Documentos de Académico
Documentos de Profesional
Documentos de Cultura
ANALYSIS ONFRACTALS
ALEXANDER TEPLYAEV
UNIVERSITY OF CONNECTICUT
HTTP://WWW.MATH.UCONN.EDU/~ TEPLYAEV/GREGYNOG/
LMS/EPSRC SHORT COURSE
10 - 15 JANUARY 2007
GREGYNOG HALL
IN CONJUNCTION WITH
ANALYSIS ON GRAPHS AND ITS APPLICATIONS
ISAAC NEWTON INSTITUTE FOR MATHEMATICAL SCIENCES
CAMBRIDGE, UK, 8 JANUARY - 29 JUNE 2007
Major general references (books)
[1] M. T. Barlow, Diusions on fractals. Lectures on Probability Theory and
Statistics (Saint-Flour, 1995), 1121, Lecture Notes in Math., 1690, Springer,
Berlin, 1998.
[2] J. Kigami, Analysis on fractals. Cambridge Tracts in Mathematics 143, Cam-
bridge University Press, 2001.
[3] M. L. Lapidus and M. van Frankenhuysen, Fractal geometry, complex dimen-
sions and zeta functions. Geometry and spectra of fractal strings. Springer
Monographs in Mathematics. Springer, New York, 2006.
[4] R. S. Strichartz, Dierential equations on fractals: a tutorial. Princeton
University Press, 2006.
Contents
Major general references (books) 2
Lecture 1
Laplacians on self-similar graphs
and relation to self-similar groups 6
Abbreviated list of references 30
Lecture 2
Laplacians on self-similar fractals
and spectral zeta functions 36
Abbreviated list of references 54
Lecture 3
Kigamis resistance forms on fractals
and relation to quantum graphs 64
Abbreviated list of references 87
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
0
v v
v v v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
v v v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
0
Lecture 1
Laplacians on self-similar graphs
and relation to self-similar groups
R. Rammal and G. Toulouse, Random walks on fractal structures and percolation
clusters. J. Physique Letters 44 (1983), L13L22.
R. Rammal, Spectrum of harmonic excitations on fractals. J. Physique 45 (1984).
S. Alexander, Some properties of the spectrum of the Sierpi nski gasket in a mag-
netic eld. Phys. Rev. B 29 (1984).
E. Domany, S. Alexander, D. Bensimon and L. Kadano, Solutions to the Schrodinger
equation on some fractal lattices. Phys. Rev. B (3) 28 (1984).
Y. Gefen, A. Aharony and B. B. Mandelbrot, Phase transitions on fractals.
I. Quasilinear lattices. II. Sierpi nski gaskets. III. Innitely ramied lattices. J. Phys.
A 16 (19831984).
R. B. Stinchcombe, Fractals, phase transitions and criticality. Fractals in the natural
sciences. Proc. Roy. Soc. London Ser. A 423 (1989), 1733.
J. Bellissard, Renormalization group analysis and quasicrystals, Ideas and methods
in quantum and statistical physics (Oslo, 1988). Cambridge Univ. Press, 1992.
2
1
1
1
z =
3
2
1
2
1
2
1
2
1
2
1
2
1
2
2 1
1
1 1
z =
3
4
2
1
1
1
1
1
1
z =
3
2
1 1
1
1
1
1
z =
5
4
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
v v v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
v v v
v
v v
v
v v
v v v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
0
Let be the probabilistic Laplacian (generator of a simple random walk) on the
Sierpi nski lattice. If z ,=
3
2
,
5
4
,
1
2
, and R(z) = z(4z + 5), then
R(z) () z ()
() = J
R
D
where D
def
= |
3
2
|
m=0
R
m
|
3
4
|
_
and J
R
is the Julia set of R(z).
E
T
0
3
2
5
4
3
4
1
2
3
2
5
4
,
, , ,
,
, , , , ,
There are uncountably many nonisomorphic
Sierpi nski lattices.
Theorem (T). The spectrum of is pure point.
Eigenfunctions with nite support are complete.
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
v v
vv v
v
v v
v
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
0
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
r
_ _ _
_ _ _ _ _ _
`
_
`
_
`
_
`
_
`
_
`
_
Let
(0)
be the Laplacian with zero (Dirichlet) boundary condition at L. Then
the compactly supported eigenfunctions of
(0)
are not complete (eigenvalues in
E is not the whole spectrum).
_
0
0
0
1
g
1g
0
0 0
0
0 0
0
0 0
. . .
. . .
Let L
(0)
be the set of two points adjacent to L and
(0)
be the spectral
measure of
(0)
associated with 1
L
(0)
. Then supp(
(0)
) = J
R
has Lebesgue
measure zero and
d(
(0)
R
1,2
)
d
(0)
(z) =
(8z + 5)(2z + 3)
(2z + 1)(4z + 5)
Fix p, q>0, p+q=1, and dene probabilistic Laplacians
n
on the segments
[0, 3
n
] of Z
+
inductively as a generator of the random walks:
0 1
g g
E '
1 1
0 1 3
g g g g
E E E ' ' '
1 q p p q 1
0 1 3 6 9
g g g g g g g g g g
E E E E E E E E E ' ' ' ' ' ' ' ' '
1 q p p q q p q p p q p q q p p q 1
0 3
n
2(3
n
) 3
n+1
g , , , , , , , , , g g g
E E E ' ' '
1 q p p q 1
and let = lim
n
n
be the corresponding probabilistic Laplacian on Z
+
.
If z ,=1p and R(z)=z(z
2
+3z+2+pq)/pq, then R(z) (
n
)
z (
n+1
)
E
T
0
2
2
1 p
, , ,
Theorem(T). () = J
R
, the Julia set of R(z).
If p=q, then ()=[2, 0], spectrum is a.c.
If p ,= q, then () is a Cantor set of Lebesgue measure zero, spectrum is
singularly continuous.
There are uncountably many random self-similar Laplacians on Z:
For a sequence K = |k
j
|
j=1
, k
j
|0, 1, 2|, let
X
n
=
n
j=1
k
j
3
j
and
n
is a probabilistic Laplacian on [X
n
, X
n
+3
n
]:
X
n
X
n
+3
n1
X
n
+2(3
n1
) X
n
+3
n
g , , , , , , , , , g g g
E E E ' ' '
1 q p p q 1
In the previous example k
j
= 0 for all j.
Theorem (T).
For any sequence K we have () = J
R
. The same is true for the Dirichlet
Laplacian on Z
+
(when k
j
0).
R. Grigorchuk and Z. Sunik, Asymptotic aspects of Schreier graphs and Hanoi
Towers groups, preprint.
,
,
, ,
, ,
,
,
, ,
,
,
, ,
,
,
, ,
, ,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
, ,
,
,
, ,
,
,
, ,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
Sierpi nski 3-graph Sierpi nski 4-graph
(Hanoi Towers-3 group) (standard)
These three polynomials are conjugate:
Sierpi nski 3-graph (Hanoi Towers-3 group): f(x) = x
2
x 3
f(3) = 3, f
t
(3) = 5
Sierpi nski 4-graph, adjacency matrix Laplacian: P() = 5
2
P(0) = 0, P
t
(0) = 5
Sierpi nski 4-graph, probabilistic Laplacian: R(z) = 4z
2
+ 5z
R(0) = 0, R
t
(0) = 5
,
, ,
,
,
, ,
,
,
, ,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
, ,
,
,
, ,
,
,
, ,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,
,
, ,
,
,
,
,
,
,
,
, ,
,
,
,
,
,
,
,
, ,
,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
, ,
,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
Theorem. Eigenvalues and eigenfunctions on the
Sierpi nski 3-graphs and Sierpi nski 4-graphs are in
one-to-one correspondence, with the exception of the
eigenvalue z =
3
2
for the 4-graphs.
,
,
4z
2
+ 5z
,
,
,
,,
,
,
,
4
3
z
2
+
8
3
z
,
,
, ,
, ,
2z
2
+ 4z
, ,
,
, ,
,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, , ,
,
, , ,
,
, ,
,
,
, ,
,
,
, , ,
,
, ,
,
,
, ,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
,
,
,
,,
,
Sierpi nski 3-graph Sierpi nski 4-graph
(Hanoi Towers-3 group) (standard)
R(z) = 2z
2
+ 4z R(z) =
4
3
z
2
+
8
3
z
Let H and H
0
be Hilbert spaces, and U : H
0
H be an isometry.
Denition. We call an operator H spectrally similar to an operator H
0
with
functions
0
and
1
if
U
(H z)
1
U = (
0
(z)H
0
1
(z))
1
In particular, if
0
(z) ,= 0 and R(z) =
1
(z)/
0
(z), then
U
(H z)
1
U =
1
0
(z)
(H R(z))
1
.
If H =
_
S
X
X Q
_
then
S zI
0
X(QzI
1
)
1
X =
0
(z)H
0
1
(z)I
0
Theorem (Malozemov, Teplyaev). If is the graph Laplacian on a self-
similar symmetric innite graph, then
J
R
(
) J
R
D
where D
j=1, 2, 3
F
j
(V
n1
) =
_
k
3
n
_
3
n
k=0
V
0
=I : _ _
c
q
V
1
: _ _ _ _
c
q
V
2
: _ _ _ _ _ _ _ _ _ _
Denition. The discrete Dirichlet (energy) form on V
n
is
E
n
(f) =
x,yV
n
yx
(f(y)f(x))
2
and the Dirichlet (energy) form on I is E(f) = lim
n
3
n
E
n
(f) =
_
1
0
|f
t
(x)|
2
dx
Denition. A function h is harmonic if it minimizes the energy given the
boundary values.
Proposition. 3E
n+1
(f) E
n
(f) and 3E
n+1
(h) = E
n
(h) = 3
n
E(h)
for a harmonic h.
Proposition. The Dirichlet (energy) form on I is self-similar in the sense that
E(f) = 3
j = 1, 2, 3
E(fF
j
)
Denition. The discrete Laplacians on V
n
are
n
f(x) =
1
2
yV
n
yx
f(y)f(x), xV
n
\V
0
and the Laplacian on I is f(x) = lim
n
9
n
n
f(x) = f
tt
(x)
GaussGreen (integration by parts) formula:
E(f) =
_
1
0
ffdx +ff
t
1
0
Spectral asymptotics: Let () be the eigenvalue counting function of
the Dirichlet or Neumann Laplacian :
() = #|j :
j
< |.
Then
lim
()
d
s
/2
=
1
where d
s
= 1 is the spectral dimension.
Denition. The spectral zeta function is
(s) =
j
,=0
_
j
_
s/2
Its poles are the complex spectral dimensions.
Let R(z) be a polynomial of degree N such that its Julia set J
R
(, 0],
R(0) = 0 and c = R
t
(0) > 1.
Denition. The zeta function of R(z) for Re(s) > d
R
=
2 log N
log c
is
z
0
R
(s) = lim
n
zR
n
|z
0
|
(c
n
z)
s/2
=
s/2
j
Theorem.
z
0
R
(s) =
f
1
(s)
1Nc
s/2
+f
z
0
2
(s), where f
1
(s) and f
z
0
2
(s) are ana-
lytic for Re(s)>0. If J
R
is totally disconnected, then this meromorphic continuation
extends to Re(s)>, where >0.
In the case of polynomials this theorem has been improved by Grabner et al.
d
R
the poles of
z
0
R
_
2 log N+4in
log c
: nZ
_
0
E
d
R
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
Theorem.
(s) =
0
R
(s) where R(z) = z(4z
2
+12z+9).
The Riemann zeta function (s) satises (s) =
s
0
R
(s) The only complex
spectral dimension is the pole at s = 1.
A sketch of the proof: If z,=
1
2
,
3
2
, then
R(z) (
n
) z (
n+1
)
and so
(s) =
0
R
(s) since the eigenvalues
j
of are limits of the eigenvalues
of 9
n
n
.
Also
j
=
2
j
2
and so
(s) =
j=1
_
2
j
2
_
s/2
=
s
(s)
where (s) is the Riemann zeta function. Q.E.D.
(s) =
s
lim
n
zR
n
|0|
z,=0
_
9
n
z
_
s/2
Denition.
is Laplacian if
E(f) =
_
1
0
|f
t
(x)|
2
dx=
_
1
0
f
fd +ff
t
1
0
.
Denition. A probability measure is self-similar with weights m
1
, m
2
, m
3
if =
j=1, 2, 3
m
j
F
j
.
Proposition.
f(x)=
f
tt
= lim
n
_
1+
2
pq
_
n
n
f(x).
n
f(
k
3
n
)=
_
pf(
k1
3
n
) +qf(
k+1
3
n
) f(
k
3
n
)
qf(
k1
3
n
) +pf(
k+1
3
n
) f(
k
3
n
)
where m
1
=m
3
, p=
m
2
m
1
+m
2
, q=
m
1
m
1
+m
2
, and
_ _ _ _
E E E ' ' '
1
m
1
m
2
m
3
q p p q 1
_ _ _ _ _ _ _ _ _ _
E E E E E E E E E ' ' ' ' ' ' ' ' '
1 q p p q q p q p p q p q q p p q 1
Spectral asymptotics: If () is the eigenvalue counting function of the Dirichlet
or Neumann Laplacian
, then
0 < liminf
()
d
s
/2
limsup
()
d
s
/2
<
where the spectral dimension is
d
s
=
log 9
log(1+
2
pq
)
1.
All the inequalities are strict if and only if p ,= q.
Proposition. R(z) (
n
) z (
n+1
)
where z,=1p and R(z)=z(z
2
+3z+2+pq)/pq.
Note that R
t
(0)=1 +
2
pq
, and d
s
=d
R
.
Theorem.
(s)=
0
R
(s)
Three contractions F
1
, F
2
, F
3
: R
2
R
2
,
F
j
(x) =
1
2
(x+p
j
), with xed points p
1
, p
2
, p
3
.
p
1
p
2
p
3
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
.
. .
The Sierpi nski gasket is a unique compact set S such that
S =
j=1, 2, 3
F
j
(S)
Denition. The boundary of S is
S = V
0
= |p
1
, p
2
, p
3
|
and discrete approximations to S are
V
n
=
j=1, 2, 3
F
j
(V
n1
)
V
0
:
y y
y
T
T
T
T
T
T
T
T
T
T
V
1
:
x x
xx x
x
x x
x
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
V
2
:
w w
ww w
w
w w
w
w w
ww w
w
w w
w
w w
ww w
w
w w
w
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
Denition. The discrete Dirichlet (energy) form on V
n
is
E
n
(f) =
x,yV
n
yx
(f(y)f(x))
2
and the Dirichlet (energy) form on S is
E(f) = lim
n
_
5
3
_
n
E
n
(f)
Denition. A function h is harmonic if it minimizes the energy given the
boundary values.
Proposition.
5
3
E
n+1
(f) E
n
(f)
5
3
E
n+1
(h)=E
n
(h)=
_
5
3
_
n
E(h) for a harmonic h.
Theorem (Kigami). E is a local regular Dirichlet form on S which is self-similar
in the sense that
E(f) =
5
3
j = 1, 2, 3
E(fF
j
)
Denition. The discrete Laplacians on V
n
are
n
f(x) =
1
4
yV
n
yx
f(y)f(x), xV
n
\V
0
and the Laplacian on S is
f(x) = lim
n
5
n
n
f(x)
if this limit exists and
f is continuous.
GaussGreen (integration by parts) formula:
E(f) =
_
S
f
fd +
pS
f(p)
n
f(p)
where is the normalized Hausdor measure, which is self-similar with weights
1
3
,
1
3
,
1
3
:
=
1
3
j = 1, 2, 3
F
j
.
Spectral asymptotics: If () is the eigenvalue counting function of the Dirichlet
or Neumann Laplacian
, then
0 < liminf
()
d
s
/2
< limsup
()
d
s
/2
<
where the spectral dimension is
1 < d
s
=
log 9
log 5
< 2.
Proposition. R(z) (
n
) z (
n+1
) where z,=
1
2
,
3
4
,
5
4
and R(z) = z(5 + 4z).
Theorem (Fukushima, Shima). Every eigenvalue of
has a form
=5
m
lim
n
5
n
R
n
(z
0
)
where R
n
(z
0
) is a preimage of z
0
=
3
4
,
5
4
under the n-th iteration power
of the polynomial R(z). The multiplicity of such an eigenvalue is C
1
3
m
+C
2
.
Theorem. Zeta function of the Laplacian on the Sierpi nski gasket is
(s) =
1
2
3
4
R
(s)
_
1
5
s/2
3
+
3
5
s/2
1
_
+
1
2
5
4
R
(s)
_
35
s/2
5
s/2
3
5
s/2
5
s/2
1
_
d
s
d
R
=
log 9
log 5
=
log 4
log 5
T
0
d
R
1 d
s
_
_
_
_
_
_
_
g
g
g
g
g
g
g
_
_
_
_
_
_
_
E
1
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
Denition. If L is a fractal string, that is, a disjoint collection of intervals of
lengths l
j
, then its geometric zeta function is
L
(s) =
l
s
j
.
Theorem (Lapidus). If A=
d
2
dx
2
is a Neumann or Dirichlet Laplacian on L,
then
A
(s) =
s
(s)
L
(s).
Example: Cantor self-similar fractal string.
If L is the complement of the middle third Cantor set in [0, 1], then the complex
spectral dimensions are 1 and |
log 2+2in
log 3
: nZ|,
L
(s) =
1
123
s
,
A
(s) = (s)
s
123
s
E
T
0
log 2
log 3
1
_
_
_
_
_
_
_
_
Denition. A post critically nite (p.c.f.) self-similar set F is a compact con-
nected metric space with a nite boundary F F and contractive injections
i
: F F such that
F = (F) =
k
_
i=1
i
(F)
and
v
(F)
w
(F)
v
(F)
w
(F),
for any two dierent words v and w of the same length. Here for a nite word
w |1, . . . , k|
m
we dene
w
=
w
1
. . .
w
m
.
We assume that F is a minimal such subset of F. We call
w
(F) an m-cell.
The p.c.f. assumption is that every boundary point is contained
in a single 1-cell.
Theorem (Kigami, Lapidus). The spectral dimension of the Laplacian
is
the unique solution of the equation
k
i=1
(r
i
i
)
d
s
/2
= 1
Conjecture. On every p.c.f. fractal F there exists a local regular Dirichlet form E
which gives positive capacity to the boundary points and is self-similar in the sense
that
E(f) =
k
i=1
i
E(f
i
)
for a set of positive renement weights = |
i
|
k
i=1
.
Denition. The group G of acts on a nitely ramied fractal F if each g G is
a homeomorphism of F such that g(V
n
) = V
n
for all n 0.
Proposition. Suppose a group G of acts on a self-similar nitely ramied fractal
F and G restricted to V
0
is the whole permutation group of V
0
. Then there exists
a unique, up to a constant, G-invariant self-similar resistance form E with equal
energy renormalization weights
i
and
E
0
(f, f) =
x,yV
0
_
f(x) f(y)
_
2
.
Moreover, for any G-invariant self-similar measure the Laplacian
has the
spectral self-similarity property (a.k.a. spectral decimation).
Abbreviated list of references
[1] B. Adams, S.A. Smith, R. Strichartz and A. Teplyaev, The spectrum of
the Laplacian on the pentagasket. Fractals in Graz 2001, Trends Math.,
Birkhauser (2003).
[2] M. T. Barlow, Diusions on fractals. Lectures on Probability Theory and
Statistics (Saint-Flour, 1995), 1121, Lecture Notes in Math., 1690, Springer,
Berlin, 1998.
[3] M. T. Barlow and R. F. Bass, Brownian motion and harmonic analysis on
Sierpinski carpets. Canad. J. Math., 51 (1999), 673744.
[4] M. T. Barlow and R. F. Bass, Random walks on graphical Sierpinski carpets.
Random walks and discrete potential theory (Cortona, 1997), 2655, Sympos.
Math., XXXIX, Cambridge Univ. Press, Cambridge, 1999.
[5] M. T. Barlow and R. F. Bass, Stability of parabolic Harnack inequalities.
Trans. Amer. Math. Soc., 356 (2004), 15011533.
[6] M. T. Barlow, R. F. Bass and T. Kumagai, Stability of parabolic Harnack
inequalities on metric measure spaces. J. Math. Soc. Japan 58 (2006) 485-
519.
[7] M. T. Barlow and B. M. Hambly, Transition density estimates for Brownian
motion on scale irregular Sierpinski gaskets. Ann. Inst. H. Poincare Probab.
Statist., 33 (1997), 531557.
[8] M. T. Barlow and J. Kigami, Localized eigenfunctions of the Laplacian on
p.c.f. self-similar sets. J. London Math. Soc., 56 (1997), 320332.
[9] M.T. Barlow and E.A. Perkins, Brownian motion on the Sierpinski gasket.
Probab. Theory Related Fields 79 (1988), 543623.
[10] O. Ben-Bassat, R. S. Strichartz and A. Teplyaev, What is not in the domain
of the Laplacian on Sierpinski gasket type fractals. J. Funct. Anal. 166
(1999), 197217.
[11] N. Ben-Gal, A. Shaw-Krauss, R. S. Strichartz, C. Young, Calculus on the
Sierpinski gasket. II. Point singularities, eigenfunctions, and normal
derivatives of the heat kernel. Trans. Amer. Math. Soc. 358 (2006), 3883
3936.
[12] E.J. Bird, S.-M. Ngai and A. Teplyaev, Fractal Laplacians on the Unit In-
terval, Ann. Sci. Math. Quebec 27 (2003), 135168.
[13] K. Coletta, K. Dias, R. S. Strichartz, Numerical analysis on the Sierpinski
gasket, with applications to Schrodinger equations, wave equation, and
Gibbs phenomenon. Fractals 12 (2004), 413449.
[14] K. Dalrymple, R. S. Strichartz and J. P. Vinson, Fractal dierential equations
on the Sierpinski gasket. J. Fourier Anal. Appl., 5 (1999), 203284.
[15] M. Denker and S. Koch, Hausdor dimension for Martin metrics. Algebraic
and topological dynamics, 163170, Contemp. Math., 385, Amer. Math. Soc.,
2005.
[16] M. Denker and H. Sato, Reections on harmonic analysis of the Sierpi nski
gasket. Math. Nachr., 241 (2002), 3255.
[17] G. Derfel, P. Grabner and F. Vogl, The zeta function of the Laplacian on
certain fractals, preprint (2005).
[18] M. Fukushima, Dirichlet forms, diusion processes and spectral dimen-
sions for nested fractals. Ideas and methods in Mathematical Analysis,
Stochastics, and applications (Oslo, 1988), 151161, Cambridge Univ. Press,
Cambridge, 1992.
[19] M. Fukushima and T. Shima, On a spectral analysis for the Sierpi nski
gasket. Potential Analysis 1 (1992), 1-35.
[20] M. Gibbons, A. Raj and R. S. Strichartz, The nite element method on the
Sierpinski gasket. Constr. Approx. 17 (2001), 561588.
[21] P. Grabner and W. Woess (editors) Fractals in Graz 2001. Analysis
dynamicsgeometrystochastics. Proceedings of the conference held at
Graz University of Technology, Graz, June 2001. Edited by . Trends in Math-
ematics. Birkhauser Verlag, Basel, 2003.
[22] B. M. Hambly, Heat kernels and spectral asymptotics for some random
Sierpinski gaskets. Fractal geometry and stochastics, II (Greifswald/Koserow,
1998), 239267, Progr. Probab., 46, Birkhauser, Basel, 2000.
[23] B. M. Hambly, On the asymptotics of the eigenvalue counting function for
random recursive Sierpinski gaskets. Probab. Theory Related Fields, 117
(2000), 221247.
[24] W. Hansen and M. Zahle, Restricting isotropic -stable Levy processes
from R
n
to fractal sets. Forum Math. 18, 171191.
[25] P. E. Herman, R. Peirone, R. S. Strichartz, p-energy and p-harmonic func-
tions on Sierpinski gasket type fractals. Potential Anal. 20 (2004), 125148.
[26] R. G. Hohlfeld and N. Cohen, SelfSimilarity and the Geometric Require-
ments for Frequency Independence in Antennae. Fractals 7 (1999), 7984.
[27] J. Kigami, A harmonic calculus on the Sierpi nski spaces. Japan J. Appl.
Math. 6 (1989), 259290.
[28] J. Kigami, Harmonic calculus on p.c.f. selfsimilar sets. Trans. Amer.
Math. Soc. 335 (1993), 721755.
[29] J. Kigami, Analysis on fractals. Cambridge Tracts in Mathematics 143,
Cambridge University Press, 2001.
[30] J. Kigami and M. L. Lapidus, Weyls problem for the spectral distribution of
Laplacians on p.c.f. self-similar fractals. Comm. Math. Phys. 158 (1993),
93125.
[31] J. Kigami and M. L. Lapidus, Selfsimilarity of volume measures for Lapla-
cians on p.c.f. selfsimilar fractals. Comm. Math. Phys. 217 (2001), 165
180.
[32] S. Kozlov, Harmonization and homogenization on fractals. Comm. Math.
Phys. 153 (1993), 339357.
[33] S. Kusuoka, Lecture on diusion process on nested fractals. Lecture Notes
in Math. 1567 3998, Springer-Verlag, Berlin, 1993.
[34] S. Kusuoka and X. Y. Zhou, Dirichlet forms on fractals: Poincare constant
and resistance. Probab. Theory Related Fields 93 (1992), 169196.
[35] M. L. Lapidus, Spectral and fractal geometry: from the Weyl-Berry conjec-
ture for the vibrations of fractal drums to the Riemann zeta-function. Dif-
ferential equations and mathematical physics (Birmingham, AL, 1990), 151
181, Math. Sci. Engrg., 186, Academic Press, Boston, MA, 1992.
[36] M. L. Lapidus and M. van Frankenhuysen, Fractal Geometry and Number
Theory. Complex Dimensions of Fractal Strings and Zeros of Zeta Func-
tions. Birkhauser, Boston, 2000.
[37] M. L. Lapidus and M. van Frankenhuysen, Fractality, self-similarity and
complex dimensions. Fractal Geometry and Applications: A Jubilee of Benoit
Mandelbrot, Part 1. Proceedings of Symposia in Pure Mathematics 72, Amer.
Math. Soc., (2004), 349372.
[38] M. L. Lapidus and M. van Frankenhuysen, Fractal geometry, complex dimen-
sions and zeta functions. Geometry and spectra of fractal strings. Springer
Monographs in Mathematics. Springer, New York, 2006.
[39] M. L. Lapidus and H. Maier, Hypoth`ese de Riemann, cordes fractales vi-
brantes et conjecture de Weyl-Berry modiee. Acad. Sci. Paris Ser. I Math.
313, (1991), 1924.
[40] M. L. Lapidus and H. Maier, The Riemann hypothesis and inverse spectral
problems for fractal strings. J. London Math. Soc. (2) 52, (1995), 1534.
[41] M. L. Lapidus and C. Pomerance, Fonction zeta de Riemann et conjecture
de Weyl-Berry pour les tambours fractals. C. R. Acad. Sci. Paris Ser. I
Math. 310, (1990), 343348.
[42] M. L. Lapidus and C. Pomerance, The Riemann zeta-function and the one-
dimensional Weyl-Berry conjecture for fractal drums. Proc. London Math.
Soc. (3) 66, (1993), 4169.
[43] T. Lindstrm, Brownian motion on nested fractals. Mem. Amer. Math. Soc.
420, 1989.
[44] R. S. Strichartz, Piecewise linear wavelets on Sierpinski gasket type frac-
tals. J. Four. Anal. Appl., 3 (1997), 387-416.
[45] R. S. Strichartz, Some properties of Laplacians on fractals. J. Funct. Anal.,
164 (1999), 181-208.
[46] R. S. Strichartz, Analysis on fractals. Notices AMS, 46 (1999), 11991208.
[47] R. S. Strichartz, Isoperimetric estimates on Sierpinski gasket type fractals.
Trans. Amer. Math. Soc. 351 (1999), 17051752.
[48] R. S. Strichartz, Taylor approximations on Sierpinski type fractals. J.
Funct. Anal. 174 (2000), 76127.
[49] R. S. Strichartz, Evaluating integrals using self-similarity. Amer. Math.
Monthly 107 (2000), 316326.
[50] R. S. Strichartz, Function spaces on fractals. J. Funct. Anal. 198 (2003),
4383.
[51] R. S. Strichartz, Fractafolds based on the Sierpinski and their spectra.
Trams. AMS 355 (2003), 40194043.
[52] R. S. Strichartz, Solvability for dierential equations on fractals. J. Anal.
Math. 96 (2005), 247267.
[53] R. S. Strichartz, Laplacians on fractals with spectral gaps have nicer
Fourier series. Math. Res. Lett. 12 (2005), 269274.
[54] R. S. Strichartz, Analysis on products of fractals. Trans. Amer. Math. Soc.
357 (2005), 571615.
[55] R. S. Strichartz, Dierential equations on fractals: a tutorial. Princeton
University Press, 2006.
[56] R. S. Strichartz and M. Usher, Splines on fractals. Math. Proc. Cambridge
Philos. Soc. 129 (2000), 331360.
[57] R. S. Strichartz, C. Wong, The p-Laplacian on the Sierpinski gasket. Non-
linearity 17 (2004), 595616.
[58] A. Teplyaev, Spectral Analysis on Innite Sierpi nski Gaskets, J. Funct.
Anal., 159 (1998), 537-567.
[59] A. Teplyaev, Spectral zeta functions of fractals and the complex dynamics
of polynomials, to appear in the Transactions of the American Mathematical
Society.
[60] M. Zahle, Harmonic calculus on fractalsa measure geometric approach.
II. Trans. Amer. Math. Soc. 357 (2005), 34073423.
Lecture 3
Kigamis resistance forms on fractals
and relation to quantum graphs
Denition. A compact connected metric space F is called a nitely ramied
self-similar set if there are injective contraction maps
1
, ...,
m
: F F
and a nite set V
0
F such that
F =
m
_
i=1
i
(F) = (F)
and
F
w
F
w
t = V
w
V
w
t
for any two distinct words w, w
t
W
n
= |1, ..., m|
n
, where F
w
=
w
(F),
V
w
=
w
(V
0
) and
w
=
w
1
...
w
n
.
The vertices of generation n are dened by V
n
= (V
n1
) =
n
(V
0
).
The fractal F can be uniquely reconstructed from its combinatorial skeleton or
ancestor: |F = V
0
, V
1
, |
V
0
| [Kigami, 1993, Appendix A].
A symmetric vanishing on the diagonal function c
0
: V
2
0
R
+
(set of conduc-
tances) denes a discrete Dirichlet form
E
0
(f) =
x,yV
0
(f(y) f(x))
2
c
0
(x, y).
Its renement by is
E
1
(f) =
(E
0
)(f) =
k
i=1
i
E
0
(f
i
).
and the trace map is
Tr(E
1
)(f) = inf |E
1
(g)|g : V
1
R, g|
V
0
= f|.
Theorem (Kigami). For given a set of positive renement weights = |
i
|
k
i=1
self-similar local regular Dirichlet forms E which gives positive capacity to the
boundary points are in one-to-one correspondence with the xed points E
0
of the
renormalization map
= Tr
.
Denition. A resistance form E is self-similar if
E(f, f) =
m
i=1
i
E(f
i
, f
i
).
Conjecture. Any nitely ramied self-similar set has a self-similar resistance form.
Any p.c.f. self-similar set has a regular self-similar resistance form.
Thus we are looking for nonlinear eigenvectors E
0
D P
(E
0
) = E
0
where D is the cone of Dirichlet forms on V
0
with and P is the cone of nonnegative
quadratic forms. Its interior P
: D D, P P, P
.
(2)
is continuous on D P
(3)
(E) =
(E +F)
(E) +
(F)
Hilberts projective metric (a pseudo distance on P
) is
h(E/F) = ln
M(E/F)
m(E/F)
.
where E,FP
, h)
is a complete metric space.
(3) The h- and the ] ]-topology coincide on H P
.
(4) h(E, F) = 0 if and only if E = F.
(5) lim
FP
h(E, F) = +
(6)
is h-nonexpansive on P
-invariant.
Let H = H D P
and q
: H R
+
,
q
(E) = h(
(E), E).
Proposition.
(1)
|
H
vanishes only at F.
(2)
vanishes on a connected
set which accumulates at P.
(3) When a
-eigenvector in B
3r
(E) H.
Proposition. Let |
n
| be such that
n
converges to in (C(H), ] ]
). If
q = h((), ) : H R
+
vanishes only at a single point, then there exists an
m N such that
n
has a unique eigenvector in H, for n m.
Denition. A collection of renement weights is admissible if and only if
(E
0
) = E
0
has a solution E
0
D P
.
Proposition. The set of admissible weights is open.
Theorem. (Hambly, Metz, T.) Let
n
,
(0, ]
k
and
has a
unique eigenvector in H. Then there exist nite admissible renement weights.
This result can be summarized as follows: If, by collapsing a subset of cells of F,
one can obtain a structure which has admissible weights, then F also has admissible
nite weights.
Proposition. If #V
0
= 3 then admissible weights exist.
Denition. The group G of acts on a nitely ramied fractal F if each g G is
a homeomorphism of F such that g(V
n
) = V
n
for all n 0.
Proposition. Suppose a group G of acts on a self-similar nitely ramied fractal
F and G restricted to V
0
is the whole permutation group of V
0
. Then there exists
a unique, up to a constant, G-invariant self-similar resistance form E with equal
energy renormalization weights
i
and E
0
(f, f) =
x,yV
0
_
f(x) f(y)
_
2
.
Theorem (Hambly, Metz, T.) Suppose a self-similar nitely ramied fractal
F has connected interior and a group G acts on F such that its action on V
0
is
transitive. Then there exists a G-invariant self-similar resistance form E on F.
Theorem (Hambly, Metz, T.) Suppose a self-similar nitely ramied fractal F
has connected interior and a symmetric boundary. Then there exists a G-invariant
self-similar resistance form E on F.
Examples.
Generalized non-symmetric Sierpi nski gaskets in R
2
:
1
2
+
1
3
>
1
1
1
1
+
1
2
>
1
3
1
1
+
1
3
>
1
2
Cut Sierpi nski gasket:
1
+
2
= 1
3
+
2
= 1
Unit interval:
1
+
2
= 1
Vicsek set:
1
+
3
+
5
= 1
2
+
4
+
5
= 1
A generalized Vicsek set
A generalized Sierpi nski gasket
GRAPH-DIRECTED FRACTALS
v
1
v
2
v
3
v
4
v
5
v
6
v
7
v
8
v
9
v
10
v
11
v
12
The house fractal.
Denition. A pair (E, DomE) is a resistance form on a countable set V
if
DomE is a linear subspace of (V
V
= v.
For any p, q V
.
Markov property: for any u DomE we have that E( u, u) E(u, u),
where
u(p) =
_
_
1 if u(p) 1,
u(p) if 0 < u(p) < 1,
0 if u(p) 1.
For any nite subset U V
U
= f|
and is called the trace of E on U.
If U
1
U
2
then E
U
1
is the trace of E
U
2
on U
1
.
Theorem (Kigami). Suppose that V
n
are nite subsets of V
and that
n=0
V
n
is R-dense in V
. Then
E(f, f) = lim
n
E
V
n
(f, f)
for any f DomE, where the limit is non-decreasing.
Theorem (Kigami). Suppose that V
n
are nite sets, and the nite dimensional
resistance forms E
V
n
on V
n
are compatible: each E
V
n
is the trace of E
V
n+1
on V
n
.
Then there exists a resistance form E on V
n=0
V
n
such that
E(f, f) = lim
n
E
V
n
(f, f)
for any f DomE, and the limit is non-decreasing.
Denition. A nitely ramied fractal F is a compact metric space with
a cell structure F = |F
|
A
and a boundary (vertex) structure
V = |V
|
A
such that the following conditions hold.
A is a countable index set;
each F
is a nite subset of F
k
j=1
F
j
then V
k
j=1
V
j
;
there exists a ltration |A
n
|
n=0
such that
(1) A
n
are nite subsets of A, A
0
= |0|, F
0
= F;
(2) A
n
A
m
= if n ,= m;
(3) for any A
n
there are
1
, ...,
k
A
n+1
such that F
k
j=1
F
j
;
F
= V
n1
F
n
= |x|.
If these conditions are satised, then
(F, F, V) = (F, |F
|
A
, |V
|
A
)
is called a nitely ramied cell structure.
Denition. A function is harmonic if it minimizes the energy for the given set of
boundary values. A function is n-harmonic if it minimizes the energy for the given
set of values on V
n
.
Theorem. Suppose that all n-harmonic functions are continuous. Then any con-
tinuous function is R-continuous, and any R-Cauchy sequence converges in the
topology of F. Also, there is a continuous injective map : F which is the
identity on V
.
Then we can (and will) consider as a subset of F. Then is the R-closure of
V
t F
, then
M
,
t : (V
) (V
t)
is the linear map which is dened as follows. If f
is a function on V
then let h
f
on V
. Then we
dene
M
,
tf
= h
f
t
.
Proposition. If F
j
then D
,
j
D
j
M
,
j
and
(F
) = Tr M
where M
= M
0,
and D
on V
.
We denote Z
=
M
(F
)
if (F
if (F
) ,= 0, A
n
and x F
\V
. Note that
Tr Z
n
(x) = 1 by denition.
Theorem. For -almost all x there is a limit Z(x) = lim
n
Z
n
(x).
Proof. One can see, following Kusuokas idea, that Z
n
is a bounded -martingale.
fd,
for any function g DomE, vanishing on the boundary V
0
, and any function
f Dom
, where
,
and such functions are dense in Dom
f = Tr (ZD
2
f)
where D
2
f is the matrix of the second derivatives of f.
Conjecture. On the Sierpi nski gasket, if f Dom
x,yV
n
c
n,x,y
E
Q
x,y
(f, f)
where
E
Q
x,y
(f, f) =
_
1
0
_
d
dt
f
_
x(1 t) +ty
_
_
2
dt
is the integral of the square of the derivative
d
dt
f
_
x(1 t) +ty
_
= {f
_
x(1 t) +ty
_
, y x)
of f along the straight line segment connecting x and y. Thus E
Q
x,y
(f, f) is the
usual one dimensional energy of a function on a straight line segment.
If f is linear then E
Q
x,y
(f, f) =
_
f(x) f(y)
_
2
. Therefore if f is piecewise
harmonic then E
Q
n
(f, f) = E
n
(f, f) for all large enough n.
Therefore for any C
1
(R
m
)-function we have
lim
n
E
Q
n
(f, f) = E(f, f)
It is easy to see that if g is a C
1
(R
m
)-function vanishing on V
0
and f is a
C
2
(R
m
)-function then
E
Q
n
(f, g) =
x,yV
n
c
n,x,y
_
1
0
g
_
x(1 t) +ty
_
_
d
2
dt
2
f
_
x(1 t) +ty
_
_
dt
because after integration by parts all the boundary terms are canceled. Then if
A
n
then
x,yV
c
n,x,y
d
2
dt
2
f
_
x(1 t) +ty
_
=
x,yV
c
n,x,y
m
i,j=1
D
2
ij
f
_
x(1 t) +ty
_
(y
i
x
i
)(y
j
x
j
) =
Tr
_
M
_
D
2
f(x
) +R
n
(x, y, t, f, , x
)
_
_
where x
and
lim
n
|R
n
(x, y, t, f, , x
)| = 0
uniformly.
Let H
x
be the space of harmonic functions on F that vanishes at x.
Denition. If h H
x
then the intrinsic derivative
df
dh
(x) R exists if
f(y) = f(x) +h(y)
df
dh
(x) +o
h(y)
yx
.
The intrinsic gradient Grad
x
f H
x
exists if for any non constant h H
x
f(y) = f(x) + Grad
x
(y) +o
h(y)
yx
.
Theorem (Pelander, T). Let be a self-similar measure on a p.c.f. s-s set with
weights
j
. Let
+
and
m
j=1
j
log(r
j
j
).
If
+
> then
df
dh
(x) exists for any f Dom