Lie algebra is the Lie algebra of a simply connected Lie
group.
(b) Let be a connected Lie group. The subalgebras of () are in bijection with
the connected Lie subgroups of . The bijection sends a Lie subalgebra to the
subgroup generated by exp() (its tangent space is ).
In addition, if is a connected Lie subgroup of and the corresponding subalgebra, then is normal if and only if is an ideal.
Proposition. Let be a connected Lie group. We have that is a solvable group if
and only if () is solvable (see Lie algebras for the definition of solvable Lie algebra).

Definition. We say that a connected Lie group is simple if there are not in nontrivial
normal connected Lie subgroups.
We say that a connected Lie group is semisimple if there are not in nontrivial
normal solvable connected Lie subgroups.

By the proposition above, is (semi)simple if () is (semi)simple.

Example. (, ) is simple.

114 | Limits, direct and inverse Definition. We say that a subalgebra of a semisimple Lie algebra is Borel if it is a
maximal solvable Lie subalgebra.
We say that a Lie subgroup of a semisimple Lie group is Borel if the corresponding
subalgebra is Borel (or equivalently if it is a maximal solvable connected subgroup).
We say that a Lie subgroup of a semisimple Lie group is parabolic if it contains a
Borel subgroup.
Proposition. Let be a semisimple Lie group. A Lie subgroup is parabolic if and only
if / is a projective algebraic variety.

Example. Let = (3); the subgroup = { (3)| 2,1 = 3,1 = 0} is parabolic;


it contains the Borel subgroup of the upper triangular matrices. The quotient (3)/
is 2 .
Definition. We say that a Lie group is reductive if all its representations are completely
reducible (see Representations).

Definition. A complex Lie group is a group endowed with a structure of complex


manifold such that the product and the map associating to any element of its inverse
are holomorphic.
Theorem. (Unitary trick). Let be a complex Lie group. If there is a real compact Lie
group such that () = ( ) , then is reductive.

Limits, direct and inverse -.

([12], [91], [208]). Let (, ) be a partially ordered


set. We say that it is a filtering set if, for all 1 , 2 , there exists such that 1 ,
2 .

Let (, ) be a filtering set; a family of -modules { } is said to be a direct system


of modules if and only if, for all 1 , 2 with 1 2 , we have a homomorphism
2 ,1 : 1 2

such that
(i) if 1 2 3 , then 3 ,2 2 ,1 = 3 ,1 ;
(ii) , is the identity for all .

Let { } be a direct system of modules on a filtering set (, ) (with homomorphisms


, ); a direct (or inductive) limit of { } is an -module with a homomorphism
:

for all , such that, for all 1 , 2 with 1 2 , we have 2 2 ,1 = 1 and


the following universal property holds: if is an -module and, for all , we
have homomorphisms : such that for all 1 , 2 with 1 2 we have

Limits, direct and inverse -

115

2 2 ,1 = 1 , then there exists a homomorphism : such that = for


all .
By the universal property, the direct limit must be unique; in addition we can prove
that the direct limit always exists.
We denote the direct limit of a direct system of modules { } by
lim

Let { } and { } be two direct systems of -modules on a filtering set (, ), and


suppose that for any there exists a homomorphism : ; suppose
that such homomorphisms commute with the homomorphisms of the direct systems.
Then for every we have a homomorphism lim (the composition of the
maps and lim ) and then a homomorphism
: lim lim

by the universal property.

Proposition . Let { } , { } and { } be three direct systems of -modules on a


filtering set (, ), and suppose that for all there is an exact sequence of modules

such that for all 1 , 2 with 1 2 we have the commutative diagram


1


2

1
2

/
1

/
2

1
2

/
1

/
2

where the vertical arrows are the maps of the direct systems. Then we have an exact
sequence

lim lim lim .

Let (, ) be a filtering set; we say that a subset of is cofinal if is a filtering set


with the order induced by and if for all there exists such that .

Proposition. Let { } be a direct system of modules on a filtering set (, ) and


let be a cofinal subset of ; then the limit of { } is isomorphic to the limit of
{ } .
Reverting the arrows in the definition of direct system and direct limit, we get the notion of inverse (or projective) system and inverse (or projective) limit, which we

116 | Linear systems


denote by
lim .

For the inverse limit, we have analogous propositions except Proposition .

Linear systems. ([93], [107], [129], [228]). Let be an algebraically closed field.
Let be a complex manifold, respectively an algebraic variety over that is smooth
in codimension 1 (i.e., the codimension of the set of the singular points is greater than
or equal to 2), e.g., a normal variety (see Normal).
Let 1 and 2 be two Weil divisors on . We say that 1 and 2 , are linearly equivalent
if there exists a meromorphic, respectively rational, function on such that 1
2 = (), where () is the Weil divisor associated to (see Divisors, Equivalence,
algebraic, rational, linear -, Chow, NeronSeveri and Picard groups). If is a Weil
divisor on , the complete linear system || is defined in the following way:
|| = { | effective Weil divisor on and },

where denotes the linear equivalence. Furthermore, we define

L() = {| = 0 or merom., resp. rat., function on s.t. () + effective}.

One can easily prove that L() is a vector space over . Analogous definitions can be
given for Cartier divisors.
Let be a compact complex manifold or a complete normal algebraic variety over .
In this case, if and are two meromorphic, respectively rational, functions such that
() = (), then and are multiple by a constant. Therefore, for any divisor there
is an obvious bijection
|| (L()).

A linear system on a is a family of divisors corresponding to a projective subspace


of (L()); its dimension is its dimension as projective subspace; we say that it is a
pencil if its dimension is 1, a net if its dimension is 2, and a web if its dimension is 3.
The base locus of a linear system is the intersection of the supports of the divisors in
the linear system.
An often used notation about linear systems on Riemann surfaces is : it denotes
a linear system of degree and projective dimension .
Let be a complex manifold or a smooth algebraic variety over . For any divisor ,
we have that
0
L() (, O()),

where () is the line bundle associated to . The bijection can be described as follows.
If { } is an open covering of and on is given by = 0 ( meromorphic,
respectively rational, function), the transition functions of the bundle associated to

are , := .

We can associate to L() the section of O() given by on .

Localization, quotient ring, quotient field

117

Vice versa, let 0 (, O()) be given by : , respectively : ,

such that = , = ; we can associate to the meromorphic, respectively

rational, function in L() given by


See also Bertinis theorem.

Linkage.
Local.

on .

See Liaison or linkage.

([12], [62]). We say that a ring is local if it has a unique maximal ideal.

Localization, quotient ring, quotient field.

([12], [62]). Let be a commutative ring with unity 1. We say that is a multiplicative part of if 1 and is closed
under multiplication. Let
1 := / ,

where is the equivalence relation defined in the following way:

(, ) ( , ) | ( ) = 0.

We denote by / the equivalence class of (, ). With the usual definition of sum and
product of fractions,
/ + / := ( + )/ ,

(/)( / ) := / ,

we have that 1 is a commutative ring with unity. We call 1 the quotient ring
of with respect to .
Let be a prime ideal of ; then is a multiplicative part of ; we define the
localization of in to be 1 , where = , and we denote this by . It is a
local ring: the unique maximal ideal is {/ | , }.

If is an integral domain and = {0}, then 1 is equal to


where

( {0})/ ,

(, ) ( , ) = 0;

this is a field (with the usual definition of sum and product of fractions) and is called
the quotient field of and denoted by ().

Let be a commutative ring with unity. If is an -module and is a multiplicative


part of , we define
1 = / ,

where (, ) ( , ) if and only if there exists such that ( ) = 0. The


set 1 , with the obvious sum and product, is an 1 -module. A homomorphism of

118 | Lroth problem


-modules : 1 2 induces a homomorphism of 1 -modules 1 : 1 1
1 2 . If is a prime ideal of and = , we denote 1 by .

Proposition. Let be a commutative ring with unity and be a multiplicative part of .


(i) If 1 2 3 is an exact sequence of -modules, then the induced sequence
1 1 1 2 1 3

is exact.
(ii) Let be an -module. Then 1 1 as 1 -modules.
(iii) Let : be a homomorphism of -modules. Then is injective (respectively
surjective) : is injective (respectively surjective) for every prime
ideal : is injective (respectively surjective) for every maximal
ideal .
(iv) Let be an -module. We have: is a flat -module is a flat -module
for every prime ideal is a flat -module for every maximal ideal (see
Flat (module, morphism)).

Lroth problem. See Unirational, Lroth problem.

M
Manifolds. ([171], [200]) A topological manifold of dimension is a connected Haus-

dorff topological space such that there exists a countable open covering { }
and, for any , a homeomorphism
: .

We say that is a differentiable (respectively ) manifold if, for any , , the


map 1 (where it is defined) is differentiable (respectively ). The maps are
called local coordinates. A complex manifold of complex dimension is a manifold
whose local coordinates are
:
and such that, for any , , the map 1 (where is defined) is holomorphic.

Proposition. Every manifold is paracompact.

Mapping cone lemma.

([186], [180]). Let

/ 1

/ 2

/ 3

/0

be an exact sequence of -modules for some commutative ring with unity. Let
0

/ ......

/ 0

/ 1

/0

Minimal free resolutions

and
0

/ ......

/ 0

/ 2

119

/0

be two free resolutions of 1 and 2 respectively (that is, two exact sequences with
and free -modules for any ). Then we have a free resolution of 3
...

/ 1 2

...

/ 2

and a commutative diagram

...
...


/ 2


/ 1 2

2
2
2

/ 0 1
/ 1


/ 1


/ 0 1

1
1
1

/ 0

/ 0


/ 0


/ 0

0
0
0

/ 3

/0

/ 1

/0


/ 2


/ 3

/0
/ 0,

where the maps are the maps (0, ) for 1 and the identity for = 0 and the
maps : 1 2 1 are defined by
(, ) = (1 (), 1 () + ())

for 2, 0 := 0 , and 1 is defined by 1 (, ) = 0 () + 1 ().

Minimal set of generators.

([51]). We say that a set of generators of a module


is minimal if no proper subset generates the module.

Minimal free resolutions. ([51], [89], [164], [209]). For any graded module
over a ring and for any , we denote by () the graded module whose part of
degree is + (the part of degree + of ). Let = [0 , . . . , ], where is a field
and let be a finitely generated graded -module. Let

1 1 0 0

be a free graded resolution of , that is, an exact sequence where the are twisted
free graded modules (i.e., every is equal to (1 ) ( ) for some ,
1 , . . . , ) and the homomorphisms are graded homomorphisms of degree 0.
We say that the resolution above is minimal if for every 1 the constant entries of
the matrix of : 1 are zero.
Proposition. The resolution above is minimal if and only if for every 0 the map
takes the standard basis of to a minimal set of generators of the image of (see
Minimal set of generators).

120 | Minimal degree


Thus, to construct a minimal free resolution we can take as 0 the free module generated by a minimal set of generators of , then as 1 the free module generated by a
minimal set of generators of the kernel of 0 : 0 , and so on.
If we write as , () for some vector spaces , , we can prove that the dimensions of the , depend only on .
Furthermore one can prove that every finitely generated graded module over has a
minimal free resolution of length (see Hilbert syzygy theorem).

A minimal free resolution of a coherent sheaf F on is defined in the following way


(see Hyperplane bundles, twisting sheaves for the defintion of O ()):
Let = 0 (O ()) = [0 , . . . , ] and let F be the graded -module associated
to F (see Serre correspondence), i.e., the -module
0 (F());

a minimal free resolution of F is the sheafified of a minimal free resolution of F .


Observe that is O , i.e., it is the module associated to O, and it is true in general
that FF = F, and thus the terms of the sheafification of a minimal free resolution of
F are of the kind O( ) for some .

Minimal degree.

([93], [104]). Let be an algebraically closed field.

Theorem. The degree of a nondegenerate algebraic variety of dimension in is


greater than or equal to + 1.
A rational normal scroll of dimension in in has degree + 1.
If is a nondegenerate algebraic variety of dimension in of degree + 1,
then it is one of the following varieties:
(a) a quadric hypersurface;
(b) a cone over the surface 2,2 (2 ) 5 ;
(c) a rational normal scroll.
(See Degree of an algebraic subset, Scrolls, rational normal - and Veronese embedding for the relative definitions. We recall that for us any alegbraic variety is irreducible.)

Modules.
Definition. Let be a ring. We say that a set is a (left) -module if it is an Abelian
group for an operation + : and there is an operation : such
that
( + ) = + ,
( + ) = + ,
( ) = ( ) ,

for all , , , . (In general the symbol is omitted).

Moduli spaces

121

In some works, if is a ring with unity, denoted by 1 , it is also required that 1 =


for any . In other works it is not required, and an -module is said to be unital
(or unitary) if 1 = for any .

Moduli spaces.

([7], [99], [100], [105], [111], [194], [199], [244]). Roughly speaking,
the expression moduli space means variety parametrizing, i.e., a moduli space
is a geometric space whose points represent geometric objects of some fixed kind or
isomorphism classes of such geometric objects. For example, if is a vector space, the
Grassmannian (, ) (see Grassmannians) is the moduli space of the -planes in .
Let
: (Schemes) (Sets)

be a controvariant functor from the category of schemes to the category of sets that
associates to a scheme the set of (the equivalence classes of) the families of objects
over of a certain kind (see Categories, Schemes).

We say that a scheme is a fine moduli space for if is representable by , i.e.,


if is isomorphic to (, ).
In this case, there exists a universal family over , i.e., a family over , such that
every other family over a scheme is the pull-back of by a certain map : .
To get the universal family, take the element of () corresponding to the identity in
(, ).
Fine moduli spaces seldom exist. Therefore, we consider a weaker definition.

We say that a scheme is a coarse moduli space for if there is a natural transformation from to (, ) such that
(1) the induced map from ( ) to ( , ) = is a bijection (i.e., the set
of the points of is in bijection with the set of the equivalence classes of families
over a point);
(2) if is another scheme with the properties above, then there is a unique morphism : such that the following diagram of natural transformations
commutes
/ (, )
K
KK
nn
KK
nnn
KK
n
n
KK
nn
K%
wnnn
(, )
where (, ) (, ) is induced by .

For example, let [], where is a field. A flat family over a scheme of subschemes of with Hilbert polynomial is a closed subscheme such that,
if we denote the projection by , we have
(1) | : is flat;
(2) |1
() is a subscheme with Hilbert polynomial for any

122 | Monoidal transformations


(see Flat (module, morphism) and Hilbert function and Hilbert polynomial).
We can consider the functor
, : (Schemes) (Sets),

that associates to the set of the flat families over of subschemes of with Hilbert
polynomial . In 1961 Grothendieck proved that there exists a projective scheme which
is a fine moduli space for such a functor; it is called Hilbert scheme , . In 1966
Hartshorne proved that it is connected.
One of the most studied moduli spaces is the one of smooth curves with a fixed
genus. There is not a fine moduli space for them, but only a coarse one.

Theorem (Baily, Mumford, Deligne, Knudsen). ([7], [14], [54], [145]). The set M of the
isomorphism classes of smooth projective algebraic curves of genus over has a
structure of quasi-projective normal algebraic variety. If > 1, its dimension is 3 3.
There is a natural compactification M , consisting of the isomorphism classes of the
so-called stable curves of arithmetic genus (a stable curve is a curve such that the
only singularities are nodes and whose smooth rational components contain at least
three singular points of the curve; see Genus, arithmetic, geometric, real, virtual -
for the definition of arithmetic genus). M is a projective algebraic variety.
In general M and M are singular; the singularities come from the curves with nontrivial group of automorphisms.

A subject obviously linked to the one of moduli spaces is the one of deformations (see
Deformations). Another one is the geometric invariant theory (see Geometric invariant theory (G.I.T.)); in fact, the existence of nontrivial automorphisms of the objects which we want to classify causes trouble for the construction of moduli spaces,
so one can look for the moduli space of these objects with some additional structure
in such a way that the only automorphism is the identity; the moduli space of the original objects will be the quotient of the moduli space of the objects with the additional
structure by a group, and so one can reduce the original problem to a problem of geometric invariant theory.

Monoidal transformations.

In some books the term monoidal transformation is a synonym of blowing-up (see Blowing-up (or -process)). In other books,
it denotes the blowing-up of a point in a surface.

Morphisms. See Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps.

Multiplicity of a curve in a surface at a point. ([93], [107], [196], [228]). Let


be an algebraic closed field and be an algebraic curve in 2 ; let be a generator
of the ideal of . Let . If we take coordinates such that = (0, 0), and we write

Noetherian, Artinian

123

= with homogeneous polynomial of degree , we define the multiplicity of


at , which we denote by (), to be the minimum of the such that = 0. An
equivalent definition is the following: () is the maximum such that ,
where is the maximal ideal of the local ring O 2 , . Analogously, we can define the
multiplicity at a point of a curve (or, more generally, of an effective Cartier divisor) in
a smooth surface.
See [93, Chapter 0, 2.2] and [107, Chapter V, Example 3.4] for the general definition of
multiplicity of a subvariety at a point.

Multiplicity of intersection. See Intersection of cycles.

N
NakaiMoishezon theorem.
Nef.

See Bundles, fibre -.

See Bundles, fibre - or Divisors.

NeronSeveri group.

See Equivalence, algebraic, rational, linear -, Chow,


NeronSeveri and Picard groups.

Net.

A net is a linear system (see Linear systems) of dimension 2.

Node.

See Regular rings, smooth points, singular points.

Noetherian, Artinian.

([12], [62], [164], [256]). Let be a commutative ring with


unity and let be an -module. We say that is Noetherian (respectively Artinian)
if for every chain of submodules 1 , 2 , . . . such that
1 2

(respectively 1 2 ) of there exists such that


= +1 = ,

or, equivalently, if for every nonempty set of submodules of there exists a maximal
element for (respectively ).

Proposition. A module is Noetherian if and only if every submodule of is finitely


generated.

We say that the ring is Noetherian (respectively Artinian) if it is Noetherian (respectively Artinian) as -module, and thus if, for every chain of ideals 1 , 2 . . . such that
1 2 (respectively 1 2 ) of , there exists such that = +1 = .

124 | Noethers formula


Proposition.
(i) The direct sum of a finite number of Noetherian (respectively Artinian) -modules
is Noetherian (respectively Artinian).
(ii) If is a Noetherian (respectively Artinian) ring, every finitely generated -module
is Noetherian (respectively Artinian).
(iii) The quotient of a Noetherian (respectively Artinian) ring by an ideal is Noetherian
(respectively Artinian).
(iv) If is a Noetherian ring and is a multiplicative part of , then 1 is Noetherian
(see Localization, quotient ring, quotient field).
Hilbert basis theorem. If is a Noetherian ring, then [] is Noetherian.

Corollary. If is a Noetherian ring, then [1 , . . . , ] is Noetherian.

Corollary. If is a field, then [1 , . . . , ] is Noetherian (since every field is Noetherian).

Definition. We say that an ideal in a ring is irreducible if it satisfies the following condition: if is the intersection of two ideals, and , then or = either
= .
Theorem. In a Noetherian ring every ideal is the intersection of a finite number of
irreducible ideals and every irreducible ideal is primary. So, in a Noetherian ring,
every ideal has a primary decomposition (see Primary ideals, primary decompositions).
Theorem.
(i) In an Artinian ring every prime ideal is maximal.
(ii) Every Artinian ring is Noetherian.
(iii) Every Artinian ring is the direct sum of a finite number of local Artinian
rings.
Proposition. Let be a Noetherian ring. Then it is Artinian if and only if () is
finite and discrete (see Schemes for the definition of ).

Noethers formula.

See Surfaces, algebraic -.

Nondegenerate. We say that an affine or a projective algebraic variety is nondegenerate if it is not contained in any hyperplane.
Normal.

([107], [129], [228]). Let be an algebraically closed field. We say that an


algebraic variety over is normal if and only if for any the stalk O, of
O in is integrally closed, where O is the sheaf of the regular functions on (see

Normal crossing and log complex

125

Integrally closed and Varieties, algebraic -, Zariski topology, regular and rational
functions, morphisms and rational maps).
Proposition. An affine algebraic variety is normal if and only if its coordinate ring is
integrally closed.
A quasi-projective algebraic variety is normal if and only if every point has a normal
affine neighborhood.
Theorem. Smooth algebraic varieties are normal.
Theorem. The codimension of the set of singular points of a normal algebraic variety
is 2. In particular, an algebraic curve is normal if and only if it is smooth.

A normalization of an algebraic variety is a normal algebraic variety together


with a finite birational morphism .
Theorem. An affine algebraic variety has a (unique) affine normalization.
A quasi-projective algebraic curve has a quasi-projective normalization. If the curve is
projective, the normalization is projective.

(We recall that for us all algebraic varieties, in particular all algebraic curves, are irreducible).

Normal, projectively -, -normal, linearly normal. ([107]). Let be an


algebraically closed field and let be a projective algebraic variety in .
We say that is -normal if and only if the map
0 (, O ()) 0 (, O ())

given by the restriction is surjective, that is, if and only if the hypersurfaces of
degree cut on a complete linear system.
We say that is linearly normal if and only if it is 1-normal.
We say that is projectively normal if and only if the projective coordinate ring
is integrally closed (see Integrally closed).

We can prove that is projectively normal if and only if it is normal (see Normal)
and it is -normal for any . Thus a smooth projective algebraic variety is projectively
normal if and only if it is -normal for any .

Normal crossing and log complex. ([66], [93]). Let be a complex manifold
of dimension and a divisor on . We say that has normal crossings if the irreducible components of are smooth and meet transversally (i.e., for every point in
the intersection of some components and for any , any of the tangent spaces at of
these components intersect in a subspace of codimension ).

126 | , Property Let

() = ()

be the sheaf on of the meromorphic forms that are holomorphic on and


have poles on .
Let be a point such that exactly of the components of contain . Let 1 , . . . ,
be local coordinates in a neighborhood of such that = (0, . . . , 0) and locally is
given by 1 . . . = 0. Let () be the subsheaf of () generated by holomor
phic forms and logarithmic differentials for = 1, . . . , . The complexes ( (), )

and ( (), ) are used to study the singular cohomology of . The complex
( (), ) is called log complex.

Np , Property -.

See Syzygies.

Null correlation bundle.

relation bundle on
morphism

([207]). Let be an odd natural number. A null coris a bundle on that is the kernel of a surjective bundle
1,0 (1) O(1),

where 1,0 is the holomorphic tangent bundle on and O(1) is the hyperplane bundle
(see Hyperplane bundles, twisting sheaves). So there is an exact sequence
0 1,0 (1) O(1) 0.

By using Euler sequence, one can easily see that the Chern classes of a null correlation
bundle are the following: () = 1 for even, 1, () = 0 otherwise (see Euler
sequence and Chern classes).

Proposition. Every null correlation bundle is simple (see Simple bundles).


Theorem. For every odd, there exists a null correlation bundle on .

Observe that to find a surjective bundle morphism 1,0 (1) O(1) it is sufficient to
construct a section without zeroes of 1 (2) and then dualize.

O
O(s).

See Hyperplane bundles, twisting sheaves.

Orbit Lemma, Closed -. ([27], [126], [210], [235] 4.3). Let be an algebraically
closed field and be an algebraic group over (see Algebraic groups). Suppose that

Pfaffian

127

acts on an algebraic variety over , that is, there is a morphism


(, )

such that = and ( ) = ( ) for any , , , where is the identity element of . Then every orbit is a smooth variety and it is open in its closure.
Moreover, the boundary of any orbit is the disjoint union of orbits of lower dimension.
In particular, the orbits of minimum dimension are closed; therefore, there exists a
closed orbit.

Orientation. Let be a differentiable manifold. We say that is orientable if there


exists a local system of coordinates ( , ) (see Manifolds for the notation) such
that, for any , , the determinant of the Jacobian of 1 is positive in every
point where 1 is defined.
See Singular homology and cohomology for a topological approach.

P
Pencil.

A pencil is a linear system (see Linear systems) of dimension 1.

Pfaffian. ([188, Appendix C, Lemma 9], [211]). Let {0}. One can prove that,
if is a 2 2 antisymmetric matrix, then the determinant of can be written as
the square of a polynomial in the entries of . One of such polynomials is called the
Pfaffian of .
Precisely, if is a 2 2 antisymmetric matrix, we define the Pfaffian of , written
(), in the following way:
() := ()1 ,1 . . . , ,

where the sum is over all the partitions of {1, . . . , 2} into disjoint 2-subsets
{1 , 1 }, . . . , { , } and () is defined as follows: we can suppose < for = 1, . . . , ;
we define () to be the sign of the permutation sending 1, . . . , 2 respectively to
1 , 1 , . . . , , . It holds that
()2 = ().

Examples.

0
) = ,
0

(

0

) = + .

128 | Picard groups


Remark. The determinant of an antisymmetric square matrix with an odd number of
rows is 0. Let , be the matrix obtained from by taking off the -th row and the -th
column; the determinant of , is the product of the Pfaffian of , and of the Pfaffian
of , .

Picard groups. See Equivalence, algebraic, rational, linear -, Chow, NeronSeveri


and Picard groups.

Plurigenera. ([93], [107], [241]). Let be a compact complex manifold or a smooth


projective algebraic variety over a field. We define the plurigenera to be the numbers

() := 0 (,
),

for = 1, 2, . . . , where is the canonical sheaf (see Canonical bundle, canonical


sheaf).
See also Kodaira dimension (or Kodaira number).

Positive.

([93], [169]). Let be a complex manifold.


We say that a (1, 1) real form on is positive if ()( ) > 0 for any
and for any 1,0 (where 1,0 is the holomorphic tangent space of at ). In
other words, if we write
= , ()
,

in local holomorphic coordinates, the matrix (, ()), (which is a Hermitian matrix since is real) is positive definite for any .
We say that a holomorphic line bundle on is positive if in its first Chern class
1 () 2 (, ) there is a positive (1, 1)-form (see Bundles, fibre - and Chern
classes).

By Kodaira Embedding theorem (see Kodaira Embedding theorem), a holomorphic


line bundle on a compact complex manifold is positive if and only if it is ample.
Observe that if is a line bundle associated to a divisor (see Bundles, fibre - or
Divisors for the definition), then the positivity of is not equivalent to the effectivity
of and no implication is true:
To show is false, one can take on a Riemann surface a divisor of positive degree but
not effective;
to show is false, one can take the exceptional divisor of the blow-up of a point
in a surface: it is effective but the associated line bundle is not positive because, by
Nakai-Moishezon criterion (see Bundles, fibre -), it is not ample since 2 = 1.

Primary ideals, primary decompositions, embedded ideals.


[107]). Let be a commutative ring with unity.

([12], [62],

Primary ideals, primary decompositions, embedded ideals

129

Definition. We say that an ideal in is primary if implies or


(where is the radical ideal of ).
Proposition.
(a) Let be an ideal in .
If is primary, then is prime, more precisely is the smallest prime ideal
containing .
If is maximal, then is primary.
(b) Let 1 , . . . , be primary ideals such that 1 = = = ; then the intersection
1 is a primary ideal whose radical is .

Definition. A primary decomposition of an ideal is an expression of as a finite


intersection of primary ideals.
We say that an ideal is decomposable if it has a primary decomposition.
We say that a primary decomposition of an ideal ,
= =1,...,

is minimal if the radical ideals of the are distinct and =1,...,,= for any =
1, . . . , .
From a primary decomposition of an ideal we can easily get a minimal primary decomposition.

Theorem. Let be a decomposable ideal and let = =1,..., be a minimal primary decomposition. Let = . The prime ideals 1 , . . . , do not depend on the particular
decomposition of .

Definition. The prime ideals 1 , . . . , in the theorem above are said the prime ideals
associated to . The minimal ones (minimal for the inclusion) are called minimal
prime ideals, the others are called embedded prime ideals.

Proposition. In a Noetherian ring (see Noetherian, Artinian) every ideal has a primary
decomposition.

Let = [1 , . . . , ] for some algebraically closed field . Then the minimal prime
ideals associated to an ideal correspond to the irreducible components of the scheme
given by and the embedded prime ideals correspond to the subschemes of the irreducible components, the so-called embedded components. If a component is given
by only one point, we speak of embedded point.
Example. Let = (2 , ) in [, ]. Then = 1 22 where 1 = () and 2 = (, ).
Both 1 and 22 are primary, in fact: 1 is prime and thus primary, besides 22 = 2 is

maximal so 22 is primary. Thus the prime ideals associated to are 1 and 2 . Since

130 | Principal bundles


1 2 , we have that 1 is minimal and 2 is embedded. The variety defined by is the
line = 0 and the embedded ideal 2 corresponds to the point (0, 0).
See Schemes.

Principal bundles.
Process, -.

See Bundles, fibre -.

See Blowing-up (or -process).

Product, Semidirect -.

([164]). Let and be two groups and let


: ()

be a homomorphism, where () denotes the group of the automorphisms of . The


semidirect product of and with respect to , denoted by , is the following
group: the Cartesian product with the product
(1 , 1 ) (2 , 2 ) := (1 (1 )(2 ) , 1 2 ).

Observe that the subgroup

{(, )| }

(where is the identity element of ) is a normal subgroup of .

Remark. Let be a group and and two subgroups such that is normal, =
and = { }. Then
,
where : () is defined by

( 1 ) .

Proper.

A map from a topological space to another is proper if the inverse image of


any compact subset is compact. See Schemes for the definition of proper morphism
of schemes.

Proper mapping theorem.


Projective modules.

See Remmerts proper mapping theorem.

See Injective and projective modules.

Projective resolutions.

See Injective and projective resolutions.

Pull-back and push-forward of cycles. ([72],[74]). Let and be two varieties over an algebraically closed field and let : be a proper morphism. For
any subvariety of we define
deg(/()) = {

0
[() : (())]

if
if

() > (()),
() = (()),

Quadratic transformations, Cremona transformations

131

where () is the rational function field of , (()) is the rational function field
of () and [() : (())] is the degree of () as field extension of (()) (see
Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms
and rational maps for the definition of rational function field). In the complex case,
if () = (()), the map () is generically a finite sheeted covering and
we can also define deg(/()) to be the number of the sheets of the covering.
We define the push-forward of the cycle by , denoted by (), to be deg(/())
times the cycle (), i.e.,
() = deg(/()) ()

(see Cycles); we extend the definition of to all the cycles by linearity. One can
prove that the image through of any cycle that is rationally equivalent to 0 is rationally equivalent to 0. So induces a map, called again , between the Chow groups:
: () ()

for any (see Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and
Picard groups for the definition of Chow group).

Let and be two varieties and : be a flat morphism (see Flat (module,
morphism)). Suppose that there exists an integer such that, for any subvariety
of , all the irreducible components of 1 () have dimension ()+ (for instance
a projection from a vector bundle to its base or an open embedding).
For any subvariety of , we define the pull-back of the cycle , which we denote
by (), to be the cycle given by the inverse image scheme 1 () (see Schemes).
We extend the definition of to all the cycles by linearity. This map induces a map
between the Chow groups:
: () + ().

Pull-back and push-forward of sheaves.

sheaves.

See Direct and inverse image

Q
Quadratic transformations, Cremona transformations. ([93], [107],
[220], [228]). Let be an algebraically closed field. A quadratic transformation of
2 is a birational map : 2 2 defined in the following way:
Let , , be three noncollinear points of 2 ; let 2 be the blow-up of 2 at , , and
,

be the strict transforms of the lines , , ; denote by : 2 2
let ,

the induced morphism and let 1 be the inverse birational map from 2 to 2 ; let
: 2 2

132 | Quotient field


,
;
define
be the morphism given by the blow-down of ,
= 1 .

If we take = [1 : 0 : 0], = [0 : 1 : 0], = [0 : 0 : 1], the map is


[0 : 1 : 2 ] [1 2 : 0 2 : 0 1 ].

A Cremona transformation of 2 is a birational map of 2 .


A classical theorem by Noether asserts that a Cremona transformation of 2 is the
composition of a projective automorphism and of quadratic transformations (see [220,
Chapter V], for a proof).

Quotient field.

See Localization, quotient ring, quotient field.

R
Rank of finitely generated Abelian groups. ([164]). Let be a finitely generated Abelian group. The elements of of finite order form a group , called torsion
group, and one can easily prove that / is a free Abelian group. The minimal number
of generators of /, i.e., the cardinality of a basis of /, is called rank of .
The rank of a module is its rank as Abelian group.
Rational normal curves. ([104], [107]). Let be an algebraically closed field.
We say that a curve in is a rational normal curve if it is projectively equivalent to
the image of 1 embedded into by the map
[0 : 1 ] [0 : 01 1 : : 0 11 : 1 ],

i.e., of 1 embedded into by the -uple Veronese embedding (see Veronese embedding). One can prove easily that it is projectively normal (see Normal, projectively -, -normal, linearly normal), and that it is the zero locus of polynomials of
degree 2, in fact 1 embedded into by the -uple Veronese embedding is the zero
locus of the polynomials 1 +1 for 1 1, where 0 , . . . , are the
projective coordinates on .
Any rational normal curve has degree . One can show that any nondegenerate algebraic curve in has degree greater than or equal to and that any nondegenerate
algebraic curve in of degree is a rational normal curve (we recall that, for us, an
algebraic curve is irreducible).

Rational normal scrolls.

See Scrolls, rational normal -.

Rational functions, rational maps. See Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps.

Regular rings

133

Rational varieties. We say that an algebraic variety is rational if it is birational


to a projective space. See Varieties, algebraic -,Zariski topology, regular and rational
functions, morphisms and rational maps, Unirational, Lroth problem.
Reduced.

See Schemes.

Regular functions. See Varieties, algebraic -,Zariski topology, regular and rational functions, morphisms and rational maps.
Regular rings, smooth points, singular points. ([12], [62], [73], [104], [107],

[113], [114], [117], [156], [228], [246]).

Definition. Let (, ) be a Noetherian local ring of dimension (see Dimension,


Local, Noetherian, Artinian). We say that is regular if
where is the residue field /.

(/2 ) = ,

Now let be an algebraically closed field.

Definition. Let be an algebraic variety over . For any , denote by O, the


stalk in of the sheaf of the regular functions on and by the maximal ideal of
O, . We say that a point of is a smooth (or nonsingular or regular) point of
if O, is a regular ring, i.e., the dimension of /2 over is equal to the dimension
of O, (which is equal to the dimension of ). We say that a point is a singular point
(or a singularity) of if it is not a smooth point of . We say that is smooth if all
its points are smooth points of .
Proposition. If is an affine algebraic variety in and the ideal () of is generated by 1 , . . . , , we have that a point of is a smooth point of if and only if the

rank of the matrix ( )=1,...,,=1,..., in is equal to the codimension of .

Proposition. The set of the singular points of an algebraic variety is a proper closed
subset.

Definition. If and are two algebraic varieties over and and , we say
that in and in are analytically isomorphic if and only if the completions of
O, and O, are isomorphic as -algebras (see Completion).

Let be an affine plane algebraic curve and let be a generator of the ideal of . Let
. By changing coordinates, we can suppose = (0, 0). If we write = =0,..., ,
with homogeneous polynomial of degree , we define () to be the minimum
of the such that = 0 (see Multiplicity of a curve in a surface at a point). Obviously
is singular if and only if () > 1.

134 | Regular sequences


We now give some examples of singularities of multiplicity 2 on plane curves.

Fig. 13. A node (on the left) and a


cusp (on the right).

A node is a singularity analytically equivalent to the point (0, 0) in the curve = 0


(or, equivalently, in the curve 2 = 2 ), i.e., a point of multiplicity 2 with the tangent
cone given by two distinct lines (see Cone, tangent -). In other words, it is a point of
multiplicity 2 such that, by blowing it up, we have that its fibre in the strict transform
are two distinct smooth points; see Blowing-up (or -process).
A cusp is a singularity analytically equivalent to the point (0, 0) in the curve 2 = 3 ,
thus it is a point of multiplicity 2 with the tangent cone given by a line. If we blow it
up, we have that its fibre in the strict transform is a single smooth point.

A tacnode is a singularity analytically equivalent to the point (0, 0) in the curve 2 =


4 , thus it is given by two arcs meeting one other with multiplicity two.
Theorem. (See [35]). Let the characteristic of be different from 2. A singular point of
multiplicity 2 on a plane algebraic curve over is analytically equivalent the point
(0, 0) in the curve 2 = for some (unique) {0, 1}.

We say that an algebraic variety has a resolution of singularities if there exists a


nonsingular algebraic variety and a proper birational map from to . In [117],
Hironaka proved that every algebraic variety in characteristic 0 has a resolution of
singularities. For curves it is unique, while in higher dimension this is not true. There
are several methods to resolve singularities. See [113], [114], [156], [1].
One of the main tools to resolve singularities is the blow-up (see Blowing-up (or process), Genus, arithmetic, geometric, real, virtual -; see also Normal).

Regular sequences. ([62], [159], [185]). Let be a commutative ring with unity
and let be an -module.
An -regular sequence is a sequence of elements (1 , . . . , ) in such that
(1 , . . . , ) = and +1 is not a zerodivisor in /(1 , . . . , ) for = 0, . . . , 1.

Representations

135

Proposition. If (, ) is a Noetherian local ring (see Noetherian, Artinian and


Local), and (1 , . . . , ) is an -regular sequence contained in , then also
((1) , . . . , () ) is an -regular sequence for any permutation of {1, . . . , }.

Proposition. Let be a finitely generated module over a Noetherian ring . Let be an


ideal in such that = . Then any two maximal -regular sequences in have the
same number of elements (where we say that an -regular sequence (1 , . . . , ) in is
maximal in if there does not exist in such that (1 , . . . , , ) is -regular).
See Depth.

Regularity.

regular if

([62], [89], [198]). We say that a sheaf F on a projective space is (F( )) = 0 > 0.

CastelnuovoMumford theorem. Let F be a coherent sheaf on (see Coherent


sheaves). If F is -regular, then it is ( + 1)-regular.

We define the regularity of F to be min{| F is -regular}.


Greens theorem. Let F be a coherent sheaf on . If

0 E E1 E0 F 0

is a minimal free resolution of F (see Minimal free resolutions) and we write E =


O () , for some complex vector spaces , , then the regularity of F is
max{ | , = 0} .

Remmerts proper mapping theorem.

([93], [241]). Let and be two complex manifolds, and let : be a holomorphic map. Let be an analytic subvariety of such that | is proper (see Varieties and subvarieties, analytic - and
Proper). Then () is an analytic subvariety of .

Representations. ([76], [115], [130], [164], [225], [248]). A representation of a


group on a vector space is a homomorphism
: ().

Sometimes we say that is a -representation, and we write instead of ().

A representation of a Lie algebra (see Lie algebras) on a vector space is a Lie


algebra morphism
: (),

where () is the Lie algebra defined to be the vector space of the endomorphisms
of , endowed with the bracket [, ] = .

136 | Residue field


A representation of a Lie group (see Lie groups) on a finite-dimensional real
vector space is a morphism of Lie groups
: ().

We say that a representation of a group is irreducible if any -invariant subspace


of is equal either to {0} or to (where we say that a subspace is -invariant if
for any and for any ).
We say that a representation of a group is completely reducible if it is the direct
sum of irreducible representations.
Let and be two vector spaces over the same field, and let : be a linear map. If and are representations of a group , we say that is a -homomorphism or is -equivariant if () = () for all and for all .
Schurs lemma. Let be a group. Let be a -homomorphism between two irreducible
-representations, and , vector spaces over the field . Then, either is an isomorphism or = 0. Furthermore, if is algebraically closed, = and the dimension
of is finite, then = for some .

Maschkes theorem. Let be a field of characteristic 0. If is a finite group, any representation over is completely reducible.
Definition. Let be a vector space of finite dimension over a field and let be a
finite group. We define the character of a representation : () to be
: ,

where denotes the trace.

(() : ),

Theorem. Let be a field of characteristic 0 and let be a finite group. Then any
irreducible finite-dimensional -representation over is determined by its character.

Theorem. All the representations of the complex Lie groups ( ) and of ( ) are
completely reducible.

See Schur functors for a description of the representations of the complex Lie groups
( ) and of ( ). See also Lie algebras and Lie groups.

Residue field. Let be a local ring (that is, a ring with a unique maximal ideal)
and let be its maximal ideal. The field / is called the residue field of .
Resolutions.

See Exact sequences.

Riemann surfaces (compact -) and algebraic curves

Riemanns existence theorem.

137

([71], [189]).

Riemanns existence theorem. Let be a compact Riemann surface. Let be a finite


subset of , and {0}. The following sets are in a natural bijective
correspondence:
(i) the set of equivalences classes of connected -sheeted branched covering spaces
of with branch locus contained in ;
(ii) the set of connected equivalence classes of -sheeted topological covering spaces
of ;
(iii) the set of the elements of (1 ( , ), ) whose images are transitive subgroups of , where 1 denotes the first fundamental group and is the group of
permutations on elements.

(See Riemann surfaces (compact -) and algebraic curves, Covering projections and
Fundamental group for the definitions of such terms).
The equivalence between (ii) and (iii) can be described in the following way:
Let : be a topological covering; then 1 ( , ) acts transitively on
1 (), so we get a homomorphism from 1 ( , ) to whose image is a transitive
subgroup of . Conversely, let (1 ( , ), ) such that its image is a transitive subgroup of ; consider the topological covering space given by the following
subgroup of 1 ( , ):
= { 1 ( , )| ()(1) = 1}.

See Riemann surfaces (compact -) and algebraic curves for another Riemanns existence theorem.

RiemannRoch theorem.

See HirzebruchRiemannRoch theorem.

Riemann surfaces (compact -) and algebraic curves. ([8], [35], [69], [73],
[93], [101], [102], [107], [129] [189], [195], [196], [246]). A Riemann surface is a complex
manifold of (complex) dimension 1. An algebraic curve is an algebraic variety of dimension 1 (see Varieties, algebraic -, Zariski topology, regular and rational functions,
morphisms and rational maps).
From the topological viewpoint, a compact Riemann surface is a topological torus with
holes, for some unique 0, since it is an orientable, compact, real manifold of real
dimension 2 (see [184] for instance). The number is called the genus of the Riemann
surface.

Fig. 14. A topological torus with three holes.

138 | Riemann surfaces (compact -) and algebraic curves


Hurwitz formula. Let and be two compact Riemann surfaces of genus respectively
and , and let : a nonconstant holomorphic map; we have that
2 2 = deg()(2 2) + ( () 1),

where the numbers () are defined in the following way:


If we choose a local coordinate around a point and a local coordinate around ()
such that = { = 0} and locally () = , we define the ramification index () to
be (and we say that is a ramification point if () > 1).
We recall that, for any holomorphic line bundle , we denote by O() the sheaf of the
(holomorphic) sections of (see Bundles, fibre -).

Riemanns theorem. Let be a compact Riemann surface of genus . Let be the


canonical bundle (see Canonical bundle, canonical sheaf). We have
0 (, O( )) = .

Observe that for any holomorphic line bundle on a compact Riemann surface we
have that (, O()) = 0 for 2; to prove this, for instance we can apply the abstract
de Rhams theorem (see Sheaves) to the exact sequence

0 O() () 0,1 () 0,

where () is the sheaf of the sections of , , () is the sheaf of the (, )forms with values in and the map O() () is given by the inclusion.
A divisor on a compact Riemann surface is a finite formal sum

=1,...,

with and (see Divisors). We define the degree of , written deg(), in


the following way:
deg() := .
=1,...,

Let be a holomorphic line bundle on a compact Riemann surface . Any meromorphic section of gives a divisor on : the linear combination of the set of the zeroes
of with coefficients their multiplicities as zeroes minus the linear combination of the
set of the poles of with coefficients their multiplicities as poles. We define deg() to
be the degree of any divisor whose associated line bundle is (see Bundles, fibre -)
or, equivalently, to be the degree of the divisor given by any meromorphic section of .

RiemannRoch theorem. Let be a compact Riemann surface of genus and be a


holomorphic line bundle on . We have
0 (, O()) 1 (, O()) = () + 1 .

Riemann surfaces (compact -) and algebraic curves

139

Corollary 1. If is a compact Riemann surface of genus , then


deg( ) = 2 2 .

In fact, by Riemanns theorem, we have that 0 (, O( )) = and, by Serre duality


(see Serre duality), we have that 1 (O( )) = 0 (O), which is equal to 1, since any
holomorphic map from to is constant.

Remark 2. Let be a holomorphic line bundle of degree on a compact Riemann


surface of genus .
(i) If < 0, then obviously 0 (O()) = 0.
(ii) If = 0, then either 0 (O()) = 0 or O() = O.
(iii) If > 2 2, then obviously 1 (O()), which is equal to 0 (O( 1 )) by Serre
duality, is equal to 0.
(iv) If = 2 2, then either 1 (O()) = 0 or = .
Remark 3. Let be a compact Riemann surface and be a holomorphic line bundle
on . Let
: (0 (, O())

be the map associated to (see Bundles, fibre -). We have that


the map is holomorphic, i.e., is base point free (see again Bundles, fibre -
for the definition), if and only if 0 (O( )) = 0 (O()) 1 for any , where
O( ) is O() O() and O() is the sheaf of the sections of the holomorphic
bundle () associated to the divisor and analogously in the sequel;
the map is injective if and only if 0 (O( )) = 0 (O()) 2 for any ,
with = ;
the differential of is injective if and only if 0 (O( 2)) = 0 (O()) 2 for any
.
By using the previous remark and the RiemannRoch theorem, we can prove

Theorem 4. Let be a compact Riemann surface of genus and be a holomorphic


line bundle on of degree .
(i) If 2, then is holomorphic.
(ii) If 2 + 1, then is an embedding.

Riemanns existence theorem. The image of a compact Riemann surface by an embedding in a projective space is an algebraic curve.
(See also Chows theorem and Remmerts proper mapping theorem).
Therefore, we can embed every compact Riemann surface in a projective space (by
Theorem 4) and the image is an algebraic curve. Vice versa we can associate to

140 | Riemann surfaces (compact -) and algebraic curves


any algebraic curve in a projective space over a Riemann surface: the desingularization of , which is equal to the normalization of (see Normal and Regular rings, smooth points, singular points). We have seen that every compact Riemann
surface can embedded in a projective space, but one can prove a stronger result:
Proposition 5. Any compact Riemann surface can be embedded in 3 .

Definition 6. Let be a compact Riemann surface of genus 2. We say that is


hyperelliptic if and only if there exists a holomorphic map 1 of degree 2.
Remark 7. A compact Riemann surface of genus 2 is hyperelliptic if and only if
there exist , such that
0 (O( + )) = 2,

i.e., if and only if there is a 21 , i.e., a linear system of degree 2 and projective dimension 1 (see Linear systems).

Let be a holomorphic line bundle of degree on a compact Riemann surface of


genus . By Remark 2 and by the RiemannRoch theorem, in the case < 0 and in
the case > 2 2 we can calculate 0 (O()) and 1 (O()). In the case 0 2 2
the following theorem can help.
Cliffords theorem. Let be a holomorphic line bundle of degree on a compact Riemann surface of genus such that 0 (O()) > 0 and 1 (O()) > 0. Then
0 (O()) 1

,
2

If we have the equality 0 (O()) 1 = 2 , then we are in one of the following cases:
(i) is the trivial bundle;
(ii) = ;
(iii) X is hyperelliptic and the linear system given by is a 21 .

Cliffords index. Let be a compact Riemann surface and be a holomorphic line


bundle on ; we define
In addition we define

Cliff() = deg() 20 (O()) + 2.

Cliff() = min{()| s.t. 0 (O()) 2, 1 (O()) 2} .

Hence, by Cliffords theorem, we can say that the Cliffords index of , when 0 (O()) >
0 and 1 (O()) > 0, is a measure of how special is; precisely it is greater than or equal
to 0 and it is equal to 0 only in special cases.
We want now to investigate the map .

Riemann surfaces (compact -) and algebraic curves

141

Remark 8. Let be a compact Riemann surface of genus and . We have


1 0 (O()) 2

and, if 0 (O()) = 2, then = 0 and is biholomorphic to 1 . In other words, if 1,


then 0 (O()) = 1. By using this, Remark 3, and the RiemannRoch theorem, one can
prove easily the following statements:
(i) if 1, then is base point free, i.e., is holomorphic;
(ii) if 2 and is not hyperelliptic, then is an embedding.
Geometric RiemannRoch theorem. Let be a compact Riemann surface of genus
2 and be an effective divisor on . We denote by the subspace of
(0 (, O( )) ) generated by , precisely the intersection of the hyperplanes

in (0 (, O( )) ) such that () or the divisor


() is effective. Then

we have
= deg 0 (O()) .

Corollary. Let be a Riemann surface of genus 1. We have


0 (O()) = {

1
+1

if
if

0 ,
,

if
if

0 ,
.

for a general effective divisor of degree .

In particular, for general , we have


0 (O()) = {

1
+1

The points that do not have the general behaviour are called Weierstrass points (see
Weierstrass points).
Castelnuovos theorem. Let be a nondegenerate smooth projective algebraic curve
in of genus and degree (see Degree of an algebraic subset). Let be the
integer part of 1
and = 1 ( 1). Then
1
( 1)

( 1)
+ .
2

Now let us consider algebraic curves over algebraic closed fields. The notion of genus
of a Riemann surface is replaced by the ones of geometric genus and arithmetic
genus (see Genus, arithmetic, geometric, real, virtual -). We have:
Proposition. Let be a smooth complete algebraic curve over an algebraic closed field.
Then () = () = 1 (, O ).
We will call this number simply genus of .

142 | Saturation
For smooth complete algebraic curves over an algebraic closed field, the theory is
very similar to the one of Riemann surfaces. Precisely, by replacing holomorphic line
bundles with algebraic line bundles and holomorphic maps with morphisms in the
RiemannRoch theorem, in Cliffords Theorem, and in the corollaries, remarks, propositions. and definitions above from 1 to 8, we get the analogous statements for any
smooth complete curve over an algebraic closed field.
As to the Hurwitz formula we have the following theorem (see Varieties, algebraic -,
Zariski topology, regular and rational functions, morphisms and rational maps for
the definition of rational functions field):
Hurwitz formula. Let and be two smooth complete algebraic curves over an algebraic closed field of characteristic 0 and let and be their respective genera.
Let : be a finite morphism such that () is a separable field extension of
(), where () and () are the rational functions fields respectively of and .
We define deg() to be the degree of () as field extension of (). Then
2 2 = deg()(2 2) + ( () 1),

where the numbers () are defined as follows: let be a uniformizing parameter of


the discrete valuation ring O() (see Discrete valuation rings), let be the image of
through the map from O() to O induced by ; let : O the valuation map
of the discrete valuation ring O ; we define () = ().

See also Genus, arithmetic, geometric, real, virtual -, Jacobians of compact Riemann surfaces.

S
Saturation.

([104], [107]). Let = [0 , . . . , ] for some algebraically closed field


and let be a homogeneous ideal of ; the saturation of , we denote by (), is
defined as follows:
() := { | (0 , . . . , ) for some },

where (0 , . . . , ) is the ideal generated by 0 , . . . , .


We say that is saturated if = ().

Remark. For any homogeneous ideal of , we have


(i) () is homogeneous;
(ii) and () coincide from a certain degree on;
(iii) in the zero locus of and the zero locus of () coincide; moreover, and
() determine the same ideal locally, precisely, for any = 0, . . . , and for any

homogeneous polynomial , let be the polynomial deg()


in the variables :=

Schemes

143

/ for = 0, . . . , , = ; we have that, for any = 0, . . . , , the ideal generated


by the for homogeneous polynomial of is equal to the ideal generated by the
for homogeneous polynomial of ();
iv) two homogeneous ideals give the same closed subscheme of () if and only
if they have the same saturation and there is a bijection between saturated ideals
of and closed subschemes of () (see Schemes).
Thus, if two ideals have the same saturation, not only do they have the same zero locus
(that is, they cut out the same variety set-theoretically), but they also determine the
same ideals locally, and they determine the same subscheme (i.e., they cut out the
same variety scheme-theoretically).

Schemes.

[107].

([12], [64], [95], [107], [129], [228]). We strictly follow the exposition in

Definition. Let be a commutative ring with unity.


We denote by () the set of the prime ideals of , endowed with following
topology:
For any ideal of , we define

() = {| prime ideal of , };

the closed subsets of () are the () for varying in the ideals of .


Let O() (O for short) be the sheaf of rings on () defined in the following
way:
For any open subset of (), let
O() = { : | section, locally quotient of elements of },

where
denotes the localization of in (see Localization, quotient ring, quotient field);
section means that () for any ;
locally quotient of elements of means that, for all , there is a
neighborhood of in and 1 , 2 such that for each
and 2 .

( ) =

1
2

The spectrum of is the pair ((), O() ). Sometimes we use the notation
() to denote the spectrum.

144 | Schemes
Definition.
A ringed space (, O ) is a topological space with a sheaf of rings O . We say
that it is a locally ringed space if the stalk of the sheaf in every point is a local
ring (see Local).
A morphism of ringed spaces from a ringed space (, O ) to a ringed space
(, O ) is given by a continuous map : and a morphism of sheaves
O O (see Direct and inverse image sheaves for the definition of O ).
A morphism of locally ringed spaces is a morphism of ringed spaces such that
the maps induced on the stalks are local homomorphisms of local rings (i.e.,
homomorphisms such that the inverse image of the maximal ideal of the second
ring is the maximal ideal of the first ring).
Proposition.
(i) Let be a commutative ring with unity and let ((), O) be its spectrum. Then
(a) for every (), the subset {} in () is closed if and only if is
maximal;
(b) for every (), the closure of the subset {} is ();
(c) O for every ();
(d) O(()) = .
In particular, the spectrum of a ring is a locally ringed space.
(ii) Let and be two commutative rings with unity. A homomorphism of rings
induces a morphism of locally ringed spaces
((), O() ) ((), O() ).

Any morphism of locally ringed spaces ((), O() ) ((), O() ) is


induced by a homomorphism .

Definition. An affine scheme is a locally ringed space isomorphic to the spectrum


of a ring. A scheme is a locally ringed space (, O ) locally isomorphic to an affine
scheme, i.e., such that for every point of there exists a neighborhood such that
(, O | ) is an affine scheme. The sheaf O is called structure sheaf. A morphism of
schemes is a morphism of locally ringed spaces.

Example. Let be an algebraically closed field. Then ([1 , . . . , ]) is denoted by


(as the affine space of dimension over ) and called affine space of dimension
over (scheme): in fact the closed points of ([1 , . . . , ]) are the maximal ideals of [1 , . . . , ], which are the ideals of [1 , . . . , ] of the kind (1 1 , . . . , )
for some 1 , . . . , ; thus they correspond to the points of the affine space of dimension over . Obviously ([1 , . . . , ]) also contains the points corresponding to the prime ideals of [1 , . . . , ] that are not maximal; the closure of such a
point is all () (which contains all the points that are in the zero locus of the ideal
in the affine space); is said to be a generic point for (). In particular the clo-

Schemes

145

sure of the zero ideal is all ([1 , . . . , ]), and so it is called the generic point of
([1 , . . . , ]).
Definition. Let = 0 be a graded commutative ring with unity.
We define () to be the set

{| homogeneous prime ideal of , >0 }.

endowed with the following topology: the closed subsets are the () for homogeneous ideal of , where

() := { ()| }.

Let O() (O for short) be the sheaf of rings on () defined in the following
way: for any open subset of () let
O() = { : () | section, locally quotient of elements of },

where
() is the set of the elements of degree 0 in the localization 1 where =
{ | homogeneous, };
section means () () for all ;
locally quotient of two elements of means that, for all , there
exists a neighborhood of in and there exist 1 , 2 homogeneous
with (1 ) = (2 ) such that for all
in ( ) and 2 .

( ) =

1
2

Proposition. Let be a graded commutative ring with unity. We have


(a) O () for every ();
(b) let >0 ; then the set
:= { ()| }

is an open subset of () and ( , O() | ) is isomorphic, as locally ringed


space, to the spectrum of () (the subring of the elements of degree 0 of the localized ring ). The for >0 cover (). Hence ((), O() ) is
a scheme.
Sometimes we use the notation () to denote this scheme.

Example. Let be an algebraically closed field. Then ([0 , . . . , ]) is denoted


by (as the projective space of dimension over ) and called projective space of

146 | Schemes
dimension over (scheme): in fact, the closed points of ([0 , . . . , ]) correspond to the points of the projective space of dimension over .

Definition. Let be a scheme. A scheme over is another scheme and a morphism


of schemes . If is a ring, a scheme over is a scheme over (). If and
are two schemes over , an -morphism from to is a morphism such that
the morphism is the composition of the morphism with the morphism
.

Let be an algebraically closed field. One can associate to any algebraic variety
over (see Varieties, algebraic -, Zariski topology, regular and rational functions,
morphisms and rational maps) a scheme () over such that is homeomorphic
to the subset of the closed points of () and the sheaf of the regular functions on is
obtained by restricting the structure sheaf O() to the subset of closed points identified
with . The definition of is as follows:
define () to be the set of the nonempty irreducible closed subsets of ;
define () to be the set of the nonempty irreducible closed subsets of for any
closed subset of ; endow () with the topology such that a subset of () is
closed if and only if it is equal to () for some closed subset of ;
define : () to be () := {} for all ; let O be the sheaf of regular
functions on .
One can prove that () with the sheaf (O ) is a scheme with the properties we want.
Moreover, we can prove that the map is a full faithful functor (see Categories) from
the category of algebraic varieties over to the category of schemes over .
Definition. We say that a scheme is irreducible (respectively connected) if and only
if the corresponding topological space is irreducible (respectively connected).
Let be a scheme and let O be its structure sheaf. We say that is reduced if O() has
no nilpotent elements for any open subset of . This is equivalent to the condition
that the stalk O has no nilpotent elements for any .
We say that is integral if O() is an integral domain for any open subset of ; one
can prove that this holds if and only if is reduced and irreducible.
We say that is locally Noetherian if it can be covered by open affine subsets ( )
with Noetherian rings.
We say that is Noetherian if if it can be covered by a finite number of open affine
subsets ( ) with Noetherian rings.
Let : be a morphism of schemes. We say that is respectively
locally of finite type,
of finite type,
finite,
if there exists an open affine covering {( )} of such that, respectively,
there exists an open affine covering of 1 (( )), {( , )} with , finitely
generated -algebra,

Schemes

147

there exists a finite open affine covering of 1 (( )), {( , )} with ,


finitely generated -algebra,
1 (( )) = ( ) for some -algebra that is a finitely generated module.

We say that a morphism of schemes : is affine if there is an open affine


covering { } of such that 1 ( ) is affine for every .

An open subscheme of a scheme is an open subset of the topological space of


with the induced topology and with the sheaf given by the restriction of the structure
sheaf of to . We can prove that it is a scheme.
An open embedding of schemes is a morphism of schemes : that induces
an isomorphism of with an open subscheme of .
A closed embedding of schemes is a morphism of schemes : inducing a
homeomorphism between the topological space of and a closed subset of the topological space of and the induced map of sheaves O O is surjective.
A closed subscheme is an equivalence class of closed embeddings, where two closed
embeddings : and : are equivalent if there is an isomorphism
: such that = .
We say that a scheme over is projective if it is a closed subscheme of the scheme
= ([0 , . . . , ]) for some .
Remark. Let be a ring and let be an ideal of . The homomorphism / induces
a closed embedding from the spectrum of / to the spectrum of and the image of
the map between the topological spaces is (). Thus any ideal induces a structure
of closed subscheme on ().
For instance, the ideals (), (2 ), (2 , ) in [, ] have the same zero locus (the
-axis) in the affine plane over , but induce different structures of closed subscheme
on the -axis. Observe that the ideal (2 ) gives a subscheme such that for every point
of -axis the stalk O has nilpotent elements, while the ideal (2 , ) gives a subscheme such that only the stalk of the sheaf in the origin has nilpotent elements (in
this case we say that the origin is an embedded point; see Primary ideals, primary
decompositions, embedded ideals).

Definition.
Let 1 and 2 be two schemes over another scheme . The fibred product of 1
and 2 over , denoted by 1 2 , is a scheme with morphisms : 1 2
for = 1, 2 such that, if : are the given morphisms, we have 1 1 = 2 2
and, given another scheme over and morphisms : for = 1, 2 such
that 1 1 = 2 2 , there exists a unique morphism : 1 2 such that
= for = 1, 2. We can prove that the fibred product exists and that it is
unique up to isomorphisms.

148 | Schur functors

Given a morphism of schemes : and a subscheme of , we define


the inverse image scheme of through to be the fibred product . Its
underlying topological space is homeomorphic to 1 (). Sometimes the inverse
image scheme of by is denoted by 1 ().
A morphism of schemes : is said to be separated (and is said to
be separated over ) if the diagonal morphism is a closed embedding, where the diagonal morphism is the unique morphism whose
composition with both the projections is the identity. We say that a
scheme is separated if it is separated over ().
A morphism of schemes : is said to be proper if it is separated, of finite
type, closed and, for any morphism , the induced morphism
is closed (closed means that the image of any closed subset is a closed subset).
A morphism of schemes : is said to be projective if it is the composition
of a closed embedding () and the projection ()
(where = ([0 , . . . , ])).

We can prove that the properties of separatedness and properness for schemes, the
definitions of which may sound a bit strange, correspond respectively to the Hausdorff property and to the usual notion of properness through the functor from the category of the schemes of finite type over to the category of complex analytic spaces
(see G.A.G.A.).

Theorem. The image of the map described above from the category of algebraic varieties over an algebraically closed field to the category of schemes over is the set of
the quasi-projective integral schemes and the image of the set the projective algebraic
varieties is the set of the projective integral schemes.

Schur functors. ([75], [76]). Let be a natural number and let = ( 1 , . . . , ) be a


partition of , that is, let for = 1, . . . , with 1 + + = and 1 .
We consider ( 1 , . . . , ) and ( 1 , . . . , , 0, . . . , 0) the same partition and usually we
dont write the zeroes.
We can associate to a diagram, which we call the Young diagram of , with boxes
in the -th row for = 1, . . . , and the rows lined up on the left.
Fig. 15. Young diagram of (4, 3, 1).

The conjugate partition is the partition of whose Young diagram is obtained from
the Young diagram of by interchanging rows and columns.

Schur functors

149

A tableau with entries in {1, . . . , } on the Young diagram of a partition = ( 1 , . . . , )


of is a numbering of the boxes by the integers 1, . . . , , allowing repetitions (we also
say that it is a tableau on ).
(a)

(b)

2
8
1

6 1
4

Fig. 16. (a) The Young diagram of (3, 2, 2, 1), conjugate of (4, 3, 1); (b) a tableau on (4, 3, 1).

Definition. Let be a complex vector space of dimension . Let , and let


= ( 1 , . . . , ) be a partition of . Number the boxes of the Young diagram of with
the numbers 1, . . . , from left to right, beginning from the top row; let be the group
of permutations on elements. Let be the subgroup of given by the permutations preserving the rows, and let be the subgroup of given by the permutations
preserving the columns. We define
:= (

() : ) ,

where stands for image and () is the sign of the permutation . The are
called Schur representations.

In particular, denotes the symmetric product of , and (1,...,1) denotes the alternating product. We can prove that is isomorphic to
(

() : ) .

Theorem. Let be a complex vector space of dimension . For every = ( 1 , . . . , )


with for = 1, . . . , and 1 , is an ()-irreducible representation and all the (complex) irreducible ()-representations are of this form.
Proposition. Let be a complex vector space of dimension . Let = ( 1 , . . . , ) with
for = 1, . . . , and 1 . The dimension of is 0 if +1 = 0, and is
if = 0 for + 1.

1<

Observe that, for any = ( 1 , . . . , ) with for = 1, . . . , and 1 and


for any ,
( 1 +,..., +) ( 1 ,..., )

150 | Schur functors


as ()-representations. In particular, as ()-representation,
( 1 ,..., ) ( 1 ,..., 1 ) .

Furthermore, for every = ( 1 , . . . , ) with for = 1, . . . , and 1 ,


the space is also a ()-irreducible representation. Observe that, for all ,
( 1 +,..., +) ( 1 ,..., ) ( )

as ()-representations. By defining ( ) = (( ) ) if negative, this allows us


to define ( 1 ,..., ) for every = ( 1 , . . . , ) with 1 , not necessarily greater
than or equal to 0.
Every (complex) irreducible ()-representation is equal to ( 1 ,..., ) with 1
, for some 1 , . . . , .
The following theorem, called the LittlewoodRichardson rule, shows the decomposition in irreducible components of the tensor product of two irreducible representations of (). To state it, we need the following definition:

A = (1 , . . . , )-expansion of a Young diagram is obtained by first adding 1 boxes,


not two in the same column, with a 1 in each box, then adding 2 boxes, not two in
the same column, with a 2 in each box, and so on. We say that the expansion is strict
if the list of the numbers in the boxes we add, read from the top row to lowest one and
every row from right to left, has the following property: for every , in the first entries
of the list, the number of the is greater than or equal to the number of the + 1 for
every = 1, . . . , 1.

LittlewoodRichardson rule. Let , . For every = ( 1 , . . . , ) partition of and


= (1 , . . . , ) partition of , we have
= partition of

,, ,

as ()-representation, where ,, is the number of ways in which the Young diagram of can be obtained from the Young diagram of by a strict -expansion.

For instance, if is a partition of a natural number and is a natural number, the


Young diagrams of the irreducible components of are obtained from the
Young diagram of adding boxes not two in the same column and all the multiplicity
are 1 (Pieris formula).

Finally we observe that, if () = , ( 1 ,..., ) is isomorphic as ()representation to 1 , 1 1 ,..., 1 2 . Moreover ( ) .

Now we want to describe the irreducible representations of (the group of permutations on elements).
Let , and let = ( 1 , . . . , ) be a partition of . As before, number the boxes of
the Young diagram of with the numbers 1, . . . , from left to right beginning from the

Scrolls, rational normal -

151

top row. Let be the subgroup of given by the permutations preserving the rows
and let be the subgroup of given by the permutations preserving the columns.
We define
:= () ,
,

where is the group algebra associated to (i.e., the algebra whose underlying
vector space is , endowed with the product = ). Let
= .

Theorem. The only irreducible representations of are the for partition


of .

Scrolls. A scroll is a projective bundle (i.e., a bundle whose fibres are projective
spaces) embedded into a projective space in such a way that the fibres are projective
subspaces of .
Scrolls, rational normal -.

any .
Let , , 0 , . . . , {0} with

([93], [104]). Let be a field and let denote for

0 and

= + 1.

=0,...,

Let , for = 0, . . . , , be complementary subspaces and be (nondegenerate) rational normal curves in (see Rational normal curves); let : 0 ,
for = 1, . . . , , be isomorphisms. Define
0 ,..., = 0 , 1 (), . . . , (),

where , 1 (), . . . , () is the minimal subspace containing , 1 (), . . . , ().


The variety 0 ,..., is called a rational normal scroll. Its dimension is obviously + 1. It
is determined up to projective equivalence by the integers .
Caution! The name may be somewhat misleading: it is not true that if a scroll is rational
and normal, then it is a rational normal scroll.

Example of a rational normal scroll. 1 embedded into ++ = ((+1)(+1) ) by


the bundle O1 (, 1), i.e., by the map
([0 : 1 ], [0 : : ])
[0 0 : 01 1 0 : : 1 0 : : 0 : 01 1 : : 1 ].

152 | Segre classes


The above is , = for any , and the are:

0 = {[0 0 : 01 1 0 : : 1 0 : 0 : : 0]},
......

= {[0 : 0 : 0 : 01 1 : : 1 ].

See Minimal degree.

Segre classes.

([72], [74]). Let be a holomorphic vector bundle of rank on a


complex smooth projective algebraic variety of dimension ; let ()() be the Chern
polynomial (see Chern classes)
1 + 1 () + 2 ()2 + .

We define the Segre polynomial

()() = 1 + 1 () + 2 ()2 +

to be the polynomial such that

()() ()() = 1.

Thus 1 () = 1 (), 2 () = 1 ()2 2 (), . . . . The element () 2 (, ) is called


the -th Segre class of .
We have that
( ()) = ((1 ()1+ )),

where
is the Poincar duality (see Singular homology and cohomology);
is the projection () and is the map induced by in homology;
= O() (1), i.e., is the line bundle on () whose restriction to ( ) (
fibre on ) is O(1) for all (i.e., is the dual of the tautological bundle; see
Tautological (or universal) bundle).
The Segre classes can be defined also for cones and subvarieties, see [72], [74].

Segre embedding.

([104], [107]). Let be an algebraically closed field and, for


every , let = be the projective space over of dimension .
Let , . The Segre map on is the map
, : (+1)(+1) 1 ,

([0 : : ], [0 : : ]) [ : : ],

i.e., the map sending ([0 : : ], [0 : : ]) to the point of (+1)(+1) 1 whose


coordinates are all the products for = 0, . . . , , = 0, . . . , . Obviously, the map
, is an embedding (see Embedding).

Serre correspondence

153

Let , for = 0, . . . , , = 0, . . . , be the coordinates on (+1)(+1) 1 ; the image of the


Segre map , is the zero locus of the polynomials of degree 2:
, , , , ,

with , {0, . . . , }, , {0, . . . , }. It is a smooth variety of dimension + and


) (see Degree of an algebraic subset), called Segre variety. It is a deterdegree (+

minantal variety (see Determinantal varieties), since it is the zero locus of the determinants of the (2 2)-submatrices of the matrix (, ), .
A coordinate-free way to describe the Segre map , is the following: Let be a vector
space over of dimension + 1 and be a vector space over of dimension + 1;
the map , is the map
() () ( ),
(, )

for , , where () is the line generated by .


One can define also the Segre map on the product of more than two projective spaces
in the obvious way: let 1 , . . . , ; the map 1 ,..., is the map
1 (1 +1)...( +1) 1 ,

([01 : : 11 ], . . . , [0 : : ]) [ : 11 : ],

where {0, . . . , } for = 1, . . . , (the point on the right is the point whose coordinates are all the products 11 , where {0, . . . , } for = 1, . . . , ).

Semicontinuity theorem.

See Grauerts semicontinuity theorem.

Serre correspondence.

([64], [107], [223]). Let be a graded commutative ring


with unity and let be the scheme () (see Schemes).
For any graded -module , let F be the following sheaf on : for every ,
let () be the group of the elements of degree 0 in the localization of with
respect to the multiplicative system of the homogeneous elements in (see
Localization, quotient ring, quotient field); for every open subset of , let
F () be the set of the functions
: ()

such that () () for every and is locally a fraction, that is, for all
, there exists a neighborhood of in and homogeneous elements ,
of the same degree such that, for all , we have and ( ) = .
The functor F is called sheafification.

For any sheaf F of O -modules on , let

F := 0 (, F()).

154 | Serre duality


We can make this a graded -module, in fact, given 0 (, F()) and
, which gives an element in 0 (, O ()) (see Hyperplane bundles, twisting sheaves), we can easily define their product 0 (, F( + )) by using
the isomorphism F() O () F( + ).

Proposition. Let be a field and let = [0 , . . . , ]. For any quasi-coherent sheaf F


on = () (see Coherent sheaves), we have
FF F .

It is not true that for any finitely generated graded -module we have that
F , it is only true that F and are isomorphic from a certain degree on.
We say that two graded -modules are equivalent if they are isomorphic from a certain
degree on. We say that a graded -module is quasi-finitely generated if it is equivalent
to a finitely generated graded -module. The functors F , F F induce an
equivalence of categories between the following categories:
the category of the graded quasi-finitely generated -modules modulo the equivalence above, where = [0 , . . . , ];
the category of coherent sheaves of O -modules.
See [107, Chapter II, 5 and Example 5.9] for more general statements.

Serre duality.

([93], [107], [224]).

Theorem. Let be a compact complex manifold of dimension and a holomorphic


vector bundle on . The following isomorphism holds:
(, ()) (, ( )) .

The isomorphism above is called Serre duality. A completely analogous statement


holds for a vector bundle on a smooth projective algebraic variety of dimension over
an algebraically closed field. See Dualizing sheaf.

Serres theorems A and B.

See CartanSerre theorems.

Sheafify.

See Serre correspondence.

Sheaves.

([34], [84], [93], [103], [107], [119], [146], [152], [223], [228], [234]).

Definition. A presheaf of Abelian groups F on a topological space is the datum of


a map associating to any open subset of an Abelian group F() and, for any ,
open subsets of with , a homomorphism (called restriction map) : F()

Sheaves

155

F() such that


(a) F(0) = {0};

(b) = for every open subset of ;

=
.
(c) if , then

Thus a presheaf of Abelian groups is a controvariant functor from the category of the
open subsets of with the inclusion maps to the category of Abelian groups.
Obviously we can consider also presheaves with values in an other category, say C,
i.e., such that F() is an object of C for any open subset of (for instance presheaves
of modules over some ring).
For any open subsets and of with and any F(), we often denote
() by | .
For any open subset of , the group F() is often denoted also by (, F). Its elements are called sections of F over . The elements of F() are called global sections.

Definition. We say that a presheaf F on a topological space is a sheaf if and only if


the two following conditions hold:
(1) (Locality) Let be an open subset of and { } be an open covering of . If
, F() and | = | for all , then = .
(2) (Gluing) Let be an open subset of and { } be an open covering of . Let
F( ) for all such that | = | ; then there exists F() such
that = | .
Let . The stalk of a presheaf F on is
F =

lim

F()

(see Limits, direct and inverse -). We denote the map from F() to F by .
Let
F = F .

We can associate to any presheaf F on a topological space , a sheaf F on in the


following way: let F be the sheaf
() F ;

{
{

{ : F , , neighborhood of , and

{
{

F() s.t. () = ()

}
}
}
}
}
}
}

for any open subset .


Equivalently we can define F in the following way: endow F with the strongest topology such that, for any open subset and for any F(), the map F , ()
is continuous (with this topology, F is called espace tal of F); define F to be the

156 | Sheaves
following sheaf:
{ : F | () F , continuous}.

If F is a sheaf, then F = F .

Examples.
Let be a topological space and let be an Abelian group. Endow with the
discrete topology.
We define a sheaf (or simply ) on in the following way: For any open
subset , let () be the group of the continuous functions from to and
let the restriction maps be the usual restriction maps. We call a constant
sheaf. For instance, we can take = , , . . . .

If is a point of , the skyscraper sheaf on is defined to be the sheaf


such that, for any open subset , () is if , while it is 0 if .

Let be a manifold. We define the sheaves , , and, for any bundle


on , the sheaves () and , () in the following way:
() is the group of the functions from to ;
, () is the group of the (, )-forms on ;
()() is the group of the sections of on ;
, ()() is the group of the (, )-forms on with values in ,
,

, , () , , ()
for any open subset . We should write respectively
when the manifold is not clear from the context.
Let be a complex manifold. We define the sheaf O, the sheaf , and for any
holomorphic bundle on , the sheaf O() in the following way:
O() is the group of the holomorphic functions from to ;
() is the group of the holomorphic -forms on ;
O()() is the group of the (holomorphic) sections of on ;

for any open subset . We should write respectively O , , O() when the manifold is not clear from the context.

If is an algebraic variety, O denotes the sheaf of the regular functions (see Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms
and rational maps). If is an (algebraic) vector bundle on , we denote by O()
the sheaf on associating to any open subset of the group of the sections of
on .

A subsheaf of a sheaf F is a sheaf G such that, for any open subset of , G() is a
subgroup of F() and the restriction maps of G are induced by the ones of F.

Sheaves

157

Definition. Let F, G be two sheaves on a topological space . A morphism of sheaves


: F G is the datum of a homomorphism
: F() G()

for any open subset of such that, if and are open subsets with and
is the restriction map of F and is the restriction map of G, the diagram
F()


F ()

/ G()


/ G()

commutes. Denote by (F, G) the set of the morphisms from F to G.

Obviously, a morphism of sheaves : F G induces a map : F G for any


.
We say that is injective if and only if : F G is injective for all and this is true
if and only if the maps : F() G() are injective for any open subset of .
We say that is surjective if and only if : F G is surjective for all . This is not
equivalent to requiring that the maps : F() G() are surjective for any open
subset of . For instance consider the morphism (exponential morphism) : O
O on {0} that sends O() to 2 O () for any open subset (where O
is the sheaf associating to any open subset the group of the holomorphic functions
on that are never zero); it is surjective but, if we take = {0}, the map exp :
O() O () is not surjective; in fact O (), but is not in the image of exp .
We define to be the sheaf

We define to be the sheaf associated to the presheaf


and to be the sheaf associated to the presheaf


We say that

.
1

F1 F F+1

is a complex if +1 = 0 and is exact if


is exact for all .

(1 )

( )

(F1 ) (F ) (F+1 )

158 | Sheaves
For any subsheaf G of a sheaf F, we define the quotient sheaf F/G to be the sheaf
associated to the presheaf F()/G(); so we get an exact sequence 0 G
F F/G 0. Obviously we can define the quotient sheaf only for sheaves with
values in categories where the quotient is defined.
Cohomology
Let be a topological space. Let F be a sheaf on and U = { } be an open covering
of . For any 0, let

C (U, F ) =

Let

(0 , . . . , ) s.t.
0 , . . . , and 0 = 0

F(0 ).

: C (U, F) C+1 (U, F)

be the following map: for = (0 ,..., )0 ,..., C (U, F) we define


( )0 ,...,+1 =

=0,...,+1

where means that we omit .


We have that 1 = 0. Define

(1) 0 ,..., ,...,+1 |

(U, F) =

+1 ,

Let U = { } and V = { } be two open coverings of . We say that V is finer than


U if and only if there exists a map : such that () for any . The
map induces a map

C (U, F ) C (V, F ),
(0 ,..., )0 ,..., ((0 ),...,( ) |

)0 ,...,

and, since this map commutes with , it induces a map


(U, F) (V, F).

One can show that this map does not depend on ; therefore we can define the ech
cohomology of F to be
(, F) := lim (U, F)
U

(see Limits, direct and inverse -); sometimes it is denoted simply by (, F).
Remark. 0 (U, F) = F() for any U open covering of .

Sheaves

159

Lerays theorem. Let F be a sheaf on a topological space .


Let U = { } be an acyclic open covering of , i.e., an open covering such that (1
, F| ) = 0 for all 1, for all and for all 1 , . . . , . Then
1

for all .

(U, F) (, F)

Proposition. Let
0FGH0

be an exact sequence of sheaves on a paracompact topological space . Then there is


a long exact sequence
(, F) (, G) (, H) +1 (, F) .

Definition. We say that a sheaf F on a topological space is acyclic if, for any 1,
we have (, F) = 0.
Abstract De Rhams theorem. Let

0 F A 0 A 1 A 0

be an exact sequence of sheaves on a paracompact topological space and suppose


that the sheaves A are acyclic. Then, for any natural number ,
where we set A1 = 0.

(, F)

(, A ) (, A+1 )
(, A1 ) (, A )

Definition. We say that a sheaf F on a topological space is fine if, for any open subset
of and any locally finite open covering { } of , there exist maps : F( )
F() such that
(1) if F( ), then () has support in , that is, ( ()) = 0 for any ;
(2) (| ) = for any F().

Theorem. Let be a paracompact topological space. A fine sheaf on is


acyclic.
Definition. We say that a sheaf F on a topological space is flasque (or flabby) if, for
all , open subsets of with , the map : F() F() is surjective.

Theorem. Let be a paracompact topological space. A flasque sheaf on is


acyclic.
There are several sheaves cohomology theories besides ech cohomology.

160 | Sheaves
One can prove that ech cohomology of constant sheaves and singular cohomology
(see Singular homology and cohomology) coincide on manifolds (see [234, Chapter 6]).
Another cohomology is the so-called derived functor cohomology. Let be a topological space; we define the cohomology functors (, ) to be the classical right derived functors of the global section functor (, ) from the category of sheaves of
Abelian groups on to the category of Abelian groups (see Derived categories and
derived functors):
For any sheaf of Abelian groups F, take an injective resolution I (we can prove that
it exists), i.e, a complex of injective sheaves I0 I1 with a morphism F I0
such that
0 F I0 I1
is an exact sequence (we say that a sheaf N is injective if (, N) is an exact functor);
define (, F) to be the cohomology in of the complex (, I ).
For any exact sequence of sheaves on a topological space
there is a long exact sequence

0 F G H 0,

(, F) (, G) (, H) +1 (, F) .

ech cohomology and derived functor cohomology generally coincide (but on pathological spaces they an be different). In particular we have the following theorems.
Theorem . (See [107, Example 4.11 and Theorem 3.5, Chapter III)]).
(a) Let F be a sheaf (of Abelian groups) on a topological space . Let U = { } be an
open covering of acyclic for the derived functor cohomology, that is, such that
(1 , F| ) = 0 for all 1, for all and for all 1 , . . . , , where

1
denotes the derived functor cohomology. Then
(U, F) (, F)

for all .
(b) Let F be a quasi-coherent sheaf on a Noetherian affine scheme (see Coherent
sheaves and Schemes). Then the derived functor cohomology (, F) is zero
for all 1.
Serres theorem. (See [223, part 46]) Let F be a coherent sheaf on an affine algebraic
variety. Then the ech cohomolgy (, F) is zero for all 1.

Thus affine open subsets of an algebraic variety give an acyclic open covering both for
ech cohomology of coherent sheaves (by Serres Theorem) and derived cohomology
of coherent sheaves (by Theorem , part (b)). Thus, by part (a) of Theorem and by

Sheaves

161

Lerays Theorem, the two cohomologies coincide for coherent sheaves on algebraic
varieties.
Furthermore, derived cohomology and ech cohomology coincide on paracompact
topological spaces (see [84, Theorem. 5.10.1, p. 228, Example 7.2.1, p. 263]):
Sheaves of O -Modules

Definition. Let (, O ) be a ringed space (see Spaces, ringed -).


We say that a sheaf of groups F on is a sheaf of O -modules if, for any open
subset of , F() is an O ()-module and, for any open subset of , the diagram
O () F()

O () F()

/ F()


/ F()

commutes (where the vertical maps are the maps induced by the restriction maps).
We say that a sheaf of O -modules is free if it is isomorphic to a direct sum of copies
of O . The number of the copies is called the rank.
We say that a sheaf of O -modules is locally free of finite rank if there is an open
covering { } of and there exist such that F| is isomorphic to (O | ) . If
is connected all the are equal.
We say that a sheaf of O -modules is invertible if it is locally free of rank 1.
Let F and G be sheaves of O -modules.
Define the tensor product F O G (or simply F G) to be the sheaf associated to the
presheaf
F() O () G()

for any open subset .


Define O (F, G) to be the group of O -module homomorphisms and define
HOM(F, G) to be the sheaf associated to the presheaf
O| (F| , G| ).

For any sheaf F of O -modules, let F = HOM(F, O) (dual sheaf).


We say that a sheaf F of O -modules is torsion-free if every stalk is torsion-free.
We say that F is reflexive if the natural map F (F ) is bijective.

Proposition. (see [146]). Let be a complex manifold. Let F be a coherent sheaf of


O -modules (see Coherent sheaves).
(1) The sheaf F is torsion-free if and only if the natural map F (F ) is injective.
(2) The sheaf F is reflexive.
(3) We have
locally free reflexive torsion-free .

162 | Siegel half-space


Let be a smooth algebraic variety and let O be the sheaf of the regular functions.
There is a bijection between the set of vector bundles on up to isomorphisms and
the set of locally free sheaves of O -modules of finite rank up to isomorphisms: the
bijection is given by the map sending a vector bundle to the sheaf O() associating
to any open subset of the group of the sections of on (see [228, Chapter 6], or
[107, Chapter 2, Exercise 5.18]).
The fibre of the vector bundle can be recovered from the stalk of the sheaf in the following way:
= (O()) / (O()) ,
where is the maximal ideal of O for any .

A sheaf of ideals on a ringed space (, O ) is a sheaf of O -modules I such that for


any open subset of , I() is an ideal of O ().
If is a subvariety of an algebraic variety , we define

Siegel half-space.

I, () = { O ()| | = 0}.

The Siegel half-space of degree is

H = { ( , )| = , () > 0},

where () > 0 means that the imaginary part of is positive definite. It is the moduli space of polarized Abelian varieties of fixed type with symplectic basis. See Tori,
complex - and Abelian varieties.

Siegels theorem.

([228], [232]).

Siegels theorem. The transcendence degree over (see Transcendence degree) of


the field of meromorphic functions on a compact complex manifold is less than or
equal to the dimension of .

See also Dimension.

Simple bundles. ([146], [207]). We say that a holomorphic vector bundle on


a compact complex manifold is simple if any holomorphic morphism of bundles
from to is a scalar multiple of the identity, that is, if 0 (, O( )) = , where
= .

Proposition. Every stable vector bundle (see Stable sheaves) on a compact Khler
manifold is simple.

Singular homology and cohomology. ([33], [91], [112], [183], [215], [234],
[247]). Throughout the item, will denote a commutative ring with unity.

Singular homology and cohomology

163

Homology
For any , let {1 , . . . , } be the canonical basis of and let 0 be the zero element
of . We define to be 0 , . . . , , i.e., the simplex spanned by 0 , . . . , . Thus 0 is
a point, 1 is a segment, 2 a triangle and so on.
Let be a topological space. A (singular) -simplex in is a continuous map :
. We define
(, ) = {

-simplex in

all are zero apart from a finite number } .

Thus (, ) is the set of the finite formal linear combinations of the -simplexes in
with coefficients in . It is called the set of the -chains in with coefficients in .
We define : 1 to be the affine map sending to for any {0, . . . , 1}
and to +1 for any {, . . . , 1}, so sending 1 onto the face of spanned by
0 , . . . , , . . . , .
For any 1, we define the border operator
: (, ) 1 (, )

to be the linear map such that, if is a singular -simplex in ,


= =0,..., (1)

and we define 0 : 0 (, ) {0} to be the zero map. We should write () instead of


, but the brackets are generally omitted.

Lemma. +1 = 0.

Thus

+1

+1 (, ) (, ) 1 (, )

is a complex; we denote it by ( (, ), ). We define the singular homology module of of degree with coefficients in , which we denote by (, ), to be the
homology of the complex ( (, ), ) in (, ), i.e.,
(, ) :=


.
+1

The elements of are called -cycles, the elements of +1 are called borders. Two -cycles are said homologous if their difference is a -border. The rank
of (, ) is denoted by (, ) or by (, ) (Betti numbers).
Remark. If is a topological space with path-connected components, then
0 (, ) .

164 | Singular homology and cohomology


A slight variant of the homology is the so-called reduced homology: for 1, we
define

: (, ) 1 (, )

to be equal to and we define 0 : 0 (, ) to be the map


.

We define the reduced homology to be

(, ) =

+1

So we have (, ) = (, ) for any 1; moreover, if is path-connected, then

0 (, ) = 0, if has path-connected components with 2, then 0 (, ) 1 .

Definition. Let : be a continuous map between two topological spaces. For


any , we can define
: (, ) (, )
to be the linear map such that

() =

for every singular -simplex (we should write , since there is one such map for
every , but we omit the subscript for simplicity).
Remark. If : and : are two continuous maps, then
( ) = .

Remark. Let : be a continuous map between two topological spaces. The


maps commute with the border maps, therefore they define operators
() : (, ) (, ).

These operators are more often denoted by .

Homotopy theorem. If : and : are homotopic continuous maps,


then
() = ()
for any . In particular homotopic topological spaces have the same homology.

Definition. Let be a topological space and let . Obviously the inclusion of


into induces an injection from (, ) into (, ). For any , we define

Singular homology and cohomology

165

(, , ) to be the homology in the place of the complex


+1 (, )
(, )
(, )

1
,
+1 (, )
1 (, )
(, )

where the maps are the maps induced by the border maps .

Couple exact sequence. Let be a topological space and let . Then there is an
exact sequence
(, ) (, ) (, , ) 1 (, ) ,

where the first map is induced by the inclusion of into and the second is induced
(,)
by the projection (, ) (,) .

Triple exact sequence. Let be a topological space and let . Then there is
an exact sequence
(, , ) (, , ) (, , ) 1 (, , ) ,

where the first map is induced by the map

(,)
(,)

(,)
(,)

given by the inclusion

(, ) (, ) and the second map is induced by the projection


(,)
.
(,)

(,)
(,)

Excision theorem. Let be a topological space and let . Let be a closed subset
included in the interior part of . Then
( , , ) (, , )

for any (the isomorphism is induced by the inclusion of ( , ) into


(, )).

MayerVietoris theorem. Let be a topological space and let 1 and 2 be two open
subsets such that = 1 2 . Let 1 : 1 , 2 : 2 , 1 : 1 2 1 ,
2 : 1 2 2 be the inclusions.
Then there is an exact sequence

(1 , ) (2 , )
(, )
(1 2 , )

1 (1 2 , ) ,

where
is the map ( (1 )(), (2 )());
is the map (, ) (1 )() + (2 )();
and, if a -cycle in is homologous to 1 + 2 with ( , ) for = 1, 2,

then ([])
:= [1 ] (we can prove that, given , we can always find 1 and 2 as
above).

166 | Singular homology and cohomology


Attaching cells. Define
= { | || 1},

1 = { | || = 1}.

Let be a Hausdorff topological space. Let be the topological space obtained by


attaching an -cell ( , 1 ) to by a continuous map
: 1 ,

i.e., let be the quotient of by the equivalence relation determined by the


identification of with () for every 1 .
If = 1, , then

(, ) = (, );
moreover,

and there is an exact sequence

1 (, ) 1 (, )/(1 ())

0 (, ) (, ) (1 ()) 0,

where 1 () : 1 (1 , ) 1 (, ).

Orientation

Let be a manifold of dimension . Here, a manifold will not be considered necessarily connected. By the Excision Theorem we can easily prove that
(, , ) =

for all (where we write (, , ) instead of (, {}, ) for simplicity).


An -orientation system of is given by
(a) a family of open subsets, { } , that covers ;
(b) for any , an element (, , ) such that, for all , the element
( ) is a generator of (, , ) and
for all , where

( ) = ( )

: (, , ) (, , )

is the map induced by the inclusion.


We say that two -orientation systems are equivalent if they induce the same elements
in (, , ) for all .
An -orientation of is the equivalence class of an -orientation system.
We say that is -orientable if there exists an -orientation on .

Singular homology and cohomology

167

Proposition. An open submanifold of an -orientable manifold is -orientable.


A manifold is -orientable if and only if all its connected components are -orientable.
If a manifold is connected and two -orientations coincide in a point, then they are
equal.
All manifolds are /2-orientable (and the /2-orientation is unique).
All connected manifolds have at most two -orientations.

Theorem. Let be a connected manifold of dimension . Then (, ) = 0 for all


> ; moreover, if is not compact, (, ) = 0.
If is compact and an integral domain, then
(, ) = {

if is -orientable,
if is not -orientable,

and, if is -orientable, the natural map (, ) (, , ) is an isomorphism for any .

Definition. If is a compact connected -oriented manifold of dimension , we define


the fundamental class of to be the element of (, ) inducing the orientation in
(, , ) for every .
Examples.

( , ) = {

if is even and 0 2,
otherwise.

0
if > ,
{
{
{
{

if = 0 or = if odd,
( , ) = {
{
/2
if
1 < and is odd,
{
{
if 1 < and is even,
{ 2
where 2 is { | 2 = 0}.

Let be the topological torus with holes (i.e., the topological space obtained by attaching to a bouquet of 2 circles, 1 , . . . , 2 , a 4-agon with the law
1
1 , +1 , 1
1 , +1 , . . . ). We have

0
if > 2,
{
{
( , ) = {
if = 0 or = 2,
{ 2
if
= 1.

{
Let be the nonorientable compact surface obtained by attaching to a bouquet of
circles 1 ,. . . , a 2-agon with the law 1 , 1 , 2 , 2 , . . . . We can see easily that 1 is
2 and 2 is the Kleins bottle. We have
{
{
{
{
( , ) = {
{
{
{
{

2
/2 1

if
if
if
if

> 2,
= 0,
= 2,
= 1.

168 | Singular homology and cohomology


Cohomology
Let be a topological space. We define

(, ) = ( (, ), );

its elements are called -cochains. Let

: (, ) +1 (, )

be the cobordism operator defined by

( )() = (+1 )

for any (, ), +1 (, ). Obviously +1 = 0, thus


1

+1

1 (, ) (, ) +1 (, ) +2 (, )

is a complex. We define the singular cohomology module of of degree with coefficients in , which we denote by (, ), to be the cohomology of such a complex
in (, ), i.e.,
(, ) =


.
1

The elements of are called -cocycles, the elements of 1 are called coborders. The rank of (, ) is denoted by (, ).
Definition. Let : be a continuous map between two topological spaces. We
can define
: (, ) (, )

to be the linear map such that

() =

for every (, ), where is the map : (, ) (, ) we have already


defined.
Remark. If : and : are two continuous maps, then
( ) = .

Remark. If : is a continuous map between two topological spaces, then


commutes with the coborder maps. Thus we can define operators
() : (, ) (, ) .

Singular homology and cohomology

169

Definition. Let be a topological space and let . We define (, , ) to be the


cohomology in the place of the complex
(

(, )
1 (, )
(, )
) (
) ( +1
) ,
1 (, )
+1 (, )
(, )

where denotes (, ) and the maps are induced by the coborder maps .

Homotopy theorem. If : and : are homotopic continuous maps,


then
() = ()
for any . In particular, homotopic topological spaces have the same cohomology.

Couple exact sequence. Let be a topological space and let . There is an exact
sequence
(, , ) (, ) (, ) +1 (, , ) ,
(,)

where the first map is induced by the inclusion ( (,) ) (, ) and the second

map is induced by the inclusion of into .

Triple exact sequence. Let be a topological space and let . There is an


exact sequence
(, , ) (, , ) (, , ) 1 (, , ) ,
(,)

(,)

where the first map is induced by the injection ( (,) ) ( (,) ) and the second
(,)

(,)

is induced by the map ( (,) ) ( (,) ) given by the restriction.

Excision theorem. Let be a topological space and let . If is a closed subset


included in the interior part of , then
(, , ) ( , , )

for any (the isomorphism is induced by the inclusion of ( , ) into


(, )).

MayerVietoris theorem. Let be a topological space and let 1 and 2 be two open
subsets such that = 1 2 . Let 1 : 1 , 2 : 2 , 1 : 1 2 1 ,
2 : 1 2 2 be the inclusions.
Then there is an exact sequence

(, ) (1 , ) (2 , ) (1 2 , )
+1 (, ) ,

170 | Singular homology and cohomology


where
is the map ( (1 )(), (2 )());
is the map (, ) (1 )() + (2 )().

Theorem . Let be a topological space and let . If is a PID, i.e., an integral


domain such that every ideal is principal, and 1 (, , ) is free (for instance if is
a field), then there is a canonical isomorphism
(, , ) (, , )

for every .
If is a PID and 1 (, , ) and (, , ) are finitely generated, then
(, ) 1 ,

for every , where is the free part of (, ) and 1 is the torsion part of
1 (, ).

For any , let : 0 , . . . , 0 , . . . , + be the affine inclusion sending


to for = 0, . . . , and, for any , let : 0 , . . . , 0 , . . . , + be the
affine inclusion sending to + for = 0, . . . , .

Definition. Let be a topological space. For every , , we define the cup product to
be the map
(, ) (, ) + (, ),
(, ) ,

where is the element of + (, ) such that, for any ( + )-simplex in ,


( )() = ( ) ( ). .

The cup product in (, ) is bilinear, associative and has the identity element (the
0-cochain sending every point in to 1).

Proposition. Let be a topological space. For any (, ), (, ), we have


= (1) .

Proposition. If : is a continuous map, then

( ) = () ()

for any (, ), (, ).

Remark. Let be a topological space. Then

( ) = + (1)

Singular homology and cohomology

171

for all (, ), (, ). Therefore, the cup product induces a product, called


again cup product,
(, ) (, ) + (, ) .

Definition. Let be a topological space. We define the cap product to be the map
(, ) + (, ) (, ),
(, ) ,

where

for any ( + )-simplex in .

:= ( )

Remark.
(1) For any (, ), (, ), + (, ),

( ) = ( )().

(2) For any (, ), + (, ),

( ) = (1) [( ) ( )].

Thus the cap product induces a product, called again cap product,
(, ) + (, ) (, ) .

Projection formula. If : is a continuous map, then

()[ ()() ] = + ()()

for any + (, ), (, ).

In the relative case, for topological space and subset of , we have maps

(, , ) (, , ) + (, , ),

(, , ) + (, , ) (, ),

(, ) + (, , ) (, , ).

The following theorem describes the homology and the cohomology of the product of
two topological spaces.
Knneth formula. Let be a PID. Let and be two topological spaces. There is an
exact sequence (see Tor, TOR for the definition of 1 )
0 += (, ) (, ) ( , )

+=1 1 ( (, ), (, )) 0

172 | Singular homology and cohomology


and, if all the homology modules are finitely generated, an exact sequence
0 += (, ) (, ) ( , )

+=+1 1 ( (, ), (, )) 0 .

Universal coefficient theorem. For any topological space (and commutative ring
with unity), there is an exact sequence
0 (, ) (, ) 1 (1 (, ), ) 0 .

Let be a manifold. For every , we define the cohomology with compact


supports (, ) in the following way:
(, ) =

lim

, compact

(, , )

(see Limits, direct and inverse -). If is compact, then (, ) = (, ).


If : is a proper continuous map (see Proper), then we have an induced map
: (, ) (, ).

If is an -oriented manifold of dimension , then, for every compact subset, we


have a map
(, , ) (, )
given by the cap product with the element of (, , ) induced by the orientation
and these maps induce a map
called Poincar duality.

: (, ) (, ),

Theorem. Let be an -oriented manifold of dimension . Then the Poincar duality


: (, ) (, )

is an isomorphism for any .

Corollary. Let be an -orientable connected manifold of dimension , then


(, ) .

If is a compact -orientable manifold of dimension , then


for any .

(, ) = (, )

Singular homology and cohomology

173

By the de Rhams theorem (see De Rhams theorem)

(, ) =
(, )

for any manifold and the isomorphism


(, ) ( (, )) (given by the
De Rhams theorem and Theorem ) is induced by the integration.

(, )
Moreover, the cup product in (, ) corresponds to the wedge product in
and to the intersection of cycles in (, ) through the Poincar duality, i.e.,

(1 (1 ) 1 (2 )) = 1 2

for any 1 , 2 (, ) (see Intersection of cycles for the definition of 1 2 ).

Let be an -oriented manifold. Let be a closed subset in . We define the Alexan


der cohomology of , which we denote by (, ), in the following way:

(, ) =

lim

open subset

(, ).

Theorem. If is an -oriented compact manifold of dimension , then there is an


isomorphism, called Alexander duality,

(, ) (, , ) .

Corollary.
(i) Let be a compact submanifold in . Then, for any ,

(, ) 1 ( , ).

(ii) If is a compact submanifold in with dimension 1 and connected components, then has + 1 connected components.
(iii) If is a nonorientable compact manifold of dimension 1, then it cannot be
embedded in .

Theorem. Let be a compact manifold with boundary. Denote the boundary of


by . Suppose the interior part of is -oriented. Then, for any , there are isomorphisms (called Lefschetz dualities):
(, , ) (, ),

(, ) (, , ) .

Lefschetz fixed points theorem. Let be a field and let be a compact -oriented
manifold. Let : be a continuous map. We have
= (1) (),

174 | Singularities
where is the diagonal in , is the graph of , is the intersection number
(see Intersection of cycles) and is the trace.

Singularities.
Smooth.

See Regular rings, smooth points, singular points.

See Regular rings, smooth points, singular points.

Snake lemma.

([12], [79], [116], [164]). Let


1


/ 1

1
1

/ 2


/ 2

2
2

/ 3


/ 3

/0

be a commutative diagram of Abelian groups (or of -modules for some ring ). If the
rows are exact, then there is an exact sequence
(1 ) (2 ) (3 ) (1 ) (2 ) (3 ).

Moreover, if 1 is injective, then the map (1 ) (2 ) is injective; if 2 is surjective, then the map (2 ) (3 ) is surjective.

Spaces, analytic -.

([103], [107]). Let be the polydisc


{ | | | < 1 {1, . . . , }}.

Definition.
An analytic space is a locally ringed space (, O ) (see Spaces, ringed -) such
that, for any , there exist an open neighborhood of , natural numbers
, and holomorphic functions 1 , . . . , on such that
(, O | )

is isomorphic, as locally ringed space, to

( , O/(1 , . . . , )),

where O is the sheaf of the holomorphic functions on and (1 , . . . , ) is the


ideal sheaf generated by 1 , . . . , (see Sheaves).
Let (, O ) and (, O ) be analytic spaces. A map from to is said to be a morphism of analytic spaces if it is a morphism of locally ringed spaces.

Spaces, ringed -.

([103], [107]). A ringed space (, O ) is a topological space


with a sheaf of rings O (see Sheaves). The sheaf O is called structure sheaf.
We say that a ringed space (, O ) is reduced if, for any open subset of , O ()
has no nilpotent elements. We say that a ringed space is a locally ringed space if the
stalk of the structure sheaf in every point is a local ring.

Spectral sequences

175

A morphism of ringed spaces from a ringed space (, O ) to a ringed space (, O )


is given by a continuous map : and a morphism of sheaves O O
(see Direct and inverse image sheaves for the definition of O ). A morphism of
locally ringed spaces is a morphism of ringed spaces such that the maps induced on
the stalks are local homomorphisms of local rings (i.e., homomorphisms such that the
inverse image of the maximal ideal of the second ring is the maximal ideal of the first
ring).

Spectral sequences. ([32], [41], [62], [84], [93], [116], [180]). A spectral sequence
is a sequence { , } of bigraded groups
= , ,

together with differentials (see Figure 17)

such that 2 = 0 and

i.e.,

for any , , .

: , +,+1
( ) = +1 ,

( : , +,+1 )
,+1

( :

= +1

,
+,+1

Fig. 17. Spectral sequence.

If = + for all , we denote by ; in this case we say that the spectral


sequence converges to and we write .

Let be a group. Given a filtration in subgroups on

= 0 1 +1 = 0,

176 | Spin groups


we set, for any ,

:= /+1 .

A filtered complex

= 0 1 +1 = 0

is defined to be a finite sequence of complexes such that is a subcomplex


of +1 for any .

Theorem. Given a filtered complex = 0 1 +1 = 0,


there is a spectral sequence , called associated spectral sequence, such that
,

0 = + ,
,

1 = + ( ),

,
= (+ ( )).

Precisely, is the spectral sequence defined in the following way: for any , , ,
, :=

{ + | + ++1 }
.
(+1 + 1 ) + +1 +

Given a bigraded complex (, , , ), i.e., a bigraded group


= , ,

with differentials

: , +1, ,

: , ,+1

such that 2 = 0, 2 = 0 and + = 0, we can define the associated single


complex in the following way (see Figure 18):
= += ,

with the differential

It has two obvious filtrations:

= + : +1 .

= ,+= , ,
= ,+= , .
By the theorem above, there are two spectral sequences and associated to
endowed respectively with the two filtrations and . We have:
,

2 ( (, )),

( (, )).

See also Hypercohomology of a complex of sheaves, Leray spectral sequence.

Spin groups. ([47], [76], [167]). Let () be the group of the orthogonal real
matrices with determinant 1. Its fundamental group is
1 (()) = {

/2

if
if

= 2,
3.

Spin groups

177

2
1

Fig. 18. Bigraded complex.

Thus there is a double covering space of (), which is universal if 3 (see Covering projections); it is denoted by ().
To describe it and to define a group structure on it, we must define Clifford algebras.
Definition. Let be a vector space over or over and endow with the bilinear product defined by
(1 ) (1 ) = 1 1 .

We will omit the symbol in the sequel. With such a product, is a ring. Let
be a non degenerate bilinear symmetric form on . Let
Cliff (, ) = 0 /,

where is the two-sides ideal generated by the elements of the form (, ).

For every = [1 ] Cliff (, ) ( ), we denote by the element


(1) [ 1 ];

we extend the definition of to every element Cliff (, ) by linearity.

Definition. Let

(, ) = { Cliff (, )even | = 1 and },

where Cliff(, )even = 0, even /. We endow (, ) with the product induced


by the product .
Now take = and the opposite of the standard positive definite bilinear form on
. In this case (, ) is denoted () and we can prove that:
if we define, for any (, ), a map () : to be the map
,

we have that () is an element of (, );

178 | Splitting type of a vector bundle

the map
: (, ) (, ),
()

is a two-sheeted covering projection and a group homomorphism.


An analogous statement holds for complex vector space and non degenerate bilinear symmetric form on .

Splitting type of a vector bundle.

vector bundle on .

See Jumping lines and splitting type of a

Stable sheaves. ([83], [146], [182], [199], [207], [239]). Let (, ) be a compact Kh-

ler manifold of dimension and let be the Khler form. Let F be a coherent sheaf on
(see Hermitian and Khlerian metrics, Sheaves and Coherent sheaves). The
determinant bundle of F, denoted by F, is defined in the following way: let
0 E E 1 E 0 F | 0

be a resolution of F| with E locally free coherent sheaves and small open subset
of ; let be the bundle corresponding to E . We define

F| = =0,..., ( )(1) .

If F is also torsion-free and its rank is , then we can give the following equivalent
definition: F := ( F) .
We define the first Chern class of F by
1 (F) = 1 ( F)

(see Chern classes). The -degree of F is


The degree/rank ratio of F is

deg(F) = 1 (F) 1 .

(F) = deg(F)/(F).

According to Takemotos definition, we say that F is -semistable if, for every coherent
subsheaf F with 0 < (F ), we have
(F ) (F).

Star operator

179

If we have also that, for every coherent subsheaf F with 0 < (F ) < (F),
(F ) < (F),

we say that F is -stable.


We say that a holomorphic vector bundle on is -(semi)stable if and only if the
sheaf O() of the sections of is -(semi)stable.

There are other definitions of stability. Giesekers one is the following. Let be a
smooth projective algebraic variety over and let be an ample line bundle on .
Let F be a torsion-free coherent sheaf on . Let
(F()) =

(1) (, F O( ))
.
(F)

We say that F is Gieseker -stable (respectively Gieseker -semistable) if, for every
coherent subsheaf F of F with 0 < (F ) < (F), there exist 0 such that
(F ()) < (F()),

(respectively (F ()) (F()))

for any 0 .
We can prove that Takemoto -stability implies Gieseker -stability, which implies
Gieseker -semistability, which implies Takemoto -semistability, where 1 ().
Any (Takemoto or Gieseker) stable bundle is simple; see Simple bundles.

By using geometric invariant theory (see Geometric invariant theory (G.I.T.)),


Gieseker and Maruyama constructed moduli spaces of Gieseker stable sheaves.

Star operator. ([44], [90], [93], [135], [250]). Let (, ) be a riemannian manifold
of dimension and let () denote the set of the -forms on . We define the
star operator
: () ()

in the following way: let 1 , . . . , be an orthonormal frame on an open subset, i.e.,


for every point of the open subset, let it give a basis of () (the dual of the tangent
space at ) orthonormal for ; locally is defined by
(()1 ) = ()1 ,

where 1 1 = 1 ; we extend this definition


to any -form by linearity.
We have that
= (1)() .

If (, ) is a Hermitian complex manifold of dimension (see Hermitian and Khlerian metrics), the star operator is defined on complex forms by extending the star
operator defined above, by -linearity in some books, by -linearity in other books.

180 | Stein factorization

Stein factorization.

([107], [241]).

Theorem. (algebraic version; see [107]) Let : be a projective morphism of


Noetherian schemes (see Schemes). Then there exists a scheme , a finite morphism
: and a projective morphism with connected fibres : such that
= .

Theorem. (analytic version); (see [241]). Let : be a proper surjective morphism of reduced complex analytic spaces (see Spaces, analytic -). Then there exists
a reduced complex analytic space , a surjective morphism : such that the
fibres consist of a finite number of points and a surjective morphism with connected
fibres : such that
= .

Subcanonical. We say that a smooth subvariety of a smooth algebraic variety


is subcanonical if there exists a line bundle on such that | = , where is
the canonical bundle of (see Canonical bundle, canonical sheaf).

Surfaces, algebraic -. ([18], [22], [25], [65], [93], [107], [253]). In the sequel the
word surface will denote a smooth projective algebraic variety of dimension 2 over .
We will denote the projective space by .

We start by stating some basic theorems and notation.


Let be a surface; then 4 (, ) and 4 (, ) are isomorphic to , since is oriented
(as it is a complex manifold), compact and of real dimension 4 (see Singular homology and cohomology). We denote by [] the fundamental class of , that is, the generator of 4 (, ) giving the orientation of . If and are two holomorphic line bundles
on , we denote by the number obtained by evaluating 1 () 1 ( ) 4 (, )
in [] (where is the cup product; see again Singular homology and cohomology).
For any divisors and on , let their intersection number (see Intersection
of cycles).
Observe that, for any holomorphic line bundle on , we have that (O()) = 0 for
3; to prove this, we can apply the Abstract de Rhams theorem (see Sheaves) to
the exact sequence

0 O() () 0,1 () 0,2 () 0,

where () is the sheaf of the sections of , , () is the sheaf of the (, )forms with values in and the map O() () is given by inclusion.
So, if we denote by (O()) the Euler characteristic of O(), i.e., the number
=0,..., (1) (O()), we have that, in this case (i.e., in the case of surfaces),
(O()) = 0 (O()) 1 (O()) + 2 (O()),

Surfaces, algebraic -

181

and so (O ) = 0 (O ) 1 (O ) + 2 (O ). Furthermore, we recall that () denotes


the Euler-Poincar characteristic of , i.e., =0,..., (1) (), where () is the Betti
number (, ); in this case,
() = 0 () 1 () + 2 () 3 () + 4 () = 20 () 21 () + 2 ()

by Poincar duality (see Singular homology and cohomology).

Index theorem. Let be a surface. The intersection form is negative definite on a subspace of codimension 1 in 1,1 (, ), precisely is negative definite on the subspace of
the primitive forms (see Lefschetz decomposition and Hard Lefschetz theorem for
the definition of primitive).
RiemannRoch theorem for surfaces. Let be a holomorphic line bundle on a surface . We have:

(O()) = (O ) +
,
2

where is the canonical bundle of (see Canonical bundle, canonical


sheaf).

Noethers theorem. Let be a surface. We denote by () the -th Chern class of the
holomorphic tangent bundle 1,0 and we denote by 1 ()2 the cup product 1 ()1 ().
We have
1
( + ())
(O ) =
12
or, in other words,
1
( ()2 + 2 ()) ,
(O ) =
12 1
where by 1 ()2 and 2 (), we mean them evaluated in [].

The two formulas are equivalent because 1 ( ) = 1 (1,0 ) and, by the GaussBonnet
Hopf theorem (see GaussBonnetHopf theorem) 2 (1,0 ) = ().

CastelnuovoEnriques criterion. Let be a surface and a smooth rational curve in


with = 1. Then there exists a surface and a map : such that is the
blow-up of in a point and 1 () = .

Structure of birational maps on surfaces. Let 1 and 2 be two surfaces and : 1


2 be a birational map. Then there exists another surface and two morphisms, 1
and 2 , both given by a sequence of blowing-up maps, such that the following diagram
commutes:

 ===
1 
==2

==
.

=



/ 2
1

182 | Surfaces, algebraic In other words, a birational map between surfaces is given by a sequence of blowingups followed by a sequence of blowing-downs.
Definition. We say that a surface is minimal if there does not exist another surface
and a blowing-up . A minimal model for a surface is a minimal surface birational to .
We recall that if is a compact complex manifold of dimension , we define
() = 1,0 () (irregularity);

() = 0 (O(
)) for any (plurigenera);
,0
() = () = 1 () (geometric genus).

One can prove that they are birational invariants. With the notation above, we have
(O ) = 0 (O ) 1 (O ) + 2 (O )

= 0,0 () 0,1 () + 0,2 () = 1 () + ()

by Dolbeaults theorem and the Hodge theorem (see Hodge theory).


The most important tool to classify surfaces is Kodaira dimension (see Kodaira
dimension (or Kodaira number)). For surfaces, it can be obviously only , 0, 1, 2.
First, we will deal with rational and ruled surfaces.

Definition. We say that a surface is rational if it is birational to 2 .

Definition. We say that a surface is ruled over a compact Riemann surface if it is


birational to 1 .

Remark. Let be a surface. Then is ruled on 1 is rational (in fact it is birational


to 1 1 if and only if it is birational to 2 ).

Definition. We say that a surface is geometrically ruled over a compact Riemann


surface if there exists a smooth morphism such that the fibres are isomorphic
to 1 .

(Please note that in some works the term ruled means geometrically ruled.)

NoetherEnriques theorem. A geometrically ruled surface is equal to the projectivized


() of a vector bundle of rank 2.

Proposition 1. Let be a ruled surface on a compact Riemann surface . Let () be


the genus of . Then
() = (),

() = 0,

Moreover, if is geometrically ruled, then

2 = 8(() + 1),

() = 0 2.
2 () = 2 .

Surfaces, algebraic -

183

Definition. Let . The -th Hirzebruch surface is


= (O1 () O1 ),

where O1 is the trivial bundle on 1 and O1 () is the -th power of the hyperplane
bundle on 1 ; see Hyperplane bundles, twisting sheaves (we are making a slight
abuse of the notation: O1 () generally denotes the sheaf of the holomorphic sections
of the -th power of the hyperplane bundle, but sometimes, as here, it denotes just the
-th power of the hyperplane bundle).

Remark. The Hirzebruch surfaces are exactly the surfaces that are geometrically ruled
on 1 . In fact, by the GrothendieckSegre theorem (see GrothendieckSegre theorem) and by the NoetherEnriques theorem, we have that a surface that is geometrically ruled on 1 is equal to ( 1 2 ) = (( 1 1
2 ) O1 ) for some 1 and 2
1
1
holomorphic line bundles on and (( 1 2 ) O1 ) = (O1 () O1 ) for some
. Obviously we can suppose .

We can prove that for all = 0 the surface is the unique 1 -bundle on 1 with an
irreducible curve with = and that the blow-up of 2 in a point is 1 . We can
also prove that 1 can be obtained from by a blowing-up followed by a blowingdown.
Let be a fibre of (O1 () OP1 ) and be the image of the section given by the zero
section of O1 () and by the section 1 of OP1 . Let , be the image of by the map
(+) associated to the linear system | + | (see Bundles fibre -). We can prove
that it is a surface in +2+1 of degree +2, which is the minimal achievable degree for
a surface in +2+1 (see Minimal degree). The surfaces , are the rational normal
scrolls of dimension 2 (see Scrolls, rational normal -).

Proposition 2. Every nondegenerate surface of minimal degree in , i.e., every nondegenerate surface of degree 1 in , is either a rational normal scroll or the
Veronese surface in 5 , i.e., the image of 2 embedded in 5 by O2 (2) (see Veronese
embedding).

Another example of rational surfaces are Del Pezzo surfaces. Let 6. Let 1 , . . . ,
be distinct points in 2 in general position (i.e., no 3 of them are collinear, and no
6 of them lie on a conic). Let : 2 be the blowing-up of 2 in 1 , . . . , . Then
defines an embedding : 9 , whose image has degree 9 and is called a
Del Pezzo surface.
If = 6, () is a smooth cubic in 3 ; if = 5, () is a complete intersection of two
quadrics in 4 .
One can show that () contains only a finite number of lines, precisely the images
under of
(a) the exceptional curves;
(b) the strict transforms of the lines for = ;

184 | Surfaces, algebraic (c) the strict transforms of the conics through 5 of the .
In the case = 6 we have exactly 27 lines.

Theorem.
(i) The minimal ruled surfaces on 1 are isomorphic either to the Hirzebruch surfaces
or to 2 .
(ii) The minimal ruled surfaces on a Riemann surface of genus 1 are the geometrically ruled ones and the minimal model is not unique.
Theorem. The nonruled surfaces have a unique minimal model (up to isomorphisms).

Remark. If is a rational surface, then, for every 1, we have () = () =


() = 0.
The following theorem tells us that also the converse is true.

CastelnuovoEnriques theorem. Let be a surface such that () = 2 () = 0. Then


is rational.
Theorem. Any unirational surface (see Unirational, Lroth problem) is rational.

Theorem (CastelnuovoDe Franchis).


Let be a minimal surface with () < 0; then is irrational ruled.
Let be a minimal surface with 1 ()2 < 0; then is irrational ruled.
Let be a minimal surface with ((O )) < 0; then is irrational ruled.

(Observe that the third statement follows trivially from the first two and Noethers theorem.)
Enriques theorem. A surface is ruled if and only if 12 () = 0.

As we have already said, the most important tool to classify surfaces is Kodaira dimension.
Enriques theorem and Proposition 1 tell us that () = if and only if is ruled.
Definition. We say that a surface is hyperelliptic or bielliptic if it is equal to
( )/,

where and are elliptic curves and is a finite group of translations of acting on
in such a way that / = 1 .

Definition. Let be a smooth curve. An elliptic surface with base is a surface such
that there exists a surjective morphism such that the generic fibre is an elliptic
(irreducible) curve.

Surfaces, algebraic -

185

Classification theorem (EnriquesKodaira).


A minimal surface with () = 0 is one of the following:
(a) a surface with = 0 and = 1 (this implies = O); such surfaces are called
K3 surfaces;
(b) a surface with = 0 and = 0 (this implies 2 = O); such surfaces are
called Enriques surfaces;
(c) a hyperelliptic surface if = 1;
(d) an Abelian variety if = 2 (see Tori, complex - and Abelian varieties).
A surface with () = 1 is elliptic.

A particular case of K3 surfaces are the Kummer surfaces. A Kummer surface is a


surface obtained in the following way: let be an Abelian surface and let be the
surface obtained from by blowing up the 16 points of order 2; define

= /,
,

where is the map induced on by the map on .

Proposition. The quotient of a K3 surface by a fixed-point free involution is an Enriques


surface.
Conversely, let be an Enriques surface. As we have already said, 2 is the trivial
bundle. Let = { | ( ) = 1} where is the isomorphism 2 . The
surface is a (nonramified) double covering space of , where the covering map is
induced by the projection of to , and it is a K3 surface.
Definition. We say that a surface is of general type if () = 2.

BogomolovMiyaokaYau inequality. For a surface of general type, it holds that


where () = (1,0 ).

12 () 3 2 (),

Noether inequality. For a minimal surface of general type, , it holds that


where () = (1,0 ).

()

1 2
() + 2,
2 1

Bombieri and Mumford extended the classification of surfaces to arbitrary algebraically closed fields (see [25]). The classification in the case where the characteristic
is different from 2, 3 is analogous to the one over the complex numbers.

186 | Surfaces, algebraic -

Symmetric polynomials ([76], [164], [178], [181], [236], [238]). Let . Let
[1 , . . . , ]; we say that is symmetric if it is invariant for the action of the
symmetric group (in other words, if interchanging any of the variables does not
modify the polynomial).
We will denote by [1 , . . . , ] the set of the symmetric polynomials in 1 , . . . ,
with coefficients in .
Let (1 , . . . , ) be the sum of the squarefree monomials of degree in 1 , . . . ,
(squarefree means not divisible by the square of a variable). The polynomials are
symmetric and they are called elementary symmetric polynomials.
For example,
0 (1 , . . . , ) = 1;
1 (1 , . . . , ) = 1 + + ;
2 (1 , . . . , ) = 1< .
Observe that (1 , . . . , ) = 0 for any + 1.

The elementary symmetric polynomials can be defined also by the following formula:
=1,..., (1 + ) = (1 , . . . , ) .

Gauss theorem. If [1 , . . . , ] , then there exists a polynomial [1 , . . . , ]


such that = (1 , . . . , ).

Let (1 , . . . , ) be the sum of the monomials of degree in 1 , . . . , . The are


called complete symmetric polynomials.
For example:
0 (1 , . . . , ) = 1;
1 (1 , . . . , ) = 1 + + ;
2 (1 , . . . , ) = 1 .
The complete symmetric polynomials can be defined also by the following formula:
=1,...,

1
= ( , . . . , ) .
1 1

Remark. The following relations hold:

(1) (1 , . . . , ) = 0

(1) = 0

{1, . . . , },
1.

Surfaces, algebraic -

187

The first follows from the second definition of taking = 1/ . The second follows
from the second definitions of and :
1 = =1,..., (1 ) =1,...,

1
1

= ( (1) (1 , . . . , ) )( (1 , . . . , ) ).

From the second relation we have [1 , . . . , ] = [1 , . . . , ], which, by Gauss


theorem, is [1 , . . . , ] . Thus both the elementary symmetric polynomials and
the complete symmetric polynomials generate the algebra of the symmetric polynomials.
Now we will define another class of symmetric polynomials such that they generate
[1 , . . . , ] as -module: the Schur polynomials.

Let = ( 1 , . . . , ) with and 1 (we call it a partition of 1 + + );


we can associate to a diagram, called a Young diagram, with boxes in the -th row
for any {1, . . . , } and the rows lined up on the left; see Figure 19.
Fig. 19. Young diagram of (4, 3, 1).

Let be the matrix :

1
.
( .
.
1

1
.
.
.

12
.
.
.
2

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
. ).
.
.

Number the columns of beginning from 0. For distinct 1 , . . . , , define


1 ,..., (1 , . . . , ) to be the determinant of the matrix obtained by taking the columns
1 , . . . , of .
For = ( 1 , . . . , ) with and 1 , define
(1 , . . . , ) =

+ (1 , . . . , )
,
(1 , . . . , )

where = ( 1, 2, . . . , 0). The are symmetric polynomials and they are called
Schur polynomials.
For example: let = 3 and consider = (1, 1, 0), which we write (1, 1) (in general the
zeroes at the end of a partition are omitted); we have
(1,1) =

32 22 (3 2 ) 12 32 (3 1 ) + 22 12 (2 1 )
= 1 2 + 1 3 + 2 3 .
> ( )

188 | Syzygies
Theorem. The Schur polynomials are a basis of [1 , . . . , ] as -module.

The following formulas express the Schur polynomials in terms of elementary symmetric polynomials and in terms of complete symmetric polynomials.

JacobiTrudiGiambelli formulas. For any = ( 1 , . . . , ) with and 1


, we have
1
2 1
= det (( + ),{1,...,} ) = det (
.
.
1

= det (( + ),{1,...,} ) = det ( 2 1


.
.

1 +1
2
.
.

1 +1
2
.
.

.
.
.
.

.
.
.
.

.
.
),
.
.

.
.
),
.
.

where = (1 , . . . , ) is the conjugate of , i.e., is the number of the boxes of the -th
column of the Young diagram of (in other words, the Young diagram of is obtained
from the Young diagram of by interchanging rows and columns).
LittlewoodRichardson rule. For any = ( 1 , . . . , ) with and 1
and for any = (1 , . . . , ) with and 1 , we have
= ,, ,

where ,, is the number of the ways the Young diagram of can be expanded to the
Young diagram of by a strict -expansion, where
a = (1 , . . . , )-expansion of the Young diagram of is a Young diagram obtained from the Young diagram of by adding 1 boxes not two in the same column, then 2 boxes not two in the same column, and so on;
a -expansion is called strict if the following condition hold: put a 1 in each of the
1 boxes, a 2 in each of the 2 boxes, and so on; form a list reading the numbers
in the boxes, reading from right to left and beginning from the top row; we must
have that for every with 1 1 + + , and for every with 1 1,
in the first entries of the list, the number of the s is greater than or equal to the
number of the ( + 1)s.
See Schur functors.

Syzygies. Let be a field. A syzygy among some polynomials, 1 , . . . ,


[1 , . . . , ], is a -uple of polynomials (1 , . . . , ) with 1 , . . . , [1 , . . . , ]
such that
= 0.
=1,...,

Syzygies

189

More generally, a syzygy among some -uples of polynomials in [1 , . . . , ],


11
1
1 = ( ... ) , . . . . . . . . . , = ( ... ) ,
1

is a -uple of polynomials (1 , . . . , ) with 1 , . . . , [1 , . . . , ] such that


= 0.

=1,...,

Given a projective algebraic variety, one can study the syzygies among generators of
the ideal of the variety and then the syzygies among these syzygies, and so on. In
particular, one can study the degree of such syzygies. A definition which is often used
is the following one; it is due to Green and Lazarsfeld (see [88], [89], [170]):
Let be a smooth complex projective algebraic variety of dimension and let
be a holomorphic line bundle on defining an embedding : , where
= (0 (, ) ) (see Bundles, fibre -). Set = 0 (O()), the homogeneous coordinate ring of the projective space , and consider the graded -module
= 0 (, O( )). Let
1 0 0

be a minimal graded free resolution of (see Minimal free resolutions). For any
, we say that the line bundle satisfies Property if the two following conditions hold:
0 = ,

= ( 1)

for

1 ,

where the second condition means that is the direct sum of some copies of ( 1).
Observe that the kernel of the map is the ideal of () (take the cohomology of
the exact sequence 0 I () () O () O () () 0, where I () is the ideal
sheaf ()).
Thus, satisfies Property 0 if and only if is normally generated, satisfies Property 1 if and only if it satisfies Property 0 and the ideal of () is generated by
quadrics, and satisfies Property 2 if and only if it satisfies Property 1 and the
syzygies among these quadrics are linear, and so on.
See also Groebner bases, Hilbert syzygy theorem.

190 | Tautological (or universal) bundle

T
Tautological (or universal) bundle.

([93], [188]).
Let be a vector space of dimension and let < . The tautological (or universal bundle) on the Grassmannian of -subspaces in , (, ) (see Grassmannians), is the bundle whose fibre on (, ) is the -subspace . In
particular the tautological bundle on the projective space () is the line bundle
whose fibre on is the line ; it is the dual of the hyperplane bundle (see Hyperplane bundles, twisting sheaves).

Let be a vector bundle on a manifold (or an algebraic variety) and let :


() be the projectivized bundle, i.e., the bundle whose fibre on is
the projectivized of . The tautological (or universal) bundle on the projectivized bundle () is the following bundle: the subbundle of whose fibre
on a point of () is the line represented by .
Its dual is the line bundle O() (1), i.e., the line bundle whose restriction on ( )
is O(1) for all .

More generally: let be a vector bundle on a manifold (or an algebraic variety) ,


and let : (, ) be the bundle whose fibre on is (, ); the tautological (or universal) bundle on (, ) is the following bundle on (, ):
the subbundle of whose fibre on a point of (, ) is the -subspace of
() = ( ) given by :
()


(, )


/

Obviously taking equal to a point, we get the notion of universal bundle on


the Grassmannian and taking = 1 we get the notion of universal bundle on the
projectivized of a bundle.

Tor, TOR. ([41], [62], [93], [116]). Let be a commutative ring with unity. Let be
an -module. Let
1 0 0

be a projective resolution of (see Injective and projective resolutions); we denote


by the complex
1 . 0 0.

Let be another -module. We can consider the complex :

1 0 0.

Tor, TOR

We define
One can prove that if

191

(, ) := ( ).

1 0 0

is a projective resolution of and is the complex

1 0 0,

then (, ) is equal to ( ). Moreover, one can demonstrate that the module (, ) does not depend on the choice of the resolutions. Sometimes we
will omit the superscript in (, ).
Proposition.
(1) (, ) (, ).
(2) For any short exact sequence of -modules
we have an exact sequence:

0 0,

2 (, ) 2 (, ) 2 (, )
1 (, ) 1 (, ) 1 (, )

0.

(3) Let be an -module. It is flat (see Flat (module, morphism)) if and only if
(, ) = 0 for every -module and for every 1 and this is true if and
only if 1 (, ) = 0 for every -module .
(4) Free implies flat and also projective implies flat (see Injective and projective modules).
(5) If = , or more generally is a principal ideal domain, then is flat if and only
if it is torsion-free.
More synthetically, we can define (, ) to be the -th classical left derived functor
of the right exact functor (see Derived categories and derived functors).

Let (, O ) be a ringed space (see Spaces, ringed -). Let F be a sheaf of O -modules
on (see Sheaves). We define TORO (F, ) to be the left derived functor of F O .

Proposition. If F is locally free, then TOR (F, G) = 0 for all > 0 and for all G sheaf
of O -modules.
O

Observe that the Proposition above can deduced from properties (a) and (d) of EXT
and the fact that, if F is locally free, then HOM(F , ) = F (see , EXT ).

192 | Torellis theorem

Torellis theorem. See Jacobians of compact Riemann surfaces for Torellis theorem on Riemann surfaces. More generally we say that a theorem is of the Torelli
kind if it states the injectivity of a map associating a certain kind of varieties to another kind of varieties.
Tori, complex - and Abelian varieties. ([93], [139], [163], [165], [166], [193]).
We follow mainly the exposition in [165].
A complex torus of dimension is defined to be the quotient of a complex vector
space of dimension by a lattice of maximal rank in (we say that is a lattice
of maximal rank in if there exists a basis of over , { 1 , . . . , 2 }, such that =
1 + + 2 ):
= /.
Example. Let = , = 1 + 2 with 1 , 2 indipendent over . The quotient
= / is a complex torus of dimension 1; see Figure 20.

Fig. 20. A complex torus.

A complex Abelian variety is defined to be a complex torus embeddable into a projective space.
Remark. Let 1 = 1 / 1 and 2 = 2 / 2 be complex tori. Let : 1 2 be a
holomorphic map such that (0) = 0. Then is a group homomorphism and there is
a unique -linear map : 1 2 sending 1 into 2 and inducing .

Definition. Let 1 = 1 / 1 and 2 = 2 / 2 be two complex tori. An isogeny from


1 to 2 is a surjective holomorphic homomorphism : 1 2 with a finite kernel.
Remark. The isogeny is an equivalence relation.

Tori, complex - and Abelian varieties

193

One can easily prove that any holomorphic line bundle on a complex vector space is
trivial; thus, if = / is a complex torus, : is the projection and is a
holomorphic line bundle on , then is trivial. Therefore for any and for any
there is an isomorphism from ( ) onto ( )+ given by an element of ,
we call , . The , , , are called factors of automorphy of and satisfy
the following relation:
,+ , = +,

(6)

, , .

On the other hand, given a set {, }, such that , for any and , they
are holomorphic functions in and they satisfy (6), one can define a holomorphic line
bundle on as follows:
= ( )/ ,

where is the relation (, ) ( + , , ).

Remark . There is a bijection between the set T of the Hermitian forms on a complex
vector space and the set E of the real valued alternating forms on such that
(, ) = (, ) for any , .
The bijection can be described as follows. Send T to := (), where
denotes the imaginary part, and, conversely, send E to the form defined by
(, ) = (, ) + (, ) for any , .
Theorem (AppellHumbert). Let = / be a complex torus. There is a canonical
isomorphism of exact sequences
1

/ (, 1 )

/ 0 ()

/ ()


/ ()

/ ()


/ ()

/0
/0

where
1 is the group { | || = 1};
() = { : Hermitian s.t. (, ) };

(), : 1 ,
};
() = {(, )
( + ) = ()()(,) ,
the map (, 1 ) P() is the map (0, );
the map () () is the map (, ) ;
the map () () is the following map: let (, ) (); define

, = ()(,)+ 2 (,)

for , ; they satisfy (6), and thus they define a holomorphic line bundle
on , which we call (, ); we associate to (, ) the line bundle (, ).

194 | Tori, complex - and Abelian varieties


See Chern classes, Equivalence, algebraic, rational, linear -, Chow, NeronSeveri
and Picard groups for the definition of 1 , 0 (), (), () and of the exact sequence in the second line of the diagram.

Definition. Let = / be a complex torus. We define the dual complex torus


=

(, )
,
{ (, )| () }

where (, ) is the vector space of the additive functions : such that


() = () for any , .
Proposition. The map

(, ) (, 1 ),
2()

induces an isomorphism from to 0 ().

Definition. Let = / a complex torus. A Poincar bundle for is a holomorphic


line bundle P on such that
(1) P|{} for any (where we are identifying with 0 ());
(2) P|{0} is trivial.
Proposition. There exists a unique, up to isomorphisms, Poincar bundle on
.

Definition. Let be a holomorphic line bundle on . Let : be the map


() 1 ,

where we identify with 0 (), and, for any , the map : is defined
to be the map + . We define () = .

We can prove that is induced by the map (, ), (, ), where


is the element of () representing 1 (); thus depends only on and then it
can be denoted also by . Thus
() = { | (, ) }/.

Let = / be a complex torus. Let (). One can prove that there is a symplectic basis B of , that is, there exist for = 1, . . . , with 1 |2 | | and a basis
B of such that is expressed in the basis B by the matrix
(

),
0

Tori, complex - and Abelian varieties

195

where is the diagonal matrix with diagonal (1 , . . . , ). If is a holomorphic line


bundle on and is the element of () representing 1 (), the -uple (1 , . . . , ) is
called the type of . We say that is nondegenerate if and only if is nondegenerate.

Proposition. If is nondegenerate, then

() = =1,..., (/ )2 ,

where (1 , . . . , ) is the type of .

Let : / = be the canonical projection. Let be a holomorphic line bundle


on . Since is trivial, having chosen a system of factors of automorphy , , one
can identify 0 (O()) with the set of holomorphic functions on such that, for any
and ,
( + ) = , ().
Such functions are called theta functions.

Theorem. Let be a complex torus of dimension and a holomorphic line bundle


on such that 1 () has positive eigenvalues and negative eigenvalues. Then

{ ( ) 1 +
(O()) = {
{ 0

if and |()0 is trivial,

otherwise,

where (1 , . . . , + , 0, . . . , 0) is the type of and ()0 is the connected component of


() containing 0. In particular, if 1 () is positive definite, then
0 (O()) = 1 ,

where (1 , . . . , ) is the type of .

Corollary (RiemannRoch theorem). If is a holomorphic line bundle on a complex


torus of dimension , then
1
(O()) = ,
!

where (O()) is the EulerPoincar characteristic of O(), i.e., the number


=0,..., (1) (, O()), and is the cup product 1 () 1 () ( times) evaluated in the fundamental class of , that is, in the element of 2 (, ) giving the
orientation of (see Singular homology and cohomology).
By using Kodaira embedding theorem (see Kodaira Embedding theorem) or directly,
one can prove the following proposition.

Proposition. A complex torus is embeddable in a projective space, i.e., it is an Abelian


variety, if and only if there is a (1, 1) closed positive integer translation-invariant form.
This is equivalent to requiring that there be a positive holomorphic line bundle (the
first Chern class 1 () gives the required form).

196 | Tori, complex - and Abelian varieties


The form above is called polarization. The polarization is said to be principal if is
of type (1, . . . , 1). We say that two polarized Abelian varieties are isomorphic if there
is a biholomorphism such that the pull-back of the polarization of the second Abelian
variety is the polarization of the first.
More generally, following [166], we can give the following definition: Let be a nondegenerate holomorphic line bundle on a complex torus ; let its first Chern class be
given by a Hermitian form with exactly negative eigenvalues; then we say that the
first Chern class of is a polarization of index on .

Definition. Let be a complex torus and a nondegenerate holomorphic line bundle


on defining a polarization of type (1 , . . . , ) with an index. One can prove that the
map
1 1 :

is an isogeny and that there is a line bundle on such that = 1 1 ; the polarization with an index on given by is called dual polarization (with index) (see
[166]).

Theorem (Riemann relations). Let = / be a complex torus of dimension ; let


{1 , . . . , } be a complex basis of , { 1 , . . . , 2 } be a basis of and be the matrix
expressing { 1 , . . . , 2 } in function of {1 , . . . , } (thus /2 ). We have that
is an Abelian variety if and only if there is a nondegenerate alternating matrix
(2 2, ) such that
(i) 1 = 0;
(ii) 1 is positive definite;
is the matrix expressing the alternating form defining the polarization in the basis
{ 1 , . . . , 2 }.
Theorem. Let be an Abelian variety with an ample holomorphic line bundle of type
(1 , . . . , ). Let be the rational map on associated to (see Bundles, fibre -).
(1) If 1 2 the map is holomorphic.
(2) (Lefschetz) If 1 3, then is an embedding.
Definition. We define the Siegel upper-half space

H := { ( , )| = and > 0},

where > 0 means that the imaginary part of is positive definite.

Suppose that = / is an Abelian variety of dimension and let be a Hermitian


form on defining a polarization of type (1 , . . . , ); let { 1 , . . . , , +1 , . . . , 2 } be
a symplectic basis of , i.e., a basis in which = is expressed by the matrix
(

0
),
0

Toric varieties

197

where is the diagonal matrix with diagonal (1 , . . . , ); let := 1 + for =

1, . . . , ; one can prove that {1 , . . . , } is a complex basis of ; the matrix expressing { 1 , . . . , 2 } in function of {1 , . . . , } is of the form = (, ) for some
( , ). We can prove that H .
Conversely, an element H determines a polarized Abelian variety with polarization of type (1 , . . . , ) with a symplectic basis: the torus = /(, )2 , with the
0

polarization whose imaginary part is expressed by the matrix ( 0 ) in the basis of the lattice given by the columns of the matrix (, ) (the imaginary part of the
polarization determines the whole polarization by Remark ).

Theorem. Fix a type (1 , . . . , ). The space H is the moduli space of polarized


Abelian varieties of type (1 , . . . , ) with symplectic basis. Moreover, if 1 , 2 H ,
the polarized Abelian varieties of type (1 , . . . , ) determined by 1 and 2 are isomorphic if and only if
2 = (1 + )(1 + )1
for some (

) (2 2, ) such that

0
0

) 2 (
)(
) 2 ,
0
0

where is the diagonal matrix with diagonal (1 , . . . , ), and

0

0

).
)=(
)(
)(
0

0

More generally, we can define an Abelian variety over an algebraic closed field
to be a complete algebraic variety which is also a group and such that the group
operation and the map associating to any element its inverse are morphisms.
We can prove that the group operation, which is usually called a sum, is necessarily
Abelian. Many properties we have seen for complex Abelian varieties also hold for
Abelian varieties over . We refer to [163] and [193].

See also Albanese varieties, Elliptic Riemann surfaces, elliptic curves Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard groups, Jacobians of compact Riemann surfaces, Jacobians, Weil and Griffiths intermediate -.
Sometimes (but not in this book) the term complex torus is used also to denote ( )
for some , where is {0}.

Toric varieties.

([52], [56], [67], [70], [78], [141], [205], [206]). A (complex) toric variety is an algebraic variety over such that there is an action of the group ( ) on
for some and such an action has a dense orbit.
A normal toric variety can be constructed as follows.

198 | Toric varieties


Let be a -module isomorphic to and let = (, ). If and
, we say that if () = 0.
Let be a rational strongly convex polyhedral cone in , i.e., a subset of
such that there exists a finite number of elements of , 1 , . . . , , such that
= + 1 + + +

and the unique linear subspace contained in is {0} (here + = { | 0}). We


define
:= { | () 0 },
:= .

We can prove that the semigroup is finitely generated (Gordons Lemma). Let
1 , . . . , be a set of generators of the semigroup .
The affine toric variety associated to can be defined in the following way. Consider [ ], the -algebra generated by for all with the property

for any , . Thus

= + ,

[ ] = [1 , . . . , ]/,

where is the ideal generated by the binomials:

for , such that

(1 )1 ( ) (1 )1 ( )
= .

=1,...,

=1,...,

Define to be the zero locus in of the ideal ; in the language of the schemes (see
Schemes)
= ([ ]) = ([1 , . . . , ]/).

The set of the closed points of correspond to the set of the maximal ideals of [ ],
([ ]), and there is a bijection from ([ ]) to ( , ) (the set of the
semigroup homomorphisms from to ) defined by
where () := () for all .

Definition. Let be a rational strongly convex polyhedral cone. We say that is a


face of if there exists such that = {} .
Thus we have
= + ().

Toric varieties

199

Then
[ ] [ ]

and is an open subset of complementary of the zero locus of the equation


= 0.

Definition. A fan in is a set of rational strongly convex polyhedral cones in


such that
(1) every face of an element of is an element of ;
(2) the intersection of two elements of is a face of each of them.

The toric variety () associated to the fan is obtained by considering the union
of for and gluing them by identifying the images of in and .

One can prove that in this way we obtain a normal toric variety of dimension and
every normal toric variety is equal to () for some finite fan (see [141, 1.2, Theorem 6]).
The action of ( ) is the following: identify ( ) with (, ); identify every
closed point of with a semigroup homomorphism . Let (, );
is the semigroup homomorphism such that ()() = ()().
Example. Let be the following fan: 1 = 1 , 2 , 2 = 1 , 2 (see Figure 21).

Fig. 21. A fan.

Then 1 = 1 , 2 , 2 = 1 , 2 and 1 2 = 1 , 1 , 2 .
Thus [1 ] = [1 , 2 ] and 1 = 2 , [2 ] = [1 , 2 ] and 2 = 2 ; finally,
[1 2 ] = [1 , 2 , 1 ]/1 1 = 1 and then 1 2 = . Therefore () =
1 .
Theorem. Let be a rational strongly convex polyhedral cone in . Then is
smooth if and only if is generated by a part of a -basis of . If is smooth, then
= ( ) where = .
Notation. For every fan , let || be .

Theorem. Let be fan. Then () is compact if and only if is finite and || =


.

200 | Toric varieties


Theorem. Let be a finite fan in . Then

1 (()) /|| ,

where 1 denotes the first fundamental group (see Fundamental group). In particular, if contains a cone of dimension , then () is simply connected. Moreover, for
any cone in ,
( , ) ( ),

where denotes the singular cohomology (see Singular homology and cohomology).

Theorem. Let be a finite fan. The set of the Weil divisors (see Divisors) in () that
are invariant for the action of ( ) is
edges of

(()),

where () is the fan given by the cones of containing .


A ( ) -invariant Cartier divisor is the zero locus of for some .
If is a fan such that || is not contained in any proper subspace of , then there
is a commutative diagram
0
0


/

/ {( ) -invar. C-divisors}


/ {( ) -invar. W-divisors}

/ (()) = {C-divisors}/

/ {W-divisors}/

/0
/0

whose rows are exact, the vertical maps are injective and is defined by () =
{ = 0} and where C-divisors stands for Cartier divisors, W-divisors stands for
Weil divisors, stands for linear equivalence and (()) is the Picard group of
() (see Equivalence, algebraic, rational, linear -, Chow, NeronSeveri and Picard
groups). In particular
((())) ( edges in ) .

Moreover, (()) is torsion free.

For any ( ) -invariant Cartier divisor on (), define


: ||

in the following way: for any cone of , let

| () = ()(),

where () is the element such that = {() = 0}.

Toric varieties

201

Theorem. Let be a ( ) -invariant Cartier divisor on (). Let O() be the sheaf of
the sections of the bundle associated to (see Divisors).
If () is compact, then is ample if and only if is strictly upper convex (i.e., for
all , the graph of on the complementary of is strictly under the graph of ()).
Moreover,
0 ((), O()) = , on || .
More generally, for all 0,

((), O()) (||, ||, , ),

where ||, := { || | () < ()} and the at the second member denotes the
singular relative cohomology.

Now we are describing a way to construct toric varieties from convex polytopes.
A subset of is said to be a rational polytope if it is the convex hull of a finite set
of points of . Let be an -dimensional rational polytope in .
For every face of , let be the cone defined by
= { | () () and }.

We can easily prove that the cones , for varying among the faces of , form a fan .
We define the toric variety associated to to be ( ).
Finally we sketch another approach to toric varieties (see [78]).

For any ( ) and any , let denote 11 . . . . Let = {1 , . . . , } be a

finite subset of ; we define to be


1

:= {[ : : ] | ( ) }

(the closure in 1 ) and we define to be


1

:= {(+1 , . . . , +1 ) | = (1 , . . . , ) ( ) , +1 }

(the closure in ).
Obviously is the closure of the orbit of [1 : : 1] under the action of ( ) on 1
given by the formula
1

[1 : : ] = [ 1 : : ]

for ( ) , [1 : : ] 1 , so it is a toric variety. Vice versa one can easily


prove the following proposition.

Proposition. Let be a projective toric variety in 1 such that the action of the torus
( ) extends to the whole 1 . Let 1 be the dimension of the minimal projective
space in 1 containing . Then there exists a subset of containing exactly

202 | Transcendence degree


elements such that there exists an isomorphism from to equivariant under the
torus action and extending to an equivariant projective isomorphism from 1 to the
minimal projective space in 1 containing .

The link between this approach and the first description we have given of normal toric
varieties is the following:
let = {1 , . . . , } a finite subset of ; let be the semigroup generated by the
elements (, 1) for ; we can prove that = [ ]; besides we can prove
that, if is a finitely generated semigroup of +1 containing 0, then [] is an
affine toric variety and it is normal if and only if is the intersection of the convex hull
of and the Abelian group generated by .
The theory of toric varieties has been developed more generally over algebraically
closed fields, see [141] and [206].

Transcendence degree. ([12], [62], [164], [256]). Let be a field and let be a
subfield of . A subset {1 , . . . , } of is said to be algebraically independent over
if there doesnt exist a nonzero polynomial in variables with coefficients in such
that (1 , . . . , ) = 0. The transcendence degree of over is the largest cardinality
of a subset of algebraically independent over .

Theorem. If is an integral domain and a finitely generated -algebra for some field
, then the dimension of is the transcendence degree over of the quotient field
of .
See Dimension, Localization, quotient ring, quotient field, Siegels theorem.

Transcendental. The word transcendental in algebraic geometry means concerning complex analysis.

U
Unirational, Lroth problem. ([10], [48], [107], [132], [228], [254]) We say that
an algebraic variety of dimension over a field is unirational if there exists a dominant rational map (see Varieties, algebraic -, Zariski topology, regular and
rational functions, morphisms and rational maps for the definition of dominant). The
Lroth problem can be formulated as follows: is any unirational variety rational?
(see Rational varieties).
If the dimension of is 1, the answer is yes for any field (Lroths theorem).
If the dimension of is 2, the answer is yes if = (Castelnuovos theorem).

Vanishing theorems

203

In higher dimension, the answer is no, also over : the first counterexamples were
given by Iskovskikh and Manin, Artin and Mumford, and Clemens and Griffiths. In
particular, a smooth cubic threefold in 4 is unirational but not rational, see [48].

Universal bundle.

See Tautological (or universal) bundle.

V
Vanishing theorems.

([4], [66], [93], [98], [107], [138], [149], [169], [173], [174],
[231], [243]). (See Bundles, fibre -, Positive, Dimension, Canonical bundle,
canonical sheaf and Sheaves for the definitions of nef, big, ample, positive, dimension, canonical bundle, cohomology of sheaves.)
Grothendiecks theorem. Let F be a sheaf of Abelian groups on a Noetherian topological space of dimension (we say that a topological space is Noetherian if for every
chain of closed subset 1 2 , there exists such that = +1 = ). Then
(, F) = 0

for > ,

where denotes the derived functor cohomology.

AkizukiKodairaNakano theorem. Let be a positive (thus ample) holomorphic line


bundle on a compact complex manifold of dimension (thus is projective algebraic, see Kodaira Embedding theorem, Remmerts proper mapping theorem,
Chows theorem). Then
(, ()) = 0

for + > .

Observe that in particular (, O( )) = 0 for > 0, where is the canonical


bundle. The last statement is generalized by Kawamata-Viehweg theorem (the case
= 2 was proved first by Ramanujam):

KawamataViehweg theorem. Let be a nef and big holomorphic line bundle on a


smooth complex projective algebraic variety of dimension . Then
(, O( )) = 0

for > 0 .

Le Potiers theorem generalizes the AkizukiKodairaNakano theorem to a bundle of


any rank.
Le Potiers theorem. Let be an ample holomorphic vector bundle of rank on a compact complex manifold of dimension . Then
(, ()) = 0

for + + .

204 | Varieties, algebraic More generally, for 1 ,

(, ( )) = 0

See also CartanSerre theorems.

for + > + ( ) .

Varieties, algebraic -, Zariski topology, regular and rational functions, morphisms and rational maps. ([73], [104], [107], [140], [159], [228]).

Let be a field. Let be the affine -space over . For any [1 , . . . , ], we


denote the zero locus of by (), i.e.,
() = { | () = 0 }.

Furthermore, for any , we define

() = { [1 , . . . , ]| () = 0 }

We can prove easily that () is an ideal; we call it the ideal of .

A subset of is said to be an (affine) algebraic set if it is equal to () for some


[1 , . . . , ].
The Zariski topology on is the topology such that the closed subsets are the algebraic sets (it is a topology by items 3 and 4 of the following theorem).

For any ideal in a ring , we define the radical ideal of , which we denote by , as
follows:
:= { | {0} s.t. }.

We say that an ideal is radical if and only if = .

Theorem.
(1) For any subset of [1 , . . . , ], we have that () = (), where is the ideal
generated by .
(2) If 1 2 [1 , . . . , ], then (1 ) (2 ).
(3) If is a collection of ideals in [1 , . . . , ], then ( ) = ( ), so the intersection of algebraic sets is an algebraic set.
(4) () () = () for any , ideals in [1 , . . . , ], so the finite union of algebraic sets is an algebraic set.
(5) If 1 2 , then (1 ) (2 ).
(6) (1 2 ) = (1 ) (2 ) for any 1 , 2 .
(7) (()) for any [1 , . . . , ].
(8) (()) for any , more precisely (()) = (the closure of for the
Zariski topology).
(9) () is a radical ideal for any .

Varieties, algebraic -

205

(10) (Hilbert basis theorem) If is a Noetherian ring, then [1 , . . . , ] is Noetherian; in particular for any field , [1 , . . . , ] is Noetherian, thus any ideal of
[1 , . . . , ] is finitely generated.
(11) (Hilberts Nullstellensatz) Let be an algebraically closed field. For any ideal in
[1 , . . . , ] we have:
(()) = ;
therefore, if () = 0, then = [1 , . . . , ] and then = [1 , . . . , ].
In particular there is a bijection between the set of the algebraic subsets of and
the set of the radical ideals of [1 , . . . , ].
(12) An algebraic subset of is irreducible in the Zariski topology (see Irreducible
topological space) if and only if () is prime.

Let be the -projective space over . For any subset of homogeneous elements of
[0 , . . . , ], we denote the zero locus of by (), i.e.,
() = { | () = 0 }.

For any homogeneous ideal (see Homogeneous ideals), we define () to be the


zero locus of the set of the homogeneous elements of . Moreover, for any , we
denote by () the ideal generated by
{ [0 , . . . , ]| homogeneous and () = 0 }.

Obviously it is a homogeneous ideal. We call it the ideal of .

A subset of is said to be a (projective) algebraic set if it is equal to () for some


subset of homogeneous elements of [0 , . . . , ].
For the projective algebraic sets, we have properties analogous to the ones for the
affine algebraic sets. See Hilberts Nullstellensatz for the projective version of
Hilberts Nullstellensatz.
The Zariski topology on is the topology such that the closed subsets are the algebraic sets.
From now on, let be an algebraically closed field.
A subset of is said to be an affine algebraic variety if it is an irreducible algebraic
set.
A quasi-affine algebraic variety in is an open subset of an affine algebraic variety.
A projective algebraic variety is an irreducible algebraic subset of .
A quasi-projective algebraic variety is an open subset of a projective algebraic variety in .

The quasi-affine algebraic varieties in and the quasi-projective algebraic varieties


in are called algebraic varieties (obviously affine and projective algebraic vari

eties are included). More generally, an algebraic variety in 1 is an


open subset of an irreducible algebraic subset.

206 | Varieties, algebraic

The Zariski topology on an algebraic variety in 1 is the topology

induced by the Zariski topology of 1 (i.e., by the topology where the


closed subsets are the algebraic subsets).

A topological space is said to be Noetherian if for any sequence 1 2 of closed


subsets, there is an integer such that = +1 = .

Proposition. In a Noetherian topological space, every nonempty closed subset can


be written uniquely as a finite union of irreducible closed subsets, one not contained
in the other. These irreducible closed subsets are called the irreducible components
of .

The spaces and and more generally 1 with the Zariski topology
are Noetherian topological spaces, so that the proposition above can be applied to
them with the Zariski topology, and so every algebraic set can be expressed as a finite
union of irreducible closed subsets (i.e., varieties), each not contained in another.
If is an algebraic set in , then we define the affine coordinate ring () to be
[1 , . . . , ]/().

Observe that any element of () defines a function from to .


If is an algebraic set in , then we define the projective coordinate ring ()
to be
[0 , . . . , ]/().

We say that a function : on a quasi-affine algebraic variety is regular in a


point if there is an open neighborhood of in such that = / for some
and polynomials with nowhere zero on .

We say that a function : on a quasi-projective algebraic variety is regular in


a point if there is an open neighborhood of in such that = / for some
and homogeneous polynomials of the same degree with nowhere zero on .

We say that a function : is regular on if it is regular in any point of .


The sheaf of the regular functions on a variety is denoted by O .
The stalk O, of O in a point is a local ring (see Local) and the unique
maximal ideal, denoted by , is given by the elements of O, that vanish in (it is
the unique maximal ideal because every element of O, is invertible). Obviously
O, / .

A rational function on an algebraic variety is an equivalence class of couples (, )


where is an open subset and a regular function on and (, ) and ( , ) are
equivalent if and only if = on .
The set of the rational functions on a variety form a field, which is called the rational
functions field of and denoted by ().

Varieties and subvarieties, analytic -

207

Theorem.
(a) If is an affine algebraic variety in , we have
O() = () (which is a domain since () is prime);
() is the quotient field of ();
for any , the stalk O, is the localization of () with respect to (),
where () is the ideal in () given by the elements that are 0 in (see
Localization, quotient ring, quotient field). Thus O, is a Noetherian (see
Noetherian, Artinian) local domain. The map is a bijection between the points of and the maximal ideals of ().

(b) If is a projective algebraic variety in , we have


O() = ;
() is the subring of the elements of degree 0 in the localization of ()
with respect the multiplicative system given by the nonzero homogeneous elements of ();
for any , O, is the subring of the elements of degree 0 in the localization of () with respect the multiplicative system given by the homogeneous
elements of () such that () = 0.

Let and be two algebraic varieties. A morphism : is a continuous map


such that, for any open subset of and for any regular function on , the map
: 1 () is regular.
A rational map : is an equivalence class of couples (, ) where is an
open subset and a morphism on and (, ) and ( , ) are equivalent if =
on . We say that it is dominant if there exists one of these couples, (, ), such
that the image of is dense in .
We say that a rational map : is birational if there exists a rational map
: such that = as rational map.
We say that two varieties and are birational equivalent (or birational) if there is
a birational map from to .
Example. The blowing-up is the typical example of birational map (see Blowing-up
(or -process)).

Theorem. Two algebraic varieties (over ) are birational equivalent if and only if their
rational functions fields are isomorphic as -algebras.

See Regular rings, smooth points, singular points, Zariski tangent space, differential forms, tangent bundle, normal bundle, Schemes, Primary ideals, primary
decompositions, embedded ideals, G.A.G.A., Hironakas decomposition of birational maps, Unirational, Lroth problem.

Varieties and subvarieties, analytic -.

([93], [103], [121]). We say that a subset of is an analytic variety if locally it is the zero locus of a finite number holo

208 | Veronese embedding


morphic functions, precisely, if, for all , there exists an open subset of
containing and a finite number of holomorphic functions on , 1 , . . . , , such that
= { | 1 () = = () = 0}. More generally, an analytic subvariety
of a complex manifold is a subset that is locally the zero locus of a finite number of
holomorphic functions.

Veronese embedding.

([104], [107]). Let be an algebraically closed field and,


for every , let = be the projective space of dimension over .
) 1. The Veronese map of degree on is the map
Let , and = (+

, : ,

[0 : : ] [ : : ]I ,

where I := {(0 , . . . , )| 0 , . . . , 0 + + = }, that is, , is the map sending


[0 : : ] to the point whose coordinates are all the monomials of degree in
0 , . . . , . Obviously, the map , is an embedding (see Embedding).
Let , for I, be the coordinates on ; the image of the Veronese map is the zero
locus of the set of the following polynomials of degree 2:

with , , , I such that + = + .


The image of the Veronese map , is a smooth variety of dimension and degree
(see Degree of an algebraic subset), called Veronese variety.
If has characteristic 0, a coordinate-free way to describe the Veronese map , is the
following: let be a vector space over of dimension + 1; , is the map
() ( ),

for any , where () is the line generated by and is the image of


( times) through the natural map .

Web.

A web is a linear system (see Linear systems) of dimension 3.

Weighted projective spaces.

([24], [55], [57], ([104]). Let be an algebraically


closed field of characteristic 0. Let 0 , . . . , {0}. The weighted projective space of
weights 0 , . . . , over , which we denote by (0 , . . . , ), or simply by (0 , . . . , ),
is defined to be the quotient of +1 {0} by the following equivalence relation: for
any {0},
(0 , . . . , ) (0 0 , . . . , ).

Weierstrass form of cubic curves

209

We can define (0 , . . . , ) also to be the quotient of by the action of the group


/0 / generated by the automorphisms
[0 : : ] [0 : : : : ],

where is a primitive -th root of the unity.


The isomorphism from /(/0 / ) to (+1 {0})/ is given by sending

the class of [0 : : ] to the class of (00 , . . . , ).

Theorem.
(i) Let , 0 , . . . , {0}. Then the weighted projective spaces (0 , . . . , ) and
(0 , . . . , ) are isomorphic.
(ii) Let 0 , . . . , {0} with greatest common divisor equal to 1. For any = 0, . . . , ,
let
= (0 , . . . , 1 , +1 , . . . , );
= (0 , . . . , 1 , +1 , . . . , );

where stands for greatest common divisor and stands for least common
multiple. Then (0 , . . . , ) and (0 /0 , . . . , / ) are isomorphic.
(iii) For any 0 , . . . , {0}, the weighted projective space (0 , . . . , ) is a normal
projective algebraic variety (see Normal).

Weierstrass form of cubic curves.

([93], [107], [228], [246]). Let be an algebraically closed field of characteristic 0.


Let be a smooth cubic algebraic curve in 2 and be a homogeneous polynomial
in , , whose zero locus is . By Bezouts theorem applied to and to the Hessian
of , the curve must have a flex (see Bezouts theorem, Flexes). By changing
coordinates in 2 , we can suppose that = [0 : 1 : 0] and that the tangent of at ,
(), is the line = 0, thus that the coefficients of 3 and 2 are zero. Since is
irreducible, one of the coefficients of 3 and 2 must be nonzero. Furthermore, the
coefficient of 2 must be zero (otherwise would not be a flex). Thus we can suppose
that
(, , ) = 3 + (, , ),

where (, , ) is a polynomial of degree 2 with nonzero coefficient of 2 (otherwise


would be a singular point); we can suppose the coefficient of 2 equal to 1; thus
(, , ) = 2 + ( + ) + 2 + 2 +

for some , , , , . With a change of coordinates (the new equal to 12 (+)),


we can suppose
(, , ) = 3 + (2 + 2 + 2 + )

for some , , . With a new change of coordinates (the new equal to + 3 ) we


can suppose
(, , ) = 3 2 + 3 + 2

210 | Weierstrass points


for some , . One can easily see that is nonsingular if and only if the roots of
3 + 3 + 2 are distinct. We can suppose they are 0, 1, for some = 0, 1. Thus
(, , ) = 2 + ( )( ).

Therefore { = 1} is the set of the (, ) such that

2 = ( 1)( ),

which is called Weierstrass form.


See Elliptic Riemann surfaces, elliptic curves.

Weierstrass points. ([8], [69], [93], [101], [189]). By the geometric RiemannRoch
theorem (see Riemann surfaces (compact -) and algebraic curves), for a general divisor of degree on a compact Riemann surface of genus 1, we have
0 (O()) = {

1
+1

if 0 ,
if .

0 (O()) = {

1
+1

if 0 ,
if .

In particular, for general , we have

The points that have not this behaviour are called Weierstrass points.

For every let

= { | 0 (O()) = 0 (O(( 1)))}

The set ordered by increasing order is called Weierstrass gap sequence.

Remark. By using the RiemannRoch theorem one can easily prove that, for every
, the cardinality of is (in fact 0 (O()) = + 1 for every 2 1 and
0 (O(0)) = 1).

Obviously a point is a Weiersrtrass point if and only if = (1, . . . , ).

Let be a Weierstrass point, and let = (1 , . . . , ) (with 1 < < ). We define


the weight of to be
() = ( ).
=1,...,

Weierstrass preparation theorem and Weierstrass division theorem

211

Proposition. Let be a Riemann surface of genus 1.


If = 1, there are no Weierstrass points on .
If 2, there exist at least 2 + 2 Weierstrass points on (and the equality holds if
and only if is hyperelliptic). Besides the sum of the weights of the Weierstrass points
on is ( 1)( + 1); therefore the maximum number of the Weierstrass points on a
Riemann surface of genus is ( 1)( + 1).

Weierstrass preparation theorem and Weierstrass division theorem. ([93], [121], [151]).

Weierstrass preparation theorem. Let be a neighborhood of 0 in ; call 1 , . . . ,


the coordinates in and let = (1 , . . . , 1 ). Let : be holomorphic, not
identically zero on the -axis and such that (0) = 0. Then, in a neighborhood of 0 in
, the function can be written uniquely as
(, ) = (, )( + 1 ()1 + + ()),

for some 1, , holomorphic functions such that (0) = 0 and (0) = 0 for
all .

Corollary. The stalk of the sheaf of the holomorphic functions in a point of is a


unique factorization domain.
Weierstrass division theorem. Let be a neighborhood of 0 in and call 1 , . . . ,
the coordinates in and let = (1 , . . . , 1 ). Let : be holomorphic. Let
(, ) = + 1 ()1 + + (),

where are holomorphic functions such that (0) = 0 for all . Then, in a neighborhood of 0, the function can be written as
= + ,

where and are holomorphic and can be written as a polynomial of degree 1


in with some holomorphic functions in as coefficients.

Corollary: Weak Nullstellensatz. Let and be holomorphic maps from a neighborhood of 0 in to . Let be irreducible (in the ring of the holomorphic functions defined in some neighborhood of 0) and let vanish on the zero locus of . Then =
for some holomorphic function on a neighborhood of 0.
See Hilberts Nullstellensatz.

212 | Zariskis main theorem

Z
Zariskis main theorem. ([107], [241], [255]). There are several statements that
are called Zariskis main theorem. We report the versions of Zariskis main theorem in
[107] and [241] respectively.
Theorem. Let and be Noetherian integral schemes (see Schemes). Suppose that
is normal, that is, all the stalks of the structure sheaf are integrally closed domains
(see Normal, Integrally closed). Let : be a birational projective morphism. Then 1 () is connected for any .

Theorem. Let and be reduced irreducible complex analytic spaces (see Spaces,
analytic -) and suppose is normal (reduced means that all the stalks of the structure sheaf have no nilpotent elements and normal that they are integrally closed domains; irreducible means that the underlying topological space is irreducible; see Irreducible topological space). Let : be a proper surjective morphism with
finite fibres and suppose there is an open subset of such that 1 () is a point for
all . Then is an isomorphism.

Zariski tangent space, differential forms, tangent bundle, normal


bundle. ([104], [107], [140], [228]). Let be an algebraically closed field. The Zariski

tangent space of an algebraic variety at a point , denoted by , , is defined to be


the dual of /2 , where is the maximal ideal of O, and O, is the stalk in of
the sheaf of the regular functions on (see Varieties, algebraic -, Zariski topology,
regular and rational functions, morphisms and rational maps):
, = ( /2 ) .

If is an affine algebraic variety in and 1 , . . . , generate (), we can define the


tangent space of at as

= {(1 , . . . , )

=1,...,

()( ) = 0 for = 1, . . . , }

(observe that the partial derivatives and the differential of a polynomial can be defined
in an obvious way over any field by applying the usual rules for derivatives).
We can prove that the differential at gives an isomorphism from /2 to ,
so the Zariski tangent space , is isomorphic to . In particular for any quasiprojective algebraic variety the Zariski tangent space at is isomorphic to the
tangent space at of any of its affine neighborhoods. If is a projective variety contained in (with homogeneous coordinates 0 , . . . , ), the closure in of the tangent space at of { = 0} does not depend on , for such that = 0, and it too is
often called the tangent space of at .

Zariski topology

213

Let be an algebraic variety over . We say that a function associating to every point

is a regular differential -form if every point has


an element of ,
an open neighborhood such that restricted to can be written as linear combination with coefficients in the coordinate ring () of elements of the kind 1
with 1 , . . . , (), where is the differential (see Varieties, algebraic -, Zariski
topology, regular and rational functions, morphisms and rational maps for the definition of coordinate ring). For any open subset , denote by the set of the regular
differential -forms on . Let be the sheaf of modules over O assigning to any
open subset of .

Furthermore, consider the set of the pairs (, ) where is an open subset of and
is a regular differential -form on . We say that (, ) and ( , ) are equivalent if
= on . A rational differential -form on is an equivalence class of a pair.
Theorem. If is smooth, the sheaf is a locally free sheaf.

Thus, if is smooth, for any , the sheaf determines a vector bundle. The bundle determined by 1 is called cotangent bundle. Obviously the bundle determined by
is the -wedge product of the cotangent bundle. The dual of the cotangent bundle
is called tangent bundle; we denote it by ; its fibre in every point is the
Zariski tangent space at .
Moreover, if is a smooth variety and a smooth closed subvariety, we define the
normal bundle , to in to be the quotient of the restriction of to by :
| = | / .

Zariski topology. See Varieties, algebraic -, Zariski topology, regular and rational
functions, morphisms and rational maps.

Bibliography
[1]
[2]
[3]
[4]
[5]
[6]
[7]

[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]

[16]
[17]
[18]

[19]
[20]
[21]
[22]
[23]

D. Abramovich and A. J. de Jong, Smoothness, semistability, and toroidal geometry, J. Algebraic Geom. 6(4), 789801 (1997).
D. Abramovich, K. Karu, K. Matsuki, and J. Wlodarczyk, Torification and factorization of birational maps, J. Amer. Math. Soc. 15(3), 531572 (2002).
J. F. Adams, Lectures on Lie groups, W. A. Benjamin, Inc., 1969.
Y. Akizuki and S. Nakano, Note on KodairaSpencers proof of Lefschetz theorems, Proc.
Japan Acad. 30, 266272 (1954).
V. Ancona and G. Ottaviani, An introduction to the derived categories and the theorem of
Beilinson, Atti Accad. Peloritana Pericolanti, Cl. Sci. Fis. Mat. Natur. 67, 99110 (1989).
A. Andreotti and T. Frankel, The Lefschetz theorem on hyperplane sections, Ann. of Math. (2)
69, 713717 (1959).
E. Arbarello, M. Cornalba, and P. A. Griffiths, Geometry of algebraic curves, Vol. II, with a contribution by Joseph Daniel Harris, Grundlehren der Mathematischen Wissenschaften Vol. 268,
Springer-Verlag, 2011.
E. Arbarello, M. Cornalba, P. A. Griffiths, and J. Harris, Geometry of algebraic curves,
Grundlehren der Mathematischen Wissenschaften Vol. 267, Springer-Verlag, 1985.
E. Arrondo, A home-made Hartshorne-Serre correspondence, Rev. Mat. Complut. 20(2), 423
443 (2007).
M. Artin and D. Mumford, Some elementary examples of unirational varieties which are not
rational, Proc. London Math. Soc. 25(3), 7595 (1972).
M. F. Atiyah, A survey of K-theory, K-theory and operator algebras, in: Proc. Conf., Univ. Georgia, Athens, Ga., 1975, pp. 19, Lecture Notes in Math. 575, Springer-Verlag, 1977.
M. F. Atyah and I. G. Macdonald, Introduction to commutative algebra, Addison-Wesley Publishing Co., 1969.
M. F. Atiyah and I. M. Singer, The index of elliptic operators on compact manifolds, Bull. Amer.
Math. Soc. 69, 422433 (1963).
W. L. Baily, On the theory of -functions, the moduli of Abelian varieties, and the moduli of
curves, Ann. of Math. 75(2), 342381 (1962).
C. Banica, Smooth reflexive sheaves, in: Proceedings of the Colloquium on Complex Analysis
and the Sixth Romanian-Finnish Seminar, Rev. Roumaine Math. Pures Appl. 36 (910), 571
593 (1991).
F. Bardelli, Cicli algebrici su variet proiettive, Geometry Seminars, 19911993, 4765, Univ.
Stud. Bologna, 1994,
W. Barth and K. Hulek, Monads and moduli of vector bundles, Manuscripta Math. 25 (4), 323
347 (1978).
W. Barth, K. Hulek, C. Peters, and A. Van de Ven, Compact complex surfaces, 2nd ed., Ergebnisse der Mathematik und ihrer Grenzgebiete, 3. Folge. A Series of Modern Surveys in Mathematics 4, Springer-Verlag, 2004.
W. Barth and A. Van de Ven, A decomposability criterion for algebraic 2-bundles on projective
spaces, Invent. Math. 25, 91106 (1974).
H. Bass, On the ubiquity of Gorenstein rings, Math. Z. 82, 828 (1963).
H. Bass, Algebraic K-theory, W. A. Benjamin, Inc., 1968.
A. Beauville, Complex algebraic surfaces, 2nd ed., London Mathematical Society Student
Texts 34, Cambridge University Press, 1996.
A. A. Beilinson, Coherent sheaves on and problems of linear algebra, Functional Anal.
Appl. 12 (3), 214216 (1978).

216 | Bibliography
[24]
[25]

[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]

M. Beltrametti and L. Robbiano, Introduction to the theory of weighted projective spaces,


Exposition. Math. 4 (2), 111162 (1986).
E. Bombieri and D. Mumford, Enriques classification of surfaces in char. p., Part I, Global
Analysis (Papers in Honor of K. Kodaira), 325339, Univ. Tokyo Press, 1969; Part II, Complex
Analysis and Algebraic Geometry, 2342, Iwanami Shoten (1977); Part III, Invent. Math. 35,
197232 (1976).
F. Borceux, Handbook of Categorical Algebra I, II, III, Encyclopedia of Mathematics and its
Applications Vols. 50, 51, 52, Cambridge University Press, 1994.
A. Borel, Linear algebraic groups, 2nd ed., Graduate Texts in Mathematics 126, SpringerVerlag, 1991.
A. Borel and R. Remmert, ber kompakte homogene Khlersche Mannigfaltigkeiten, Math.
Ann. 145, 429439 (1961/1962).
A. Borel and J. P. Serre, Le thorme de Riemann-Roch, Bull. Soc. Math. France 86, 97136
(1958).
R. Bott, Homogeneous vector bundles, Ann. of Math (2), 66 (2), 203248 (1957).
R. Bott, On a theorem of Lefschetz, Michigan Math. J. 6, 211216 (1959).
R. Bott and L. V. Tu, Differential forms in algebraic topology, Graduate Texts in Mathematics
82, Springer-Verlag, 1982,
G. E. Bredon, Topology and geometry, Corrected third printing of the 1993 original, Graduate
Texts in Mathematics . 139, Springer-Verlag, 1997.
G. E. Bredon, Sheaf theory, 2nd ed, Graduate Texts in Mathematics 170, Springer-Verlag,
1997.
E. Brieskorn, H. Knrrer, Plane algebraic curves, Birkhuser, 1986.
M. Brion, Lectures on the geometry of flag varieties, in: P. Pragacz (ed.), Topics in cohomological studies of algebraic varieties, pp. 3385, Trends Math., Birkhuser, 2005.
R. Brown, Groupoids and Van Kampens Theorem, Proc. London Math. Society 17(3), 385401
(1967).
R. Brown, Topology and Groupoids, BookSurge, LLC, 2006.
W. Bruns and J. Herzog, Cohen-Macaulay Rings, Cambridge Studies in Advanced Mathematics
39, Cambridge University Press, 1993.
H. Cartan, Thorie lmentaire des functions analytiques dune ou plusieurs variables complexes, Avec le concours de Reiji Takahashi, Enseignement des Sciences, Hermann, 1961.
H. Cartan and S. Eilenberg, Homological algebra, Princeton University Press, 1956.
H. Cartan and J.-P. Serre, Un thorme de finitude concernant les varits analytiques compactes, C. R. Acad. Sci. Paris 237, 128130 (1953).
F. Catanese, Moduli of algebraic surfaces, in: E. Ernesi (ed.), Theory of Moduli, Montecatini
Terme, 1985, Lectures Notes in Mathematics 1337, pp. 183, Springer-Verlag, 1988.
S. S. Chern, Complex manifolds, Textos de Matemtica 5, Instituto de Fsica e Mtematica,
Universidade do Recife, 1959.
S. S. Chern, Characteristic classes of Hermitian manifolds, Ann. of Math. 47(2), 85121 (1946).
S. S. Chern, W. H. Chen, and K. S. Lam, Lectures on Differential Geometry, Series on University
Mathematics 1, World Scientific Publishing Co., Inc., 1999.
C. Chevalley, The algebraic theory of spinors, Columbia University Press, 1954.
C. H. Clemens and P. Griffiths, The intermediate Jacobian of the cubic threefold, Ann. of Math.
95(2), 281356 (1972),
C. H. Clemens, J. Kollar, and S. Mori, Higher dimensional complex geometry, Asterisque 166,
1988.

Bibliography
[50]

[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]

217

D. Cox, J. Little, and D. OShea, Ideals, Varieties and Algorithms. An introduction to computational algebraic geometry and commutative algebra, 3rd ed., Undergraduate Texts in Mathematics, Springer-Verlag, 1992.
D.Cox, J. Little, and D. OShea, Using Algebraic Geometry, 2nd ed., Graduate Texts in Mathematics 185, Springer-Verlag, 1998.
D.Cox, J. Little, and H. Schenck, Toric varieties, Graduate Studies in Mathematics 124, Amer.
Math. Soc., 2011.
P. Deligne, Thorme de Lefschetz et critres de dgnerescence de suites spectrales, Inst.
Hautes Etudes Sci. Publ. Math. 35, 259278 (1968).
P. Deligne and D. Mumford, The irreducibility of the space of curves of given genus, Inst.
Hautes tudes Sci. Publ. Math. 36, 75109 (1969).
C. Delorme, Espaces projectifs anisotropes, Bull. Soc. Math. France 103 (2), 203223 (1975).
M. Demazure, Sous-groupes algbriques de rang maximum du groupe de Cremona, Ann. Sci.
cole Norm. Sup. 3(4), 507588 (1970).
I. Dolgachev, Weighted projective varieties, in: Proc. Group actions and vector fields, Vancouver, B. C., 1981, pp. 3471, Lectures Notes in Math. 956, Springer-Verlag, 1982.
C. Ehresmann, Sur les espaces fibrs diffrentiables, C. R. Acd. Sci. Paris 224, 16111612
(1947).
L. Ein, Varieties with small dual varieties, I, Invent. Math. 86 (1), 6374 (1986).
L. Ein, Varieties with small dual varieties, II, Duke Math. J. 52 (4), 895907 (1985).
L. Ein, Ph. Ellia, and F. Zak, Projective varieties, linear systems and vector bundles, Notes of
courses given in Istituto Nazionale di Alta Matematica F. Severi Roma and Ferrara, 1992.
D. Eisenbud, Commutative algebra with a view toward algebraic geometry, Graduate Texts in
Mathematics 150, Springer-Verlag, 1995.
D. Eisenbud, G. Floystad, and F. Schreyer, Sheaf cohomology and free resolutions over exterior algebras, Trans. A. M. S. 355 (11), 43794426 (2003).
D. Eisenbud and J. Harris, The geometry of schemes, Graduate Texts in Mathematics 197,
Springer-Verlag, 2000.
F. Enriques, Le Superficie Algebriche, Zanichelli, 1949.
H. Esnault and E. Viehweg, Lectures on vanishing theorems, DMV Seminar 20, Birkhuser,
1992.
G. Ewald, Combinatorial convexity and algebraic geometry, Graduate Texts in Mathematics,
168, Springer-Verlag, 1996.
G. Fano, Su alcune variet algebriche a tre dimensioni razionali e aventi curve-sezioni canoniche, Comment. Math. Helv. 14, 202211 (1942).
H. Farkas and I. Kra, Riemann Surfaces, 2nd ed., Graduate Texts in Mathematics 71, SpringerVerlag, 1992.
W. Fulton, Introduction to toric varieties, Annals of Mathematical Studies 131, Princeton University Press, 1993.
W. Fulton, Hurwitz schemes and irreducibility of moduli of algebraic curves, Ann. of Math. 90
(3), 542575 (1969).
W. Fulton, Intersection theory, 2nd ed., Ergebnisse der Mathematik und ihrer Grenzgebiete,
3. Folge. A Series of Modern Surveys in Mathematics, Springer-Verlag, 1998.
W. Fulton, Algebraic curves, an introduction to algebraic geometry, Mathematics Lecture
Notes Series, W. A. Benjamin, Inc., 1969.
W. Fulton, Introduction to intersection theory in algebraic geometry, CBMS Regional Conference Series in Mathematics 54, published for the Conference Board of the Mathematical
Sciences, Washington by the American Mathematical Society, 1984.

218 | Bibliography
[75]
[76]
[77]
[78]
[79]
[80]

[81]

[82]

[83]
[84]

[85]
[86]
[87]
[88]
[89]
[90]

[91]
[92]
[93]
[94]

[95]
[96]
[97]

W. Fulton, Young Tableaux, London Mathematical Society Student Texts 35, Cambridge University Press, 1997.
W. Fulton and J. Harris, Representation theory, a first course, Graduate Texts in Mathematics
129, Springer Verlag, 1991.
W. Fulton and R. Lazarsfeld, On the connectedness of degeneracy loci and sepcial divisors,
Acta Math. 146, 271283 (1981).
I. M. Gelfand, M. M. Kapranov, and A. V. Zelevinsky, Discriminants, resultants, and multidimensional determinants, Mathematics: Theory and Applications, Birkhuser, 1994.
S. Gelfand and Y. Manin, Methods of homological algebra, 2nd ed., Springer Monographs in
Mathematics, Springer-Verlag, 2003.
F. Ghione and G. Ottaviani, A tribute to Corrado Segre, in: Proceed. Complex Projective Geometry, Trieste, 1989/Bergen, 1989, London Math. Soc. Lecture Notes Ser. 179, pp. 175188,
Cambridge University Press, 1992.
D. Gieseker, Lectures on Moduli of Curves, Tata Institute of Fundamental Research, Lectures
on Mathematics and Physics 69, published for the Tata Institute of Fundamental Research,
Bombay, Springer-Verlag, 1982.
D. Gieseker, Geometric invariant theory and applications to moduli problems, in: Proc. Invariant theory, Montecatini, 1982, pp. 4573, Lecture Notes in Math. 996, Springer-Verlag,
1983.
D. Gieseker, On the moduli of vector bundles on an algebraic surface, Ann. of Math. 106(1)
4560 (1977).
R. Godement, Topologie algbrique et thorie des faisceaux, 3rd ed., Publications de lInstitut
de Mathmatique de lUniversit de Strasbourg, XIII, Actualits Scientifiques et Industrielles
1252, Hermann, 1973.
S. I. Goldberg, Curvature and homology, Pure and Applied Mathematics XI, Academic Press,
1962.
H. Grauert, Ein Theorem de analytischen Garbentheorie und die modulrume komplexer
Structuren, Inst. Hautes tudes Sci. Publ. Math. 5, 1960.
H. Grauert and G. Mlich, Vektorbndel vom Rang 2 ber dem n-dimensionalen komplexprojektiven Raum, Manuscripta Math. 16 (1), 75100 (1975).
M. Green, Koszul cohomology and the geometry of projective varieties I,II, J. Differential Geometry 20, 125171, 279289 (1984).
M. Green, Koszul cohomology and geometry, Lectures on Riemann Surfaces, pp. 177200,
World Scientific Press, 1989.
M. Green, J. Murre, and C. Voisin, Algebraic cycles and Hodge Theory, Lectures given at the
Second C.I.M.E. Session held in Torino, June 2129, 1993, A. Albano, F. Bardelli (eds.), Lecture
Notes in Math. 1594, Springer-Verlag, 1994.
M. J. Greenberg and J. Harper, Algebraic topology: a first course, Perseus Books, 1981.
P. Griffiths, Periods of integrals on algebraic manifolds I and II, Amer. J. Math. 90, 568626,
805865 (1968).
P. Griffiths and J. Harris, Principles of algebraic geometry, John Wiley & Sons, 1978.
M. Gross, D. Huybrechts, and D. Joyce, Calabi-Yau manifolds and related geometries, Lectures from the Summer School held in Nordfjordeid, June 2001, Universitext, Springer-Verlag,
2003.
A. Grothendieck, Elments de Gomtrie Algbrique I,II, III, IV, avec la collaboration de
J. Dieudonne, Inst. Hautes tudes Sci. Publ. Math., 19601967.
A. Grothendieck, La thorie des classes de Chern, Bull. Soc. Math. France 86, 137154 (1958).
A. Grothendieck, Sur la classification des fibres holomorphes sur la sphere de Riemann,
Amer. J. of Math. 79, 121138 (1957).

Bibliography
[98]
[99]

[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
[109]
[110]
[111]
[112]
[113]
[114]
[115]
[116]
[117]
[118]
[119]
[120]
[121]
[122]

219

A. Grothendieck, Sur quelques points dalgbre homologique, Thoku Math. J. 9(2), 119221
(1957).
A. Grothendieck, Techniques de construction en gomtrie analytique. I. Description axiomatique de lespace de Teichmller et de ses variantes, Sminaire Henri Cartan 13 no. 1, Exposs
No. 7 and 8, Paris: Secrtariat Mathmatique.
A. Grothendieck, Fondaments de la gomtrie algbrique, Sminaire Bourbaki 195762, Soc.
Math. France (1962).
R. C. Gunning, Lectures on Riemann surfaces, Princeton Mathematical Notes, Princeton University Press, 1966.
R. C. Gunning, Lectures on Riemann surfaces, Jacobi varieties, Mathematical Notes 12, Princeton University Press, 1972.
R. C. Gunning, and H. Rossi, Analytic functions of several complex variables, Prentice-Hall,
Inc., 1965.
J. Harris, Algebraic geometry, a first course, Graduate Texts in Mathematics 133, SpringerVerlag, 1992.
J. Harris and I. Morrison, Moduli of curves, Graduate Texts in Mathematics 187, SpringerVerlag, 1998.
J. Harris and L. W. Tu, On symmetric and skew-symmetric determinantal varieties, Topology
23, 7184 (1984).
R. Hartshorne, Algebraic Geometry, Graduate Texts in Mathematics 52, Springer-Verlag, 1974.
R. Hartshorne, Residues and duality, with an appendix by P. Deligne, Lectures Notes in Math.
20, Springer-Verlag, 1966.
R. Hartshorne, Varieties of small codimension in projective space, Bull. Amer. Math. Soc. 80,
10171032 (1974).
R. Hartshorne, Ample Subvarieties of Algebraic Varieties, Notes written in collaboration with
C. Musili, Lectures Notes in Mathematics 156 , Springer-Verlag, 1970.
R. Hartshorne, Connectedness of the Hilbert scheme, Inst. Hautes tudes Sci. Publ. Math. 29,
548 (1966).
A. Hatcher, Algebraic topology, Cambridge University Press, 2002.
H. Hauser, The Hironaka theorem on resolution of singularities (or: A proof we always wanted
to understand), Bull. Amer. Math. Soc. (N. S.) 40 (3), 323403 (2003).
H. Hauser, Resolution of singularities 18601999, in: Proc. Resolution of singularities, Obergurgl, 1997, Progress in Mathematics 181, 341373, Birkhuser, 2000.
S. Helgason, Differential geometry, Lie groups and Symmetric spaces, Pure and Applied Mathematics 80, Academic Press, 1978.
P. J. Hilton and U. Stammbach, A course in Homolgical Algebra, 2nd ed., Graduate Texts in
Mathematics 4, Springer-Verlag, 1997.
H. Hironaka, Resolution of singularities of an algebraic variety over a field of characteristic
zero III, Ann. of Math. (2) 79, 109203, 205326 (1964).
H. Hironaka, On resolution of singularities (characteristic zero), in: Proc. Internat. Cong.
Mathematicians, Stockholm, 1962, pp. 507521, Inst. MittagLeffler, 1962.
F. Hirzebruch, Topological Methods in Algebraic Geometry, 3rd ed., with appendices by R. L. E.
Schwarzenberger and A. Borel, Classics in Mathematics, Springer-Verlag, 1995.
W. V. D. Hodge and D. Pedoe, Methods of Algebraic Geometry, Cambridge University Press,
19471954.
L. Hrmander, An introduction to Complex Analysis in several variables, Elsevier, 1973.
G. Horrocks, Vector bundles on the punctured spectrum of a local ring, Proc. London Math.
Soc. 14, 689713 (1964).

220 | Bibliography
[123] G. Horrocks and D. Mumford, A rank 2 vector bundle on 4 with 15, 000 symmetries, Topology
12 (1), 6381 (1973).
[124] K. Hulek, The Horrocks-Mumford bundle, in: Proc. Vector bundles in algebraic geometry,
Durham, 1993, London Math. Soc. Lecture Note Ser. 208, pp. 139177 Cambridge University
Press, 1995.
[125] J. E. Humphreys, Introduction to Lie Algebras and Representation Theory, Graduate Texts in
Mathematics 9, Springer-Verlag, 1972.
[126] J. E. Humphreys, Linear Algebraic Groups, Graduate Texts in Mathematics 21, Springer-Verlag,
1972.
[127] D. Husemller, Elliptic Curves, 2nd ed., with appendices by O. Forster, R. Lawrence, and
S. Theisen, Graduate Texts in Mathematics 111, Springer-Verlag, 1987.
[128] D. Husemller, Fibre Bundles, 3rd ed., Graduate Texts in Mathematics 20, Springer-Verlag,
1994.
[129] S. Iitaka Algebraic Geometry. An Introduction to Birational Geometry of Algebraic Varieties,
Graduate Text in Mathematics 76, Springer-Verlag (1982).
[130] I. M. Isaacs, Character Theory of Finite Groups (corrected reprint of the 1976 original, published by Academic Press), AMS Chelsea Publishing, 2006.
[131] V. A. Iskovskih, Fano threefolds I,II, Izv. Akad. Nauk SSSR Ser. Mat. 41 (3), 516562 (1977) and
42 (3), 506549 (1978).
[132] V. A. Iskovskikh and Yu. I. Manin, Three-dimensional quartics and counterexamples to the
Lroth problem, Math. Sb. (N. S.) 86 (128), 140166 (1971).
[133] N. Jacobson, Lie Algebras, Interscience Tracts in Pure and Applied Mathematics 10 Interscience Publishers, 1962.
[134] T. Jsefiak, A. Lascoux, and P. Pragacz, Classes of determinantal varieties associated with
symmetric and skew-symmetric matrices, Izv. Akad. Nauk SSSR Ser. Mat. 45 (3), 662673
(1981).
[135] J. Jost, Riemannian Geometry and Geometric Analysis, 6th ed Universitext, Springer-Verlag,
2011.
[136] J. P. Jouanolou, Thormes de Bertini et applications, Progress in Mathematics 42,
Birkhuser, 1983.
[137] M. M. Kapranov, On the derived category of coherent sheaves on some homogeneous spaces,
Invent. Math. 92 (3), 479508 (1988).
[138] Y. Kawamata, A generalization of Kodaira-Ramanujams vanishing theorem, Math. Ann. 261
(1), 4346 (1982).
[139] G. Kempf, Complex Abelian varieties and theta functions, Universitext, Springer-Verlag, 1991.
[140] G. Kempf, Algebraic varieties, London Mathematical Society Lecture Note Series 172, Cambridge University Press, 1993.
[141] G. Kempf, F. F. Knudsen, D. Mumford, and B. Saint-Donat, Toroidal embeddings I, Lecture
Notes in Math. 339, Springer-Verlag, 1973.
[142] S. L. Kleiman, Toward a numerical theory of ampleness, Annals of Math. (2) 84, 293344
(1966).
[143] S. L. Kleiman, About the conormal scheme, in: Proc. Complete intersections, Acireale, 1983,
pp. 161197, Lecture Notes in Math. 1092, Springer-Verlag, 1984.
[144] A. Knapp, Elliptic curves, Mathemtical Notes 40, Princeton University Press, 1992.
[145] F. F. Knudsen,The projectivity of the moduli space of stable curves III, Math. Scand. 52 (2),
200212 (1983).
[146] S. Kobayashi, Differential geometry of complex vector bundles, Publications of the Mathematical Society of Japan 15, Iwanami Shoten Publishers and Princeton University Press, 1987.

Bibliography

221

[147] S. Kobayashi and K. Nomizu, Foundations of differential Geometry, vol. I and II, Interscience
Publishers (John Wiley & Sons), 1963, 1969.
[148] N. Koblitz, Introduction to Elliptic Curves and Modular Forms, Graduate Texts in Mathematics
97, Springer-Verlag, 1984.
[149] K. Kodaira, On a differential-geometric method in the theory of analytic stacks, Proc. Nat.
Acad. Sci. USA 39, 12681273 (1953).
[150] K. Kodaira, On Khler varieties of restricted type (an intrinsic characterization of algebraic
varieties), Ann. of Math. (2) 60 (1), 2848 (1954).
[151] K. Kodaira, Complex manifolds and deformation of complex structures, Grundlehren der Mathematischen Wissenschaften 283, Springer-Verlag, 1986.
[152] K. Kodaira and J. Morrow, Complex manifolds, Holt, Rinehart and Winston, Inc., 1971.
[153] K. Kodaira and D. C. Spencer, On deformations of complex analytic structure I, II, Ann. of
Math. (2) 67, 328401, 403466 (1958).
[154] K. Kodaira and D. C. Spencer, On deformations of complex analytic structure III, Stability
theorems for complex structure, Ann. of Math. (2) 71, 4376 (1960).
[155] J. Kollar, The structure of algebraic threefolds: an introduction to Moris program, Bull. Amer.
Math. Soc. (N. S.) 17 (2), 211273 (1987).
[156] J. Kollar, Lectures on resolution of singularities, Annals of Mathematical Studies 166, Princeton University Press, 2007.
[157] J. Kollar and S. Mori, Birational geometry of algebraic varieties, with the collaboration of
C. H. Clemens and A. Corti, Cambridge Tracts in Mathematics 134, Cambridge University
Press, 1998.
[158] C. Kosniowski, A first course in algebraic topology, Cambridge University Press, 1980.
[159] E. Kunz, Introduction to commutative algebra and algebraic geometry, Birkhuser, 1985.
[160] M. Kuranishi, New proof for the existence of locally complete families of complex structures,
in: Proc. Conf. Complex Analysis, Minneapolis, 1964, pp. 142154, Springer-Verlag, 1965.
[161] M. Kuranishi, Deformations of complex spaces, Sminaire de Mathmtiques Suprieures 39,
Les Presses de lUniversit de Montreal (1969).
[162] K. Lamotke, The topology of complex projective varieties after S. Lefschetz, Topology 20 (1),
1551 (1981).
[163] S. Lang, Abelian varieties, Springer-Verlag, 1983.
[164] S. Lang, Algebra, rev. 3rd ed., Graduate Texts in Mathematics 211, Springer-Verlag, 2002.
[165] H. Lange and Ch. Birkenhake, Complex Abelian varieties, 2nd ed., Grundlehren der Mathematischen Wissenschaften 302 Springer-Verlag, 2004.
[166] H. Lange and Ch. Birkenhake, Complex tori, Progress in Mathematics 177, Birkhuser, 1999.
[167] H. B. Lawson and M. L. Michelsohn, Spin geometry, Princeton Mathematical Series 38, Princeton University Press, 1989.
[168] W. Lawvere and S. Schanuel, Conceptual mathematics: A first introduction to categories, 2nd
ed., Cambridge University Press, 2009.
[169] R. Lazarsfeld, Positivity in Algebraic Geometry I,II, Ergebnisse der Mathematik und ihrer
Grenzgebiete. 3. Folge. A Series of Modern Surveys in Mathematics 49, Springer-Verlag,
2004.
[170] R. Lazarsfeld, A sampling of vector bundle techniques in the study of linear series, in: Lectures on Riemann Surfaces, Trieste 1987, pp. 500559, World Sci. Publ., 1989.
[171] J. M. Lee, Introduction to Topological Manifolds, 2nd ed., Graduate Texts in Mathematics 202,
Springer-Verlag, 2011.
[172] S. Lefschetz, Lanalysis situs et la gomtrie algbrique, Gauthier-Villars, 1950.
[173] J. Le Potier, Annulation de la cohomolgie valeurs dans un fibr vectoriel holomorphe positif
de rang quelconque, Math. Ann. 218 (1), 3553 (1975).

222 | Bibliography
[174] J. Le Potier, Cohomologie de la grassmannienne valeurs dans les puissances extrieurs et
symtriques du fibr universe. Math. Ann. 226 (3), 257270 (1977).
[175] J. Leray, Lhomologie dun espace fibr dont la fibre est connexe, J. Math. Pures Appl. 29(9),
169213 (1950).
[176] J. Leray, Lanneau spectral et lanneau filtr dhomologie dun espace localement compact et
dune application continue, J. Math. Pures Appl. 29(9), 180, 81139 (1950).
[177] D. Lieberman, Intermediate Jacobians, in: Algebraic geometry, Oslo 1970, Proc. Fifth Nordic
Summer-School in Math., pp. 125139, Wolters-Noordhoff, 1972.
[178] I. G. Macdonald, Symmetric functions and Hall polynomials, 2nd ed., with contributions by
A. Zelevinsky, Oxford Mathematical Monographs, Oxford Science Publications, The Clarendon
Press, Oxford University Press, 1995.
[179] S. Mac Lane, Categories for the working mathematician, Second edition, Graduate Texts in
Mathematics 5, Springer-Verlag, 1998.
[180] S. Mac Lane, Homology, Die Grundlehren der mathematischen Wissenschaften 114, Academic
Press, Inc., Springer-Verlag, 1963.
[181] L. Manivel, Symmetric functions, Schubert polynomials and degeneracy loci, SMF/AMS Texts
and Monographs 6, American Mathematical Society, Socit Mathmatique de France, 2001.
[182] M. Maruyama, Moduli of stable sheaves, I, J. Math. Kyoto 17 (1), 91126 (1977).
[183] W. Massey, A basic course in algebraic topology, Graduate Texts in Mathematics 127,
Springer-Verlag, 1991.
[184] W. Massey, Algebraic topology: an introduction, Graduate Texts in Mathematics 56, SpringerVerlag, 1967.
[185] H. Matsumura, Commutative algebra, W. A. Benjamin, Inc., 1970.
[186] J. K. Migliore, Introduction to liaison theory and defeciency moduli, Progress in Mathematics
165, Birkuser, 1998.
[187] J. K. Migliore and U. Nagel, Liaison and related topics: notes from the Torino workshop-school
Liaison and related topics, Rend. Sem. Mat. Univ. Politec. Torino 59 (2), 59126 (2001).
[188] J. Milnor and J. Stasheff, Characteristic classes, Annals of Mathematics Studies 76, Princeton
University Press, 1974.
[189] R. Miranda, Algebraic curves and Riemann surfaces, Graduate Studies in Mathematics 5,
American Mathematical Society, 1995.
[190] B. G. Moishezon, A projectivity criterion of complete algebraic abstract varieties, Izv. Akad.
Nauk SSSR Ser. Mat. 28, 179224 (1964). in Russian.
[191] S. Mori and S. Mukai, Classification of Fano 3-folds with 2 2, Manuscripta Math. 36 (2),
147162 (1981/82).
[192] A. Moroianu, Lectures on Khler geometry, Cambridge University Press, 2007.
[193] D. Mumford, Abelian varieties, Tata Insitute of Foundamental Research Studies in Mathematics 5, Oxford University Press, 1974.
[194] D. Mumford, Geometric invariant theory, Ergebnisse der Mathematik und ihrer Grenzgebiete,
Neue Folge 34, Springer-Verlag, 1965.
[195] D. Mumford, Curves and their Jacobians, The University of Michigan Press, 1975.
[196] D. Mumford, Algebraic geometry. I. Complex projective varieties, Classics in Mathematics,
Springer-Verlag, 1995.
[197] D. Mumford, The red book of varieties and schemes, Lecture Notes in Mathematics 1358,
Springer-Verlag, 1988.
[198] D. Mumford, Varieties defined by quadratic equations, in: Questions in algebraic varieties,
C.I.M.E., Varenna, 1969, pp. 29100, Edizioni Cremonese, 1970.
[199] D. Mumford, J. Fogarty, and F. Kirwan, Geometric invariant theory, 3rd ed., Ergebnisse der
Mathematik und ihrer Grenzgebiete (2) 34, Springer-Verlag, 1994.

Bibliography

223

[200] R. Munkres, Topology, a first course, Prentice-Hall, Inc., 1975.


[201] M. Nagata, Existence theorems for non-projective complete algebraic varieties, Illinois J.
Math. 2, 490498 (1958).
[202] M. Nagata, Imbedding of an abstract variety in a complete variety, J. Math. Kyoto Univ. 2, 110
(1962).
[203] Y. Nakai, A criterion of an ample sheaf on a projective scheme, Amer. J. Math. 85, 1426
(1963).
[204] P. E. Newstead, Introduction to moduli problems and orbit spaces, Tata Institute of Fundamental Research Lectures on Mathematics and Physics 51, Tata Institute of Fundamental Research, Bombay, Narosa Publishing House, 1978.
[205] T. Oda, Convex bodies and algebraic geometry, an introduction to the theory of toric varieties,
Ergebnisse der Mathematik und ihrer Grenzgebiete (3) 15, Springer-Verlag, 1988.
[206] T. Oda, Torus embeddings and applications, Based on jointwork with Katsuya Miyake, Tata
Institute of Fundamental Research Lectures on Mathematics and Physics 57, Springer-Verlag,
1978.
[207] Ch. Okonek, H. Spindler, and M. Schneider, Vector bundles on complex projective spaces,
Progress in Mathematics 3, Birkhuser, 1980.
[208] M. Scott Osborne Basic Homological Algebra, Graduate Texts in Mathematics 196, SpringerVerlag, 2000.
[209] G. Ottaviani, Variet proiettive di codimensione piccola, notes of courses given
in Istituto Nazionale di Alta Matematica F. Severi in Roma, 1994, available at
http://web.math.unifi.it/users/ottavian/.
[210] G. Ottaviani, Rational Homogeneous Varieties, notes of the course SMI Geometria Algebrica,
Cortona 1995, available at http://web.math.unifi.it/users/ottavian/.
[211] S. Parameswaran, Skew-symmetric determinants, Amer. Math. Monthly 61 (2), 116 (1954).
[212] D. Perrin, Algebraic geometry. An introduction, Universitext, Springer-Verlag, 2008.
[213] C. Peskine, L. Szpiro, Liaison des varits algbriques I, Invent. Math. 26 (4), 271302 (1974).
[214] J. Roberts, Chows moving lemma, in: Algebraic Geometry, Oslo 1970, F. Oort (ed.), pp. 89
96, Wolters-Noordhoff, 1972.
[215] J. Rotman, An introduction to algebraic topology, Graduate Texts in Mathematics 119,
Springer-Verlag, 1988.
[216] J. H. Sampson and G. Washnitzer, A Knneth formula for coherent algebraic sheaves, Illinois J.
Math. 3, 389402 (1959).
[217] P. Samuel, Relations dquivalence en gomtrie algebrique, in: Proc. Internat. Congress
Math. 1958, Edinburgh, pp. 470487, Cambridge University Press, 1960.
[218] M. Schlessinger, Functors on Artin rings, Trans. Amer. Math. Soc. 130, 208222 (1968).
[219] C. Segre, Sulle rigate razionali in uno spazio lineare qualunque, Atti R. Accad. Torino 19, 265
282 (19831984).
[220] Algebraic surfaces, Trudy Math. Inst. Stekelov, 75 (1965). English translation: Algebraic surfaces (by the membres of the seminar of I. Shafarevich, Amer. Math. Soc. Translations 75
(1967).
[221] E. Sernesi, Deformations of algebraic schemes, Grundlehren der Mathematischen Wissenschaften 334, Springer-Verlag, 2006.
[222] J. P. Serre, Gomtrie algbrique et gomtrie analytique, Ann. Ist. Fourier 6, 142 (1956).
[223] J. P. Serre, Faisceaux algbriques cohrents, Ann. of Math. 61, 197278 (1955).
[224] J. P. Serre, Un thorme de dualit, Comment. Math. Helv. 29, 926 (1955).
[225] J. P. Serre, Linear Representations of Finite Groups Graduate Texts in Mathematics 42,
Springer-Verlag, 1977.

224 | Bibliography
[226] J. P. Serre, Lie algebras and Lie groups, corrected fifth printing of the 2nd (1992) ed., Lecture
Notes in Mathematics 1500, Springer-Verlag, 2006.
[227] J. P. Serre, Algbre Locale Multiplicit, rdig par P. Gabriel, Lectures Notes in Math. 11,
Springer-Verlag, 1965.
[228] I. Shafarevich, Basic Algebraic Geometry 1,2, 3rd ed., Spinger-Verlag, 2013.
[229] D. L. Shannon, Monoidal transforms of regular local rings, Amer. J. Math. 95, 294320,
(1973).
[230] R. W. Sharpe, Differential Geometry, Cartans generalization of Kleins Erlangen program,
Graduate Texts in Mathematics 166, Springer-Verlag, 1997.
[231] B. Shiffman and A. J. Sommese, Vanishing theorems on complex manifolds, Progress in Mathematics 56, Birkhuser, 1985.
[232] C. L. Siegel, Meromorphe Functionen auf kompacten analytischen Mannigfaltigkeiten, Nach.
Akad. Wiss. Gttingen, 7177 (1955).
[233] K. E. Smith, L. Kahapaa, P. Kekalainen, and W. Traves An Invitation to Algebraic Geometry,
Universitext, Springer-Verlag, 2000.
[234] E. Spanier, Algebraic Topology, corrected reprint, Springer-Verlag, 1981.
[235] T. A. Springer, Linear algebraic groups, 2nd ed., Progress in Mathematics 9, Birkhuser,
1998.
[236] R. P. Stanley, Enumerative combinatorics, vol. 2, with a foreword by Gian-Carlo Rota and appendix 1 by Sergey Fomin, Cambridge Studies in Advanced Mathematics 62, Cambridge University Press, 1999.
[237] N. Steenrod, The topology of the fibre bundles, 2nd ed., Princeton University Press, 1999.
[238] B. Sturmfels, Algorithms in invariant theory, 2nd ed., Texts and Monographs in Symbolic
Computation, Springer-Verlag, 2008.
[239] F. Takemoto, Stable vector bundles on algebraic surfaces, Nagoya Math. J. 47, 2948 (1973).
[240] H. Tango, On morphisms from projective space to the Grassmannian variety (, ),
J. Math. Kyoto Univ. 16 (1), 201207 (1976).
[241] K. Ueno, Classification Theory of algebraic varieties and compact complex spaces, Lecture
Notes in Math. 439, Springer-Verlag, 1975.
[242] J. L. Verdier, Categories derives, Etat 0 Sem. Geom. Alg. 4 1/2. Cohomologie tale, Lecture
Notes in Math. 569, Springer Verlag, 1977.
[243] E. Viehweg, Vanishing theorems, J. Reine Angew. Math. 335, 18 (1982).
[244] E. Viehweg, Quasi-projective moduli for polarized manifolds, Results in Mathematics and
Related Areas (3) 30, Springer-Verlag, 1995.
[245] C. Voisin, Hodge theory and complex algebraic geometry, I, II, Cambridge Studies in Advanced Mathematics 76, 77, Cambridge University Press, 2002, 2003.
[246] R. J. Walker, Algebraic Curves, reprint of the 1950 edition, Springer-Verlag, 1978.
[247] A. Wallace, Algebraic topology: Homology and cohomology, W. A. Benjamin, Inc., 1970.
[248] F. W. Warner, Foundations of differentaible manifolds and Lie Groups, Graduate Texts in Mathematics 94, Springer-Verlag, 1983.
[249] W. C. Waterhouse, Introduction to affine group schemes, Graduate Texts in Mathematics 66,
Springer-Verlag, 1979.
[250] A. Weil, Introduction ltude des varits Khlriennes, Hermann, 1958.
[251] R. Wells, Differential Analysis on Complex Manifolds, Graduate Texts in Mathematics 65,
Springer-Verlag, 1980.
[252] S.-T. Yau, A survey of Calabi-Yau manifolds, Surveys in differential geometry, vol. XIII. Geometry, analysis, and algebraic geometry: forty years of the Journal of Differential Geometry, Surv.
Differ. Geom. 13, 277318, 2009.

Bibliography

225

[253] O. Zariski, Algebraic surfaces, with appendices by S. S. Abhyankar, J. Lipman and D. Mumford, preface to the appendices by Mumford, reprint of the 2nd ed., Springer-Verlag, 1995.
[254] O. Zariski, The problem of minimal models in the theory of algebraic surfaces, Amer. J. Math.
80, 146184 (1958).
[255] O. Zariski, Foundations of a general theory of birational correspondences, Trans. Amer. Math.
Soc. 53, 490542 (1943).
[256] O. Zariski and P. Samuel, Commutative algebra 1,2, Rrprints of the 1958 and 1960 editions,
Graduate Texts in Mathematics 28, 29, Springer-Verlag, 1975.

List of terms
Abelian varieties. Tori, complex - and
Abelian varieties
Adjunction formula.
Albanese varieties.
Algebras.
Algebraic groups.
Almost complex manifolds, holomorphic maps,
holomorphic tangent bundles.
Ample and very ample. Bundles, fibre - or
Divisors.
Anticanonical. Fano varieties
Arithmetically CohenMacaulay or
arithmetically Gorenstein.
Cohen-Macaulay, Gorenstein,
(arithmetically -,-)
Artinian. Noetherian, Artinian

Base point free (b.p.f.) Bundles, fibre -


Beilinsons complex.
Bertinis theorem.
Bezouts theorem.
Bielliptic surfaces. Surfaces, algebraic -
Big. Bundles, fibre - or Divisors
Birational. Varieties, algebraic -, Zariski
topology, regular and rational functions,
morphisms and rational maps
Blowing-up (or -process).
Buchbergers algorithm. Groebner bases
Bundles, fibre -.

CalabiYau manifolds.
Canonical bundle, canonical sheaf.
Cap product. Singular homology and
cohomology
Cartan-Serre theorems.
CastelnuovoEnriques Criterion. Surfaces,
algebraic -
CastelnuovoEnriques theorem. Surfaces,
algebraic -
CastelnuovoDe Franchis theorem.
Surfaces, algebraic -
Categories.
Chern classes.
Chows group. Equivalence, algebraic,
rational, linear -, Chow, NeronSeveri and
Picard groups
Chows theorem.

Class group, divisor -. Equivalence,


algebraic, rational, linear -, Chow,
NeronSeveri and Picard groups
Cliffords Index and Cliffords theorem.
Riemann surfaces (compact -) and
algebraic curves
CohenMacaulay, Gorenstein,
(arithmetically -,-).
Coherent sheaves.
Cohomology, singular -. Singular homology
and cohomology
Cohomology of a complex. Complexes
Complete intersections.
Complete varieties.
Completion.
Complexes.
Cone, tangent -.
Connections.
Correspondences.
Covering projections.
Cremona transformations. Quadratic
transformations, Cremona
transformations
Cross ratio.
Cup product. Singular homology and
cohomology
Curves. Riemann surfaces (compact -) and
algebraic curves
Cusps. Regular rings, smooth points,
singular points
Cycles.
Deformations.
Degeneracy locus of a morphism of vector
bundles. Determinantal varieties
Degree of an algebraic subset.
Depth.
Del Pezzo surfaces. Surfaces, algebraic -
De Rhams theorem.
Derived categories and derived functors.
Determinantal varieties.
Dimension.
Direct and inverse image sheaves.
Discrete valuation rings.
Divisors.
Dolbeaults theorem.

228 | List of the terms


Dominant. Varieties, algebraic -, Zariski
topology, regular and rational functions,
morphisms and rational maps
Dual variety and biduality theorem.
Dualizing sheaf.
Dynkin diagrams. Lie algebras

Effective. Divisors
Elliptic Riemann surfaces, elliptic curves.
Elliptic surfaces. Surfaces, algebraic -
Embedded components. Primary ideals,
primary decompositions, embedded
ideals
Embedding.
Enriques surfaces. Surfaces, algebraic -
Equivalence, algebraic, rational, linear -, Chow,
NeronSeveri and Picard groups.
Euler sequence.
Exact sequences.
Exponential sequence.
Ext, EXT.
Fano varieties.
Fibred product.
Five Lemma.
Flag varieties.
Flat (module, morphism).
Flexes.
Fubini-Study metric.
Functors. Categories
Fundamental group.

G.A.G.A.
GaussBonnetHopf theorem.
General type, of -. Kodaira dimension (or
Kodaira number)
Genus, arithmetic, geometric, real, virtual -.
Geometric invariant theory (G.I.T.).
Globally generated. Bundles, fibre -
Gorenstein. CohenMacaulay, Gorenstein,
(arithmetically -,-)
Grassmannians.
Grauerts semicontinuity theorem.
Groebner bases.
Grothendieck group.
GrothendieckSegre theorem.
Grothendiecks vanishing theorem.
Vanishing theorems
Group algebra.

Hartogs theorem.
Hartshornes conjecture.
HartshorneSerre theorem (correspondence).
Hermitian and Khlerian metrics.
Hilbert Basis theorem.
Hilberts Nullstellensatz.
Hilbert function and Hilbert polynomial.
Hilbert schemes. Moduli spaces
Hilbert syzygy theorem.
Hironakas decomposition of birational maps.
Hirzebruch surfaces. Surfaces, algebraic -
HirzebruchRiemannRoch theorem.
Hodge theory.
Holomorphic. Almost complex manifolds,
holomorphic maps, holomorphic tangent
bundles
Homogeneous bundles.
Homogeneous ideals.
Homogeneous varieties.
Homology, Singular -. Singular homology
and cohomology
Homology of a complex. Complexes.
Horrocks Criterion.
HorrocksMumford bundle.
Horrocks theorem.
Horseshoe lemma.
Hurwitzs theorem. Riemann surfaces
(compact -) and algebraic curves
Hypercohomology of a complex of sheaves.
Hyperelliptic Riemann surfaces. Riemann
surfaces (compact -) and algebraic curves
Hyperelliptic or bielliptic surfaces. Surfaces,
algebraic -
Hyperplane bundles, twisting sheaves.
Injective and projective modules.
Injective and projective resolutions.
Integrally closed.
Intersection of cycles.
Inverse image sheaf. Direct and inverse
image sheaves
Irreducible topological space.
Irregularity.
Jacobians of compact Riemann surfaces.
Jacobians, Weil and Griffiths intermediate -.
Jumping lines and splitting type of a vector
bundle on .

List of the terms


Khler. Hermitian and Khlerian metrics
Kodaira Embedding theorem.
KodairaNakano vanishing theorem.
Vanishing theorems
Kodaira dimension (or Kodaira number).
Koszul complex.
K3 surfaces. Surfaces, algebraic -
Kummer surfaces. Surfaces, algebraic -
Lefschetz decomposition and Hard Lefschetz
theorem.
Lefschetz theorem on (1, 1)-classes.
Lefschetz hyperplane theorem.
Length of a module.
Leray spectral sequence.
Liaison or linkage.
Lie algebras.
Lie groups.
Limits, direct and inverse -.
Linear systems.
Linkage. Liaison or linkage
Local.
Localization, quotient ring, quotient field.
Lroth problem. Unirational, Lroth
problem

Manifolds.
Mapping cone lemma.
Minimal set of generators.
Minimal free resolutions.
Minimal degree.
Modules.
Moduli spaces.
Monoidal transformations.
Morphisms. Varieties, algebraic -, Zariski
topology, regular and rational functions,
morphisms and rational maps
Multiplicity of a curve in a surface at a point.
Multiplicity of intersection. Intersection of
cycles

NakaiMoishezon theorem. Bundles, fibre -


Nef. Bundles, fibre - or Divisors
NeronSeveri group. Equivalence, algebraic,
rational, linear -, Chow, NeronSeveri and
Picard groups
Net.
Node. Regular rings, smooth points,
singular points
Noetherian, Artinian.

229

Noethers formula. Surfaces, algebraic -


Nondegenerate.
Normal.
Normal, projectively -, -normal, linearly normal.
Normal crossing and log complex.
, Property -. Syzygies
Null correlation bundle.
O(). Hyperplane bundles, twisting sheaves
Orbit Lemma, Closed -.
Orientation.
Pencil.
Pfaffian.
Picard groups. Equivalence, algebraic,
rational, linear -, Chow, NeronSeveri and
Picard groups
Plurigenera.
Positive.
Primary ideals, primary decompositions,
embedded ideals.
Principal bundles. Bundles, fibre -
Process, -. Blowing-up (or -process)
Product, Semidirect -.
Proper.
Proper mapping theorem. Remmerts proper
mapping theorem
Projective modules. Injective and projective
modules
Projective resolutions. Injective and
projective resolutions
Pull-back and push-forward of cycles.
Pull-back and push-forward of sheaves.
Direct and inverse image sheaves
Quadratic transformations, Cremona
transformations.
Quotient field. Localization, quotient ring,
quotient field

Rank of finitely generated Abelian groups.


Rational normal curves.
Rational normal scrolls. Scrolls, rational
normal -
Rational functions, rational maps. Varieties,
algebraic -, Zariski topology, regular and
rational functions, morphisms and rational
maps
Rational varieties.
Reduced. Schemes

230 | List of the terms


Regular functions. Varieties, algebraic -,
Zariski topology, regular and rational
functions, morphisms and rational maps
Regular rings, smooth points, singular points.
Regular sequences.
Regularity.
Remmerts proper mapping theorem.
Representations.
Residue field.
Resolutions. Exact sequences.
Riemanns existence theorem.
RiemannRoch theorem.
HirzebruchRiemannRoch theorem
Riemann surfaces (compact -) and algebraic
curves.
Saturation.
Schemes.
Schur functors.
Scrolls.
Scrolls, Rational Normal -.
Segre classes.
Segre embedding.
Semicontinuity theorem. Grauerts
semicontinuity theorem
Serre correspondence.
Serre duality.
Serres theorems A and B. CartanSerre
theorems
Sheafify. Serre correspondence
Sheaves.
Siegel half-space.
Siegels theorem.
Simple bundles.
Singular homology and cohomology.
Singularities. Regular rings, smooth points,
singular points
Smooth. Regular rings, smooth points,
singular points
Snake lemma.
Spaces, analytic -.
Spaces, ringed -.
Spectral sequences.

Spin groups.
Splitting type of a vector bundle. Jumping
lines and splitting type of a vector bundle
on
Stable sheaves.
Star operator.
Stein factorization.
Subcanonical.
Surfaces, algebraic -.
Symmetric polynomials
Syzygies.
Tautological (or universal) bundle.
Tor, TOR.
Torellis theorem.
Tori, complex - and Abelian varieties.
Toric varieties.
Transcendence degree.
Transcendental.
Unirational, Lroth problem.
Universal bundle. Tautological (or universal)
bundle
Vanishing theorems.
Varieties, algebraic -, Zariski topology, regular
and rational functions, morphisms and
rational maps.
Varieties and subvarieties, analytic -.
Veronese embedding.
Web.
Weighted projective spaces.
Weierstrass points.
Weierstrass form of cubic curves.
Weierstrass preparation theorem and
Weierstrass division theorem.
Zariskis main theorem.
Zariski tangent space, differential forms,
tangent bundle, normal bundle.
Zariski topology. Varieties, algebraic -,
Zariski topology, regular and rational
functions, morphisms and rational maps