Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Guest Editors: Mowaffaq Hajja, Peter S. Bullen, Janusz Matkowski, Edward Neuman,
and Slavko Simic
Means and Their Inequalities
Means and Their Inequalities
International Journal of Mathematics and
Mathematical Sciences
Means and Their Inequalities
Guest Editors: Mowafaq Hajja, Peter S. Bullen,
Janusz Matkowski, Edward Neuman, and Slavko Simic
Copyright 2013 Hindawi Publishing Corporation. All rights reserved.
Tis is a special issue published in International Journal of Mathematics and Mathematical Sciences. All articles are open access articles
distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any
medium, provided the original work is properly cited.
Editorial Board
Te editorial board of the journal is organized into sections that correspond to
the subject areas covered by the journal.
Algebra
A. Ballester-Bolinches, Spain
P. Basarab-Horwath, Sweden
Howard E. Bell, Canada
Martin J. Bohner, USA
Tomasz Brzezinski, UK
Stefaan Caenepeel, Belgium
Ral E. Curto, USA
David E. Dobbs, USA
Dalibor Froncek, USA
Heinz P. Gumm, Germany
P. Haukkanen, Finland
P. E. Jorgensen, USA
V. R. Khalilov, Russia
Aloys Krieg, Germany
Robert H. Redfeld, USA
Alexander Rosa, Canada
Frank C. Sommen, Belgium
Yucai Su, China
Chun-Lei Tang, China
Ram U. Verma, USA
Dorothy I. Wallace, USA
Pei Y. Wu, Taiwan
Siamak Yassemi, Iran
Kaiming Zhao, Canada
Geometry
Teodor Bulboaca, Romania
A. Cavicchioli, Italy
Der-Chen Chang, USA
Christian Corda, Italy
M.-E. Craioveanu, Romania
Jerzy Dydak, USA
B. Forster-Heinlein, Germany
S. M. Gusein-Zade, Russia
Henryk Hudzik, Poland
R. Lowen, Belgium
Anil Maheshwari, Canada
Fr ed eric Mynard, USA
Hernando Quevedo, Mexico
Frdric Robert, France
Misha Rudnev, UK
N. Shanmugalingam, USA
Zhongmin Shen, China
Nistor Victor, USA
Luc Vrancken, France
Logic and Set Theory
R. Diaconescu, Romania
M. A. Efendiev, Germany
S. Gottwald, Germany
B. Jayaram, India
Laurent, Romania
Radko Mesiar, Slovakia
S. Montes-Rodriguez, Spain
Carles Noguera, Czech Republic
Andrzej Skowron, Poland
Sergejs Solovjovs, Czech Republic
Richard Wilson, Mexico
Mathematical Analysis
Asao Arai, Japan
Martino Bardi, Italy
Peter W. Bates, USA
H. Begehr, Germany
) in R ordered if
} M(
1
, . . . ,
) max {
1
, . . . ,
} (1)
for all
) = M(
(1)
, . . . ,
()
) (2)
for all permutations on {1, . . . , }, and 1-homogeneous in the
sense that
M(
1
, . . . ,
) = M(
(1)
, . . . ,
()
) (3)
for all permissible R.
If Mand Nare two -dimensional means on J, then we
say that M N if M(
1
, . . . ,
) N(
1
, . . . ,
) for all
) < N(
1
, . . . ,
)
for all
J for which
1
, . . . ,
,
(
1
, . . . ,
) = (
(
1
, . . . ,
(
1
, . . . ,
)
)
1/()
,
(8)
where
(
1
, . . . ,
) =
=1
. (9)
Means of the type
,1
are known as Lehmers means, and
those of the type
,0
are known as H older or power means.
Other means that have been studied extensively are the
elementary symmetric polynomial and elementary symmetric
polynomial ratio means defned by
(
)
1/
,
1
/
1
,
(10)
where
= (
). (11)
Tese are discussed in full detail in the encyclopedic work [3,
Chapters III and V].
It is obvious that the power means P
defned by
P
(
1
, . . . ,
) =
,0
(
1
, . . . ,
) = (
1
+ +
)
1/
(12)
that correspondto the values = 1 and = 1 are nothing but
the harmonic and arithmetic means Hand A, respectively. It
is also natural to set
P
0
(
1
, . . . ,
) = G(
1
, . . . ,
) = (
1
. . .
)
1/
,
(13)
since
lim
0
(
1
+ +
)
1/
= (
1
. . .
)
1/
(14)
for all
1
, . . . ,
> 0.
Te inequalities (7) can be written as P
1
< P
0
< P
1
.
Tese inequalities hold for any number of variables and they
follow from the more general fact that P
(
1
, . . . ,
), for
fxed
1
, . . . ,
) is a data set
in , then a -mean of is defned to be any element of at
which the function
() =
=1
(,
) (15)
attains its minimum. It is conceivable that (15) attains its min-
imumat many points, or nowhere at all. However, we shall be
mainly interested in distances on J for which (15) attains
its minimum at a unique point
max { : } (16)
for every data set . Such a distance is called a mean-produc-
ing or a mean-defning distance, and the point
is called the
-mean of or the mean of arising fromthe distance and
will be denoted by
0
(, ) = {
1 if = ,
0 if = ,
(17)
then the function () in (15) is nothing but the number
of elements in the given data set that are diferent from
, and therefore every element having maximum frequency
in minimizes (15) and is hence a
0
-mean of . Tus the
discrete metric gives rise to what is referred to in statistics as
International Journal of Mathematics and Mathematical Sciences 3
the mode of . Due to the nonuniqueness of the mode, the
discrete metric is not a mean-producing distance.
Similarly, the usual metric =
1
defned on R by
1
(, ) = | | (18)
is not a mean-producing distance. In fact, it is not very dif-
cult to see that if = (
1
, . . . ,
,
+1
] minimizes
=1
(19)
and is therefore a
1
-mean of . Similarly, one can show that
if is of an odd size = 2 1, then
is the unique
1
-
mean of . Tus the usual metric on R gives rise to what is
referred to in statistics as the median of .
On the other hand, the distance
2
defned on R by
2
(, ) = ( )
2
(20)
is a mean-producing distance, although it is not a metric. In
fact, it follows from simple derivative considerations that the
function
=1
(
)
2
(21)
attains its minimum at the unique point
=
1
=1
). (22)
Tus
2
is a mean-producing distance, and the corresponding
mean is nothing but the arithmetic mean.
It is noteworthy that the three distances that come to
mind most naturally give rise to the three most commonly
used means in statistics. In this respect, it is also worth
mentioning that a fourth mean of statistics, the so-called
midrange, will be encountered belowas a very natural limiting
distance mean.
Te distances
1
and
2
(and in a sense,
0
also) are mem-
bers of the family
of distances defned by
(, ) = | |
. (23)
It is not difcult to see that if > 1, then
is a mean-produc-
ing distance. In fact, if = (
1
, . . . ,
=1
,
(24)
then
() = ( 1)
=1
2
0,
(25)
with equality if and only if
1
= =
= . Tus
is convex and cannot attain its minimum at more than one
point. Tat it attains its minimumfollows fromthe continuity
of (), the compactness of [
1
,
, ) and is decreasing on (,
1
].
If we denote the mean that
defnes by
, then
() is the
unique zero of
=1
sign (
1
,
(26)
where sign() is defned to be 1 if is nonnegative and 1
otherwise.
Note that no matter what > 1 is, the two-dimensional
mean
arising from
() in closed form.
Problem 2. It would be interesting to investigate comparabil-
ity among {
: > 1}.
It is highly likely that no two means
are comparable.
5. Deviation and Sparseness
If is a mean-producing distance on , and if
is the
associated mean, then it is natural to defne the -deviation
D
() of a data set = (
1
, . . . ,
) by an expression like
D
() =
{ (
() ,
) : 1 } . (27)
Tus if is defned by
(, ) = ( )
2
, (28)
then
) =
1
+ +
, (29)
and D
(2)
(
1
, . . . ,
) =
2
+ +
.
(30)
In a sense, this provides an answer to those who are puzzled
and mystifed by the choice of the exponent 2 (and not any
other exponent) in the standard defnition of the standard
deviationgiveninthe right-handside of (30). Infact, distance
means were devised by the author in an attempt to remove
that mystery. Somehow, we are saying that the ordinary
average and the standard deviation
(2)
must be taken
or discarded together, being both associated with the same
distance given in (28). Since few people question the
sensibility of the defnition of given in (29), accepting the
standard defnition of the standard deviation given in (30) as
is becomes a must.
4 International Journal of Mathematics and Mathematical Sciences
It is worth mentioning that choosing an exponent other
than 2 in (30) would result in an essentially diferent notion
of deviations. More precisely, if one defnes
()
by
()
(
1
, . . . ,
) =
+ +
,
(31)
then
()
and
(2)
would of course be unequal, but more
importantly, they would not be monotone with respect to each
other, in the sense that there would exist data sets and
with
(2)
() >
()
() and
(2)
() <
()
(). Tus the choice
of the exponent in defning deviations is not as arbitrary as
some may feel. On the other hand, it is (27) and not (31) that
is the natural generalization of (30). Tis raises the following,
expectedly hard, problem.
Problem 3. Let be the distance defned by (, ) =
| |
defned in (27)
be denoted by D
. Is D
() > D
() D
2
() > D
2
()? (32)
We end this section by introducing the notion of sparse-
ness and by observing its relation with deviation. If is a
mean-producing distance on J, and if
is the associated
mean, then the -sparseness S
() of a data set =
(
1
, . . . ,
) in J can be defned by
S
() =
{ (
) : 1 < } . (33)
It is interesting that when is defned by (28), the standard
deviation coincides, up to a constant multiple, with the
sparsenss. One wonders whether this pleasant property char-
acterizes this distance .
ProblemSet 4. (4-a) Characterize those mean-producing dis-
tances whose associated mean is the arithmetic mean.
(4-b) If is as defned in (28), and if
is another mean-
producing distance whose associated mean is the arithmetic
mean, does it follow that D
and D
defned by
(
1
, . . . ,
= (
=1
)
1/
.
(34)
If we denote by =
the line in R
consisting of the
points (
1
, . . . ,
) with
1
= =
(
1
, . . . ,
of the point (
1
, . . . ,
)
with respect to the -norm given in (34). Here, a point
in
a subset of a metric (or distance) space (, ) is said to be a
best approximant in of if (,
) = min{(, ) :
}. Also, a subset of (, ) is said to be Chebyshev if every
in has exactly one best approximant in ; see [4, p. 21].
Te discussion above motivates the following defnition.
Defnition 1. Let J be an interval in Rand let be a distance
on J
. If the diagonal (J
) of J
defned by
(J
) = {(
1
, . . . ,
) J
:
1
= =
} (35)
is Chebyshev (with respect to ), then the -dimensional
mean
on J defned by declaring
(
1
, . . . ,
) = if
and only if (, . . . , ) is the best approximant of (
1
, . . . ,
) in
(J
defned
in Section 4.
In view of this, one may also defne
to be the best
approximation mean arising fromthe -normof
, that is,
the norm
defned on R
by
(
1
, . . . ,
= max {
: 1 } . (36)
It is not very difcult to see that
() =
1
+
2
. (37)
In viewof the fact that
() = lim
())
(38)
for every . An afrmative answer is established in [5,
Teorem 1]. In that theorem, it is also established that
lim
()) =
()
(39)
for all and all . All of this can be expressed by saying that
=
(40)
for all points in R
.
International Journal of Mathematics and Mathematical Sciences 5
Problem Set 5. Suppose that
is a sequence of distances on
a set that converges to a distance
(, ) =
, is it
necessarily true that is Chebyshev with respect to
?
(5-b) If is Chebyshev with respect to each
and with
respect to
and if
and
converges to
?
We end this section by remarking that if =
is
the -dimensional best approximation mean arising from a
distance on J
) and V (J
). Other values
of are not signifcant. Tis, together with the fact that
every mean is a best approximation mean arising
from a metric,
(41)
makes the study of best approximation means less interesting.
Fact (41) was proved in an unduly complicated manner in
[6], and in a trivial way based on a few-line set-theoretic
argument in [7].
Problem 6. Given a mean M on J, a metric on J is
constructed in [6] so that M is the best approximation
mean arising from . Since the construction is extremely
complicated in comparison with the construction in [7], it is
desirable to examine the constructionof in[6] andsee what
other nice properties (such as continuity with respect to the
usual metric) has. Tis would restore merit to the construc-
tion in [6] and to the proofs therein and provide raison d etre
for the so-called generalized means introduced there.
7. Towards a Unique Median
As mentioned earlier, the distance
1
on R defned by (23)
does not give rise to a (distance) mean. Equivalently, the 1-
norm
1
on R
1
() =
1
() = the median interval of
= the set of all medians of .
(42)
From a mathematicians point of view, however, this leaves a
lot to be desired, to say the least. Te feasibility and naturality
of defning
as the limit of
as approaches gives
us a clue on how the median
1
may be defned. It is a
pleasant fact, proved in[5, Teorem 4], that the limit of
()
(equivalently of
be
the limit of
as decreases to 1.
(7-a) Explore how the value of
continuous on R
, is the convergence of
() (as
decreases to 1) to
() monotone?
Te convergence of
() (as decreases to 1) to
()
is described in [5, Teorem 4], where it is proved that the
convergence is ultimately monotone. It is also proved in
[5, Teorem 5] that when = 3, then the convergence is
monotone.
It is of course legitimate to question the usefulness of
defning the median to be
, a sequence
, as decreases to 1, of their
associated distance means
of
. In this case,
to being
the median. However, the naturality of
and
, N, are sequences
of distances on a set that converge to the distances
and
(, ) =
is mean
producing? If so, and if the mean produced by
is
converges to
?
(8-b) If
and
and
, and
if
?
8. Examples of Distance Means
It is clear that the arithmetic mean is the distance mean
arising from the the distance
2
given by
2
(, ) = ( )
2
.
Similarly, the geometric mean on the set of positive numbers
is the distance mean arising from the distance
G
given by
G
(, ) = (ln ln )
2
. (43)
In fact, this should not be amazing since the arithmetic mean
Aon R and the geometric mean G on (0, ) are equivalent
in the sense that there is a bijection : (0, ) R, namely
() = ln , for which G(, ) =
1
A((), ()) for all
, . Similarly, the harmonic and arithmetic means on (0, )
are equivalent via the bijection () = 1/, and therefore
6 International Journal of Mathematics and Mathematical Sciences
the harmonic mean is the distance mean arising from the
distance
H
given by
H
(, ) = (
1
)
2
.
(44)
Te analogous question pertaining to the logarithmic mean
Ldefned by
L(, ) =
ln ln
, , > 0, (45)
remains open.
Problem 9. Decide whether the mean L(defned in (45)) is
a distance mean.
9. Quasi-Arithmetic Means
A -dimensional mean Mon J is called a quasi-arithmetic
mean if there is a continuous strictly monotone function
from J to an interval I in R such that
M(
1
, . . . ,
) =
1
(A((
1
) , . . . , (
))) (46)
for all
, ) . (49)
It is direct to see that (49) has a unique zero and that this zero
does indeed defne a mean.
Problem 11. Characterize deviation means and explore their
exact relationship with distance means.
If is a deviation, then (following [11]), one may defne
by
(, ) =
(, ) . (50)
Ten
(, ) 0 and
is nothing but
the unique value of at which
(
1
, ) + +
, ) attains
its minimum. Tus if
.
11. Other Ways of Generating New Means
If and are diferentiable on an open interval J, and if <
are points in J such that () =(), then there exists, by
Cauchys mean value theorem, a point in (, ), such that
()
()
=
() ()
() ()
. (51)
If and are such that is unique for every , , then we call
the Cauchy mean of and corresponding to the functions
and , and we denote it by C
,
(, ).
Another natural way of defning means is to take a
continuous function that is strictly monotone on J, and to
defne the mean of , J, =, to be the unique point in
(, ) such that
() =
1
() .
(52)
We call the mean value (mean) of and corresponding to
, and we denote it by V(, ).
Clearly, if is an antiderivative of , then (53) can be
written as
() =
() ()
. (53)
Tus V
(, ) = C
,
(, ), where is the identity function.
For more on the these two families of means, the reader
is referred to [12] and [13], and to the references therein.
In contrast to the attitude of thinking of the mean as the
number that minimizes a certain function, there is what one
may call the Chisini attitude that we nowdescribe. Afunction
on J
) = (, . . . , ) (54)
International Journal of Mathematics and Mathematical Sciences 7
has a unique solution = [
1
,
) is of interest, if the
function assumes the same value when all the
are replaced
by the mean value : (
1
, . . . ,
} M(
1
, . . . ,
) max {
1
, . . . ,
} (55)
for all
) in the indeterminates
1
, . . . ,
=1
=1
=
}
}
}
(56)
is a basis, where
=1
. (57)
Te statement above is quite easy to prove in the special case
3, and this is the case we are interested in in this paper.
We also discard the trivial case = 1 and assume always that
2.
Linear forms can be written as
1
, and they are not
worth much investigation. Quadratic forms can be written as
=
2
1
+
2
= (
=1
)
2
+ (
=1
). (58)
Cubic and quartic forms can be written, respectively, as
3
1
+
1
2
+
3
,
4
1
+
2
1
2
+
1
3
+
2
2
.
(59)
A form = (
1
, . . . ,
) is said to be copositive if
(
1
, . . . ,
) 0 for all
, 1 , be the
-tuple whose frst coordinates are 1s and whose remaining
coordinates are 0
s.
(i) If is quadratic, then is copositive if and only if 0
at the two test -tuples
k
()
1
= (1, 0, . . . , 0) , k
()
= (1, 1, . . . , 1) . (60)
(ii) If is cubic, then is copositive if and only if 0 at
the test -tuples
k
()
= (
1, . . . , 1,
0, . . . , 0), 1 . (61)
Part (i) is a restatement of Teorem 1(a) in [16]. Teo-
rem 1(b) there is related and can be restated as
(
1
, . . . ,
) 0,
R,
0 at the 3 -tuples
(1, 0, . . . , 0) , (1, 1, . . . , 1) , (1, 1, 0, . . . , 0) .
(62)
Part (ii) was proved in [17] for 3 and in [18] for all . Two
very short and elementary inductive proofs are given in [19].
It is worth mentioning that the test -tuples in (61)
do not sufce for establishing the copositivity of a quartic
form even when = 3. An example illustrating this that
uses methods from [20] can be found in [19]. However, an
algorithm for deciding whether a symmetric quartic form
in variables is copositive that consists in testing at -tuples
of the type
(
, . . . , ,
1, . . . , 1,
0, . . . , 0),
0 , , +
(63)
is established in [21]. It is also proved there that if = 3, then
the same algorithm works for quintics but does not work for
forms of higher degrees.
12.2. Internality Tests for Means Arising from Symmetric
Forms. Let F
()
)
1/()
, (64)
8 International Journal of Mathematics and Mathematical Sciences
where
, 1 , be as defned in Teorem 2.
(i) is internal if and only if it is internal at the two test
-tuples: k
()
= (1, 1, . . . , 1) and V
()
1
= (1, 1, . . . , 1, 0).
(ii) / is internal if and only if it is internal at the two test
-tuples: k
()
= (1, 1, . . . , 1) and V
()
1
= (1, 0, . . . , 0).
(iii) If 4, then
3
= (
1, . . . , 1,
0, . . . , 0), 1 . (65)
Parts (i) and (ii) are restatements of Teorems 3 and 5 in
[16]. Part (iii) is proved in [22] in a manner that leaves a lot to
be desired. Besides being rather clumsy, the proof works for
4 only. Te problem for 5, together with other open
problems, is listed in the next problem set.
ProblemSet 13. Let , , and be real symmetric cubic forms
of degrees 1, 2, and3, respectively, in non-negative variables.
(13-a) Prove or disprove that
3
= (
1, . . . , 1,
0, . . . , 0), 1 . (66)
(13-b) Find, or prove the nonexistence of, a fnite set of
test -tuples such that the internality of / at the -
tuples in gurantees its internality at all nonnegative
-tuples.
(13-c) Find, or prove the nonexistence of, a fnite set of
test -tuples such that the internality of at
the -tuples in guarantees its internality at all non-
negative -tuples.
Problem(13-b) is open even for = 2. In Section 6 of [15],
it is shown that the two pairs (1, 0) and (1, 1) do not sufce as
test pairs.
As for Problem (13-c), we refer the reader to [23],
where means of the type were considered. It is
proved in Teorem 2 there that when has the special form
1<
(
)
2
, then is internal if and only
if it is internal at the two test -tuples k
()
= (1, 1, . . . , 1)
and k
()
1
= (1, 1, . . . , 1, 0). In the general case, sufcient and
necessary conditions for internality of , in terms of
the coefcients of and , are found in [23, Teorem 3].
However, it is not obvious whether these conditions can be
rewritten in terms of test -tuples in the manner done in
Teorem 3.
13. Extension of Means, Concordance
of Means
Te two-dimensional arithmetic mean A
(2)
defned by
A
(2)
(
1
,
2
) =
1
+
2
2
(67)
can be extended to any dimension by setting
A
()
(
1
, . . . ,
) =
1
+ +
. (68)
Although very few people would disagree on this, nobody
can possibly give a mathematically sound justifcation of the
feeling that the defnition in (68) is the only (or even the best)
defnition that makes the sequence
()
of means harmonious
or concordant. Tis does not seem to be an acceptable defni-
tion of the notion of concordance.
In a private communication several years ago, Professor
Zsolt P ales told me that Kolmogorov suggested calling a
sequence M
()
of means on J, where M
()
is -dimensional,
concordant if for every and and every
in J, we have
M
(+)
(
1
, . . . ,
,
1
, . . . ,
)
= M
(2)
(M
()
(
1
, . . . ,
) , M
(
1
, . . . ,
)) .
(69)
He also told me that such a defnition is too restrictive and
seems to confrm concordance in the case of the quasi-arith-
metic means only.
Problem 14. Suggest a defnition of concordance, and test it
on sequences of means that you feel concordant. In particular,
test it on the existing generalizations, to higher dimensions,
of the logarithmic mean Ldefned in (45).
14. Distance Functions in Topology
Distance functions, which are not necessarily metrics, have
appeared early in the literature on topology. Given a distance
function on any set , one may defne the open ball (, )
in the usual manner, and then one may declare a subset
openif it contains, for every , anopenball (, ) with
> 0. If has the triangle inequality, then one can proceed in
the usual manner to create a topology. However, for a general
distance , this need not be the case, and distances that give
rise to a coherent topology in the usual manner are called
semimetrics and they are investigated and characterized in
[2429]. Clearly, these are the distances for which the family
{(, ) : > 0} of open balls centered at forms a local
base at for every in .
International Journal of Mathematics and Mathematical Sciences 9
15. Centers and Center-Producing Distances
A distance may be defned on any set whatsoever. In
particular, if is a distance on R
2
and if the function ()
defned by
() =
=1
(,
) (70)
attains its minimum at a unique point
0
that lies in the
convex hull of {
1
, . . . ,
in
R
2
, then will be called a center-producing distance.
Te Euclidean metric
1
on R
2
produces the Fermat-
Torricelli center. Tis is defned to be the point whose distan-
ces from the given points have a minimal sum. Its square,
2
, which is just a distance but not a metric, produces the
centroid. Tis is the center of mass of equal masses placed at
the given points. It would be interesting to explore the centers
defned by
(, ) =
(, ) +
(, ) +
(, )
=
.
(71)
Investigate how
() =
(), R
+
. Ten we
asked for a characterization of , such that the inequalities (, )
,
(, ) (, ) or (, )
,
(, ) (, ) hold for
each positive , , where , , , are the harmonic, arithmetic, logarithmic, and identric means, respectively. For a subclass of
with
() =
+
1
)
1
;
= (, ) :=
; = (, ) :=
log log
;
= (, ) :=
(
)
1/()
;
= (, ) :=
+
2
; = (, ) :=
/(+)
/(+)
(3)
are the harmonic, geometric, logarithmic, identric, arith-
metic, and Gini mean, respectively.
An easy task is to construct intermediate means related to
two given means and with . For instance, for an
arbitrary mean , we have that
(, ) ((, ) , (, )) (, ). (4)
Te problemis more difcult if we have to decide whether
the given mean is intermediate or not. For example, the
relation
(, )
(, ) (, ) (5)
holds for each positive and if and only if 0 1, where
the Stolarsky mean
(, ) := (
( )
)
1/(1)
.
(6)
Also,
(, )
(, ) (, ) (7)
holds if and only if 0 1, where the H older mean of order
is defned by
(, ) := (
2
)
1/
.
(8)
2 International Journal of Mathematics and Mathematical Sciences
An inverse problemis to fnd best possible approximation
of a given mean by elements of an ordered class of means
. A good example for this topic is comparison between the
logarithmic mean and the class
= and
1
= , it follows
from (2) that
0
1
. (9)
Since
0
1/2
. (10)
Finally, Lin showed in [3] that
0
1/3
(11)
is the best possible approximation of the logarithmic mean by
the means from the class
.
Numerous similar results have been obtained recently.
For example, an approximation of Seiferts mean by the class
,
(, ) :=
(, )
(, )
=
() + () 2(( + ) /2)
() + () 2(( + ) /2)
(13)
represents a mean of two positive numbers , ; that is, the
relation
min {, }
,
(, ) max {, } (14)
holds for each , R
+
, if and only if the relation
() =
() (15)
holds for each R
+
.
Let ,
() defned to
be
() =
{
{
{
{
{
{
{
+ 1
( 1)
, ( 1) =0;
log 1, = 0;
log + 1, = 1.
(18)
Since
() =
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
1
1
1
, ( 1) =0;
1
l
, = 0;
log , = 1,
() =
2
, R, > 0,
(19)
it follows that
) .
We will give in the sequel a complete answer to the above
question concerning the means
+1
(, )
(, )
:=
(, ) (20)
defned by
(, )
=
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
1
+ 1
+1
+
+1
2(( + ) /2)
+1
2(( + ) /2)
, R/ {1, 0, 1} ;
2 log (( + ) /2) log log
1/2 + 1/2 2/ ( + )
, = 1;
log + log ( + ) log (( + ) /2)
2 log (( + ) /2) log log
, = 0;
( )
2
4 ( log + log ( + ) log (( + ) /2))
, = 1.
(21)
Tose means are obviously symmetric and homogeneous
of order one.
As a consequence we obtainsome newintermediate mean
values; for instance, we show that the inequalities
(, )
1
(, ) (, )
0
(, ) (, )
1
(, ) (, )
(22)
hold for arbitrary , R
+
. Note that
1
=
2
2
log (/)
;
0
=
log (/)
log (/)
;
1
=
1
2
log (/)
.
(23)
International Journal of Mathematics and Mathematical Sciences 3
2. Results
We prove frstly the following
Teorem 1. Let ,
2
() with
> 0. Te expression
,
(, ) represents a mean of arbitrary numbers , if
and only if the relation (15) holds for .
Remark 2. In the same way, for arbitrary , > 0, + = 1,
it can be deduced that the quotient
,
(, ; , ) :=
() + () ( + )
() + () ( + )
(24)
represents a mean value of numbers , if and only if (15)
holds.
A generalization of the above assertion is the next.
Teorem 3. Let , : R be twice continuously
diferentiable functions with
},
= 1, 2, . . . ,
,
(, ) :=
) (
) (
)
, 2, (25)
represents a mean of an arbitrary set of real numbers
1
,
2
, . . . ,
() =
() (26)
holds for each .
Remark 4. It should be noted that the relation
() =
()
determines in terms of in an easy way. Precisely,
() = () 2() + + , (27)
where () :=
1
() and and are constants.
Our results concerning the means
(, ), R are
included in the following.
Teorem 5. For the class of means
(, ) (, ) for each 4;
(3) (, )
(, ) (, ) for 3 1;
(4) (, )
(, ) (, ) for 1/2 0;
(5) there is a number
0
(1/12, 1/11) such that (, )
(, ) (, ) for
0
1;
(6) there is a number
1
(1.03, 1.04) such that (, )
(, ) (, ) for
1
2;
(7) (, )
(, ) (, ) for each 2 5;
(8) there is no fnite such that the inequality (, )
,
(, ) = .
(29)
From the other hand, due to lHospitals rule we obtain
lim
,
(, ) = lim
(
()
(( + ) /2)
()
(( + ) /2)
)
= lim
(
2
()
(( + ) /2)
2
()
(( + ) /2)
)
=
()
()
.
(30)
Comparing (29) and (30) the desired result follows.
Suppose now that (15) holds and let < . Since
() >
0 [, ] by the Cauchy mean value theorem there exists
(( + )/2, ) such that
()
(( + ) /2)
()
(( + ) /2)
=
()
()
= . (31)
But,
+
2
< < , (32)
and, since
is strictly increasing,
()
((+)/2) > 0,
[, ].
Terefore, by (31) we get
(
()
(
+
2
))
()
(
+
2
)
(
()
(
+
2
)).
(33)
Finally, integrating (33) over [, ] we obtain the assertion
from Teorem 1.
3.2. Proof of Teorem 3. We will give a proof of this assertion
by induction on .
By Remark 2, it holds for = 2.
Next, it is not difcult to check the identity
) (
)
= (1
) (
1
) (
1
))
+ [(1
) () +
) ((1
) +
)] ,
(34)
4 International Journal of Mathematics and Mathematical Sciences
where
:=
1
:=
(1
)
, = 1, 2, . . . , 1;
1
= 1.
(35)
Terefore, by induction hypothesis and Remark 2, we get
) (
)
max {
1
,
2
, . . .
1
} (1
)
(
1
) (
1
))
+ max {,
} [(1
) () +
)
((1
) +
)]
max {
1
,
2
, . . . ,
}
((1
) (
1
) (
1
))
+[(1
) () +
) ((1
) +
)] )
= max {
1
,
2
, . . . ,
} (
) (
)).
(36)
Te inequality
min {
1
,
2
, . . . ,
}
,
(, ) (37)
can be proved analogously.
For the proof of necessity, put
2
=
3
= =
and
proceed as in Teorem 1.
Remark 6. It is evident from (15) that if R
+
then has to
be also convex on . Otherwise, it shouldnt be the case. For
example, the conditions of Teorem 3 are satisfedwith() =
3
/3, () =
2
, R. Hence, for an arbitrary sequence {
1
of real numbers, we obtain
min {
1
,
2
, . . . ,
)
3
3 (
)
2
)
max {
1
,
2
, . . . ,
} .
(38)
Because the above inequality does not depend on , a
probabilistic interpretation of the above result is contained in
the following.
Teorem 7. For an arbitrary probability law of random
variable with support on (, +), one has
()
3
+ 3 (min )
2
3
()
3
+ 3 (max )
2
.
(39)
3.3. Proof of Teorem 5, Part (1). We will prove a general
assertion of this type. Namely, for an arbitrary positive
sequence x = {
}, = 1, 2, . . ., denote
(p, x)
:=
{
{
{
{
{
{
{
{
{
{
{
{
{
( 1)
, R/ {0, 1} ;
log (
log
, = 0;
log
) log (
) , = 1.
(40)
For R, > 0 we have
(p, x)
++1
(p, x)
+1
(p, x)
+
(p, x) , (41)
which is equivalent to
Teorem 8. Te sequence {
+1
(p, x)/
(p, x))
(p, x))
(p, x))
(42)
holds for arbitrary sequences p, x.
Putting there = , = + 1, = + + 1 and = ,
= + , = + + 1, we successively obtain
(
+1
(p, x))
+1
(
(p, x))
++1
(p, x),
(
+
(p, x))
+1
(p, x) (
++1
(p, x))
.
(43)
Since > 0, multiplying those inequalities we get the
relation (41), that is, the proof of Teorem 8.
Te part (1) of Teorem 5 follows for
1
=
2
= 1/2.
A general way to prove the rest of Teorem 5 is to use an
easy-checkable identity
(, )
(, )
=
(1 + , 1 ) , (44)
with := ( )/( + ).
International Journal of Mathematics and Mathematical Sciences 5
Since 0 < < , we get 0 < < 1. Also,
(, )
(, )
= 1
2
;
(, )
(, )
=
1
2
;
(, )
(, )
=
2
log (1 + ) log (1 )
;
(, )
(, )
= exp (
(1 + ) log (1 + ) (1 ) log (1 )
2
1);
(, )
(, )
= exp (
1
2
((1 + ) log (1 + ) + (1 ) log (1 ))).
(45)
Terefore, we have to compare some one-variable
inequalities and to check their validness for each (0, 1).
For example, we will prove that the inequality
(, ) (, ) (46)
holds for each positive , if and only if 0.
Since
0
(, )
(, )
1. (47)
By the above formulae, this is equivalent to the assertion
that the inequality
() 0 (48)
holds for each (0, 1), with
() :=
log (1 + ) log (1 )
2
((1 + ) log (1 + ) + (1 ) log (1 ))
+ log (1 + ) + log (1 ).
(49)
We will prove that the power series expansion of ()
have non-positive coefcients. Tus the relation (48) will be
proved.
Since
log (1 + ) log (1 )
2
=
0
2
2 + 1
;
log (1 + ) + log (1 ) =
2
0
2
+ 1
;
(1 + ) log (1 + ) + (1 ) log (1 )
=
2
0
2
( + 1) (2 + 1)
,
(50)
we get
()
2
=
=0
(
1
+ 1
+
=0
1
(2 2 + 1) ( + 1) (2 + 1)
)
2
=
2
.
(51)
Hence,
0
=
1
= 0;
2
=
1
90
, (52)
and, afer some calculation, we get
=
2
( + 1) (2 + 3)
(( + 2)
1
1
2 + 1
( + 1)
1
1
2
),
> 1.
(53)
Now, one can easily prove (by induction, e.g.) that
:= ( + 2)
1
1
2 + 1
( + 1)
1
1
2
(54)
is a negative real number for 2. Terefore
0, and the
proof of the frst part is done. For 0 < < 1 we have
(, )
(, )
1
=
(1 ) ((1 + )
+1
+ (1 )
+1
2) log ((1 + ) / (1 ))
2 (1 + ) (2 (1 + )
(1 )
)
1
=
1
6
2
+ (
4
) ( 0) .
(55)
Terefore,
(, ) (, ) (56)
holds for each , ; 0 < < if and only if 1.
As before, it is enough to consider the expression
(, )
1
(, )
=
()
() := () , (57)
with
() =
(1 + ) log (1 + ) (1 ) log (1 )
2
1;
() =
(1 + ) log (1 + ) + (1 ) log (1 )
2
.
(58)
6 International Journal of Mathematics and Mathematical Sciences
It is not difcult to check the identity
() =
()
()
3
.
(59)
Hence by (48), we get
1
(, )
lim
0
+
() = 1. (60)
By monotonicity it follows that
(, ) (, ) for 1.
For > 1, ( )/( + ) = , we have
(, ) (, ) = (
1
6
( 1)
2
+ (
4
))(, )
( 0
+
) .
(61)
Hence,
(, )
(, )
1] =
( 1) (2
+1
2)
2 ( + 1) (2
2)
1 := () .
(62)
Examining the function (), we fnd out that it has the
only real zero at
0
1.0376 and is negative for (1,
0
).
Remark 10. Since () is monotone increasing, we also get
(, )
1
(, )
lim
1
() =
4 log 2
. (63)
Hence
1
(, )
1
(, )
4 log 2
. (64)
A calculation gives 4 log 2/ 1.0200.
Note also that
2
(, ) (, ) . (65)
Terefore, applying the assertion from the part 1, we get
(, ) (, ), 2;
(, ) (, ), 2.
(66)
Finally, we give a detailed proof of the part 7.
We have to prove that
(, ) (, ) for 5. Since
5
(, ) (, ) (67)
holds for each , R
+
.
Terefore, by the transformation given above, we get
log
5
= log [
2
3
(1 + )
6
+ (1 )
6
2
(1 + )
5
+ (1 )
5
2
]
= log [
2
15
15 + 15
2
+
4
2 +
2
]
log [
1 +
2
+
4
/4
1 +
2
/2
] = log (1 +
2
2
)
=
2
2
4
8
+
6
24
2
2
+
4
12
+
6
30
+
=
1
2
((1 + ) log (1 + ) + (1 ) log (1 ))
= log
,
(68)
and the proof is done.
Further, we have to show that
+ (1 )
2 = (
2
)
2
+ (
4
)
4
+ (
6
) , (69)
for > 5 and sufciently small , we get
=
1
+ 1
(
+1
2
)
2
+ (
+1
4
)
4
+ (
6
)
(
2
)
2
+ (
4
)
4
+ (
6
)
=
1 + ( 1) ( 2)
2
/12 + (
4
)
1 + ( 2) ( 3)
2
/12 + (
4
)
= 1 + (
6
1
3
)
2
+ (
4
) .
(70)
Similarly,
= exp (
1
2
((1 + ) log (1 + ) + (1 ) log (1 )))
= exp (
2
2
+ (
4
)) = 1 +
2
2
+ (
4
) .
(71)
Hence,
1
) =
1
6
( 5)
2
+ (
4
) , (72)
and this expression is positive for > 5 and sufciently
small, that is, sufciently close to .
International Journal of Mathematics and Mathematical Sciences 7
As for the part 8, applying the above transformation we
obtain
(, )
(, )
=
1
+ 1
(1 + )
+1
+ (1 )
+1
2
(1 + )
+ (1 )
2
exp (
1
2
((1 + ) log (1 + ) + (1 ) log (1 ))),
(73)
where 0 < < , = ( )/( + ).
Since for > 5,
lim
1
=
1
+ 1
2
1
2
2
, (74)
and the last expression is less than one, it follows that the
inequality (, ) <
\ {x, x, x R
} R
is called a homogeneous
function generated by f with parameters p and q if H
f
is dened by for a
/
b
H
f
_
p, q; a, b
_
_
fa
p
, b
p
fa
q
, b
q
_
1pq
, if pq
_
p q
_
/
0,
H
f
_
p, p; a, b
_
exp
_
a
p
f
x
a
p
, b
p
lna b
p
f
y
a
p
, b
p
lnb
fa
p
, b
p
_
, if p q
/
0,
1.6
where f
x
x, y and f
y
x, y denote rst-order partial derivatives with respect to rst and
second component of fx, y, respectively.
If lim
y x
fx, y exits and is positive for all x R
_
fa
p
, b
p
f1, 1
_
1/p
, if p
/
0, q 0,
H
f
_
0, q; a, b
_
_
fa
q
, b
q
f1, 1
_
1/q
, if p 0, q
/
0,
H
f
0, 0; a, b a
f
x
1,1/f1,1
b
f
y
1,1/f1,1
, if p q 0,
1.7
and H
f
p, q; a, a a.
Remark 1.2. Witkowski 17 proved that if the function x, y fx, y is a symmetric and
rst-order homogeneous function, then for all p, q H
f
p, q; a, b is a mean of positive numbers
a and b if and only if f is increasing in both variables on R
_
_
q
p
a
p
b
p
a
q
b
q
_
1/pq
, if pq
_
p q
_
/
0,
L
1/p
a
p
, b
p
, if p
/
0, q 0,
L
1/q
a
q
, b
q
, if q
/
0, p 0,
I
1/p
a
p
, b
p
, if p q
/
0,
ab, if p q 0,
1.8
where Ix, y e
1
x
x
/y
y
1/xy
if x, y > 0, with x
/
y, and Ix, x x is the identric
exponential mean see 18. Substituting A Ax, y x y/2 for f yields Gini means
H
A
p, q; a, b G
p,q
a, b dened by
G
p,q
a, b
_
_
_
a
p
b
p
a
q
b
q
_
1/pq
, if p
/
q,
Z
1/p
a
p
, b
p
, if p q,
1.9
where Za, b a
a/ab
b
b/ab
see 19.
As consequences of our results, the monotonicity properties of four ratios of mixed
Stolarsky means are presented, which generalize certain known results, and some known and
new inequalities of ratios of means are established.
2. Main Results and Proofs
In 15, 16, 20, two decision functions play an important role, that are,
I I
_
x, y
_
2
lnf
_
x, y
_
xy
_
lnf
_
x, y
__
xy
_
lnf
_
xy
,
J J
_
x, y
_
_
x y
_xI
x
_
x y
_
xI
x
.
2.1
In 14, it is important to another key decision function dened by
T
3
_
x, y
_
: xyxI
x
ln
3
_
x
y
_
, where I
_
lnf
_
xy
, x a
t
, y b
t
. 2.2
Note that the function T dened by
Tt : lnf
_
a
t
, b
t
_
, t
/
0 2.3
4 International Journal of Mathematics and Mathematical Sciences
has well properties see 15, 16. And it has shown in 14, 3.4, 16, Lemma 4 the relation
among T
t, Jx, y and T
3
x, y:
T
t t
3
T
3
_
x, y
_
, where x a
t
, y b
t
, 2.4
T
t Ct
3
J
_
x, y
_
, where C xy
_
x y
_
1
_
lnx lny
_
3
> 0. 2.5
Moreover, it has revealed in 14, 3.5 that
T
3
_
x, y
_
T
3
_
x
y
, 1
_
T
3
_
1,
y
x
_
. 2.6
Now, we observe the monotonicities of ratio of certain mixed means dened by 1.1.
Theorem 2.1. Suppose that f: R
, so T
t is continuous on p, q or q, p for
p, q R, then 2.13 in 13 holds. Thus we have
lng
1f
_
p
_
1
2
lnH
f
_
p, q
_
1
2
lnH
f
_
2k p, q
_
1
2
_
1
0
T
t
11
dt
1
2
_
1
0
T
t
12
dt, 2.7
where
t
12
tp 1 tq, t
11
t
_
2k p
_
1 tq. 2.8
Partial derivative leads to
_
lng
1f
_
p
__
1
2
_
1
0
tT
t
12
dt
1
2
_
1
0
tT
t
11
dt
1
2
_
1
0
tT
|t
12
|dt
1
2
_
1
0
tT
|t
11
|dt
_
by13, 2.7
_
1
2
_
1
0
t
_
|t
12
|
|t
11
|
T
vdv dt,
2.9
International Journal of Mathematics and Mathematical Sciences 5
and then
_
lnQ
1f
_
p
__
_
lng
1f
_
p; a, b
__
_
lng
1f
_
p; c, d
__
1
2
_
1
0
t
_
|t
12
|
|t
11
|
T
vdv dt
1
2
_
1
0
t
_
|t
12
|
|t
11
|
T
v; c, ddv dt
_
1
0
t|t
12
| |t
11
|
_
|t
12
|
|t
11
|
T
v; a, b T
v; c, ddv
|t
12
| |t
11
|
dt
:
_
1
0
t|t
12
| |t
11
|h|t
11
|, |t
12
|dt,
2.10
where
h
_
x, y
_
:
_
_
_
y
x
T
v; a, b T
v; c, ddv
y x
, if x
/
y,
T
x; a, b T
x; c, d, if x y.
2.11
Since T
3
1, u strictly increase decrease with u > 1 and decrease increase with 0 < u < 1,
2.4 and 2.6 together with b/a > d/c 1 yield
T
v; a, b T
v; c, d v
3
T
3
a
v
, b
v
T
3
c
v
, d
v
v
3
_
T
3
_
1,
_
b
a
_
v
_
T
3
_
1,
_
d
c
_
v
__
> <0, for v > 0,
2.12
and therefore hx, y > <0 for x, y > 0. Thus, in order to prove desired result, it suces to
determine the sign of |t
12
| |t
11
|. In fact, if q 0, k 0, then for t 0, 1
|t
12
| |t
11
|
t
2
12
t
2
11
|t
12
| |t
11
|
4t
q1 t kt
t
12
t
11
_
p k
_
_
> 0, if p > k,
< 0, if p < k.
2.13
It follows that
_
lnQ
1f
_
p
__
_
> <0, if p > k,
< >0, if p < k.
2.14
Clearly, the monotonicity of Q
1f
is converse if q 0, k 0.
This completes the proof.
Theorem 2.2. The conditions are the same as those of Theorem 2.1. Then, for any a, b, c, d > 0 with
b/a > d/c 1 and xed m, k with k 0, k m 0, but m, k are not equal to zero at the same time,
Q
2f
is strictly increasing (decreasing) in p on k, and decreasing (increasing) on , k.
6 International Journal of Mathematics and Mathematical Sciences
The monotonicity of Q
2f
is converse if k 0 and k m 0, but m, k are not equal to zero at
the same time.
Proof. By 2.13 in 13 we have
lng
2f
_
p
_
1
2
lnH
f
_
p, p m
_
1
2
lnH
f
_
2k p, 2k p m
_
1
2
_
1
0
T
t
22
dt
1
2
_
1
0
T
t
21
dt,
2.15
where
t
22
tp 1 t
_
p m
_
, t
21
t
_
2k p
_
1 t
_
2k p m
_
. 2.16
Direct calculation leads to
_
lng
2f
_
p
__
1
2
_
1
0
T
t
22
dt
1
2
_
1
0
T
t
21
dt
1
2
_
1
0
_
|t
22
|
|t
21
|
T
_
lng
2f
_
p; a, b
__
_
lng
2f
_
p; c, d
__
1
2
_
1
0
_
|t
22
|
|t
21
|
T
v; a, bdv dt
1
2
_
1
0
_
|t
22
|
|t
21
|
T
v; c, ddv dt
1
2
_
1
0
|t
22
| |t
21
|h|t
21
|, |t
22
|dt,
2.18
where hx, y is dened by 2.11. As shown previously, hx, y > <0 for x, y > 0 if T
3
1, u
strictly increase decrease with u > 1 and decrease increase with 0 < u < 1; it remains to
determine the sign of |t
22
| |t
21
|. It is easy to verify that if k 0 and k m 0, then
|t
22
| |t
21
|
t
2
22
t
2
21
|t
22
| |t
21
|
4
k m1 t
|t
22
| |t
21
|
_
p k
_
_
> 0, if p > k,
< 0, if p < k.
2.19
Thus, we have
_
lnQ
2f
_
p
__
_
> <0, if p > k,
< >0, if p < k.
2.20
Clearly, the monotonicity of Q
2f
is converse if k 0 and k m 0.
The proof ends.
International Journal of Mathematics and Mathematical Sciences 7
Theorem 2.3. The conditions are the same as those of Theorem 2.1. Then, for any a, b, c, d > 0 with
b/a > d/c 1 and xed m > 0, 0 k 2m, Q
3f
is strictly increasing (decreasing) in p on k,
and decreasing (increasing) on , k.
The monotonicity of Q
2f
is converse if m < 0, 2m k 0.
Proof. From 2.13 in 13, it is derived that
lng
3f
_
p
_
1
2
lnH
f
_
p, 2m p
_
1
2
lnH
f
_
2k p, 2m 2k p
_
1
2
_
1
0
T
t
32
dt
1
2
_
1
0
T
t
31
dt,
2.21
where
t
32
_
tp 1 t
_
2m p
__
, t
31
_
t
_
2k p
_
1 t
_
2m 2k p
__
. 2.22
Simple calculation yields
_
lng
3f
_
p
__
1
2
_
1
0
2t 1
_
T
t
32
T
t
31
_
dt
1
2
_
1
0
2t 1
_
|t
32
|
|t
31
|
T
v; a, bdv dt.
2.23
Hence,
_
lnQ
3f
_
p
__
_
lng
3f
_
p; a, b
__
_
lng
3f
_
p; c, d
__
1
2
_
1
0
2t 1
_
|t
32
|
|t
31
|
_
T
v; a, b T
v; c, d
_
dv dt
1
2
_
1
0
2t 1|t
32
| |t
31
|h|t
31
|, |t
32
|dt,
2.24
where hx, y is dened by 2.11. It has shown that hx, y > <0 for x, y > 0 if T
3
1, u
strictly increase decrease with u > 1 and decrease increase with 0 < u < 1, and we have
also to check the sign of 2t 1|t
32
| |t
31
|. Easy calculation reveals that if m > 0, 0 k 2m,
then
2t 1|t
32
| |t
31
| 2t 1
_
t
2
32
t
2
31
_
|t
32
| |t
31
|
42t 1
2
tk 1 t2m k
|t
32
| |t
31
|
_
p k
_
_
> 0, if p > k,
< 0, if p < k,
2.25
8 International Journal of Mathematics and Mathematical Sciences
which yields
_
lnQ
3f
_
p
__
_
> <0, if p > k,
< >0, if p < k.
2.26
It is evident that the monotonicity of Q
3f
is converse if m < 0, 2m k 0.
Thus the proof is complete.
Theorem 2.4. The conditions are the same as those of Theorem 2.1. Then, for any a, b, c, d > 0 with
b/a > d/c 1 and xed k, r, s R with r s
/
0, Q
4f
is strictly increasing (decreasing) in p on
k, and decreasing (increasing) on , k if kr s > 0.
The monotonicity of Q
4f
is converse if kr s < 0.
Proof. By 2.13 in 13, lnH
f
pr, ps can be expressed in integral form
lnH
f
_
pr, ps
_
_
_
_
1
r s
_
r
s
T
_
pt
_
dt, if r
/
s,
T
_
pr
_
, if r s.
2.27
The case r s
/
0 has no interest since it can come down to the case of m 0 in Theorem 2.2.
Therefore, we may assume that r
/
s. We have
lng
4f
_
p
_
ln
_
H
f
_
pr, ps
_
H
f
__
2k p
_
r,
_
2k p
_
s
_
1
2
1
r s
_
r
s
T
_
pt
_
dt
1
2
1
r s
_
r
s
T
__
2k p
_
t
_
dt,
2.28
and then
_
lng
4f
_
p
__
1
2
1
r s
_
r
s
tT
_
pt
_
dt
1
2
1
r s
_
r
s
tT
__
2k p
_
t
_
dt
1
2
1
r s
_
r
s
t
_
T
_
pt
_
T
__
2k p
_
t
__
.
2.29
Note that T
pt T
can be expressed as
_
lng
4f
_
p
__
1
2
r s
|r| |s|
_
|r|
|s|
t
_
T
__
_
pt
_
_
_
T
__
_
_
2k p
_
t
_
_
__
dt
1
2
r s
|r| |s|
_
|r|
|s|
t
_
|t
42
|
|t
41
|
T
vdv dt,
2.30
International Journal of Mathematics and Mathematical Sciences 9
where
t
42
pt, t
41
_
2k p
_
t. 2.31
Hence,
_
lnQ
4f
_
p
__
_
lng
4f
_
p; a, b
__
_
lng
4f
_
p; c, d
__
1
2
r s
|r| |s|
_
|r|
|s|
t
_
|t
42
|
|t
41
|
_
T
v; a, b T
v; c, d
_
dv dt
1
2
r s
|r| |s|
_
|r|
|s|
t|t
42
| |t
41
|h|t
41
|, |t
42
|dt,
2.32
where hx, y is dened by 2.11. We have shown that hx, y > <0 for x, y > 0 if T
3
1, u
strictly increase decrease with u > 1 and decrease increase with 0 < u < 1, and we also
have
sgn|t
42
| |t
41
| sgn
_
t
2
42
t
2
41
_
sgnk sgn
_
p k
_
. 2.33
It follows that
sgnQ
4f
_
p
_
sgnr s sgnk sgn
_
p k
_
sgnh|t
41
|, |t
42
|
_
> <0, if kr s > 0, p > k,
< >0, if kr s > 0, p < k,
< >0, if kr s < 0, p > k,
> <0, if kr s < 0, p < k.
2.34
This proof is accomplished.
3. Applications
As shown previously, S
p,q
a, b H
L
p, q; a, b, where L Lx, y is the logarithmic mean.
Also, it has been proven in 14 that T
3
1, u < 0 if u > 1 and T
3
1, u > 0 if 0 < u < 1. Fromthe
applications of Theorems 2.12.4, we have the following.
Corollary 3.1. Let a, b, c, d > 0 with b/a > d/c 1. Then, the following four functions are all
strictly decreasing (increasing) on k, and increasing (decreasing) on , k:
i Q
1L
is dened by
Q
1L
_
p
_
_
S
p,q
a, bS
2kp,q
a, b
_
S
p,q
c, dS
2kp,q
c, d
, 3.1
for xed q 0, k 0, but q, k are not equal to zero at the same time,
10 International Journal of Mathematics and Mathematical Sciences
ii Q
2L
is dened by
Q
2L
_
p
_
_
S
p,pm
a, bS
2kp,2kpm
a, b
_
S
p,pm
c, dS
2kp,2kpm
c, d
, 3.2
for xed m, k with k 0 and k m 0, but m, k are not equal to zero at the same
time,
iii Q
3L
is dened by
Q
3L
_
p
_
_
S
p,2mp
a, bS
2kp,2m2kp
a, b
_
S
p,2mp
c, dS
2kp,2m2kp
c, d
, 3.3
for xed m > <0, k 0, 2 m 2m, 0.
iv Q
4L
is dened by
Q
4L
_
p
_
_
S
pr,ps
a, bS
2kpr,2kps
a, b
_
S
pr,ps
c, dS
2kpr,2kps
c, d
, 3.4
for xed k, r, s R with kr s > <0.
Remark 3.2. Letting in the rst result of Corollary 3.1, q k yields Theorem 3.4 in 13 since
_
S
p,k
S
2kp,k
S
p,2kp
. Letting q 1, k 0 yields
Ga, b
Gc, d
Q
1L
<
_
S
p,1
a, bS
p,1
a, b
_
S
p,1
c, dS
p,1
c, d
< Q
1L
0
La, b
Lc, d
. 3.5
Inequalities 3.5 in the case of d c were proved by Alzer in 21. By letting q 1, k 1/2
from Q
1L
1/2 > Q
1L
1 > Q
1L
2, we have
Aa, b Ga, b
Ac, d Gc, d
>
_
La, bIa, b
_
Lc, dIc, d
>
_
Aa, bGa, b
_
Ac, dGc, d
. 3.6
Inequalities 3.6 in the case of d c are due to Alzer 22.
Remark 3.3. Letting in the second result of Corollary 3.1, m 1, k 0 yields Cheung and Qis
result see 23, Theorem 2. And we have
Ga, b
Gc, d
Q
2L
<
_
S
p,p1
a, bS
p,p1
a, b
_
S
p,p1
c, dS
p,p1
c, d
< Q
2L
0
La, b
Lc, d
. 3.7
When d c, inequalities 3.7 are changed as Alzers ones given in 24.
International Journal of Mathematics and Mathematical Sciences 11
Remark 3.4. In the third result of Corollary 3.1, letting k malso leads to Theorem 3.4 in 13.
Put m 1/2, k 1/4. Then from Q
3L
1/4 > Q
3L
1/2, we obtain a new inequality
He
1/2
a, b
He
1/2
c, d
>
_
La, bI
1/2
a, b
_
Lc, dI
1/2
c, d
. 3.8
Putting m 1/2, k 1/3 leads to another new inequality
A
1/3
a, b
A
1/3
c, d
>
_
S
1/6,5/6
a, bI
1/2
a, b
_
S
1/6,5/6
c, dI
1/2
c, d
. 3.9
Remark 3.5. Letting in the third result of Corollary 3.1, k 1/2 and r, s 1, 0, 1, 1, 2, 1,
and we deduce that all the following three functions
p
_
L
p
a, bL
1p
a, b
_
L
p
c, dL
1p
c, d
, p
_
I
p
a, bI
1p
a, b
_
I
p
c, dI
1p
c, d
, p
_
A
p
a, bA
1p
a, b
_
A
p
c, dA
1p
c, d
,
3.10
are strictly decreasing on 1/2, and increasing on , 1/2, where L
p
L
1/p
a
p
, b
p
, I
p
I
1/p
a
p
, b
p
, and A
p
A
1/p
a
p
, b
p
are the p-order logarithmic, identric exponential, and
power mean, respectively, particularly, so are the functions
_
L
p
L
1p
,
_
I
p
I
1p
,
_
A
p
A
1p
.
4. Other Results
Let d c in Theorems 2.12.4. Then, H
f
p, q; c, d c and T
\ {x, x, x R
}, then T
\ {x, x, x R
} R
0, 0, :
Ha, b Ga, b N
1
a, b N
3
a, b N
2
a, b Aa, b Sa, b, 1.1
where
Aa, b
a b
2
,
Ga, b
_
ab,
Ha, b
2ab
a b
,
N
1
a, b
_
b
2
_
2
Aa, b Ga, b
2
,
1.2
2 International Journal of Mathematics and Mathematical Sciences
N
3
a, b
a
ab b
3
2Aa, b Ga, b
3
,
N
2
a, b
_
b
2
_
_
_
_
a b
2
_
_
,
Sa, b
_
a
2
b
2
2
.
1.3
The means A, G, H, S, N
1
and N
3
are called, respectively, the arithmetic mean, the geometric
mean, the harmonic mean, the root-square mean, the square-root mean, and Herons mean.
The N
2
one can be found in Taneja 2, 3.
Furthermore Taneja considered the following dierence of means:
M
SA
a, b Sa, b Aa, b,
M
SN
2
a, b Sa, b N
2
a, b,
M
SN
3
a, b Sa, b N
3
a, b,
M
SN
1
a, b Sa, b N
1
a, b,
M
SG
a, b Sa, b Ga, b,
M
SH
a, b Sa, b Ha, b,
M
AN
2
a, b Aa, b N
2
a, b,
M
AG
a, b Aa, b Ga, b,
M
AH
a, b Aa, b Ha, b,
M
N
2
N
1
a, b N
2
a, b N
1
a, b,
M
N
2
G
a, b N
2
a, b Ga, b
1.4
and established the following.
Theorem A. The dierence of means given by 1.4 is nonnegative and convex in R
2
0,
0, .
Further, using Theorem A, Taneja proved several chains of inequalities; they are
renements of inequalities in 1.1.
International Journal of Mathematics and Mathematical Sciences 3
Theorem B. The following inequalities among the mean dierences hold:
M
SA
a, b
1
3
M
SH
a, b
1
2
M
AH
a, b
1
2
M
SG
a, b M
AG
a, b, 1.5
1
8
M
AH
a, b M
N
2
N
1
a, b
1
3
M
N
2
G
a, b
1
4
M
AG
a, b M
AN
2
a, b, 1.6
M
SA
a, b
4
5
M
SN
2
a, b 4M
AN
2
a, b, 1.7
M
SH
a, b 2M
SN
1
a, b
3
2
M
SG
a, b, 1.8
M
SA
a, b
3
4
M
SN
3
a, b
2
3
M
SN
1
a, b. 1.9
For the dierence of means given by 1.4, we study the Schur-geometric convexity of
dierence between these dierences in order to further improve the inequalities in 1.1. The
main result of this paper reads as follows.
Theorem I. The following dierences are Schur-geometrically convex in R
2
0, 0, :
D
SHSA
a, b
1
3
M
SH
a, b M
SA
a, b,
D
AHSH
a, b
1
2
M
AH
a, b
1
3
M
SH
a, b,
D
SGAH
a, b M
SG
a, b M
AH
a, b,
D
AGSG
a, b M
AG
a, b
1
2
M
SG
a, b,
D
N
2
N
1
AH
a, b M
N
2
N
1
a, b
1
8
M
AH
a, b,
D
N
2
GN
2
N
1
a, b
1
3
M
N
2
G
a, b M
N
2
N
1
a, b,
D
AGN
2
G
a, b
1
4
M
AG
a, b
1
3
M
N
2
G
a, b,
D
AN
2
AG
a, b M
AN
2
a, b
1
4
M
AG
a, b,
D
SN
2
SA
a, b
4
5
M
SN
2
a, b M
SA
a, b,
D
AN
2
SN
2
a, b 4M
AN
2
a, b
4
5
M
SN
2
a, b,
D
SN
1
SH
a, b 2M
SN
1
a, b M
SH
a, b,
D
SGSN
1
a, b
3
2
M
SG
a, b 2M
SN
1
a, b,
1.10
4 International Journal of Mathematics and Mathematical Sciences
D
SN
3
SA
a, b
3
4
M
SN
3
a, b M
SA
a, b,
D
SN
1
SN
3
a, b
2
3
M
SN
1
a, b
3
4
M
SN
3
a, b.
1.11
The proof of this theorem will be given in Section 3. Applying this result, in Section 4,
we prove some inequalities related to the considered dierences of means. Obtained
inequalities are renements of inequalities 1.51.9.
2. Denitions and Auxiliary Lemmas
The Schur-convex function was introduced by Schur in 1923, and it has many important
applications in analytic inequalities, linear regression, graphs and matrices, combinatorial
optimization, information-theoretic topics, Gamma functions, stochastic orderings, reliability,
and other related elds cf. 414.
In 2003, Zhang rst proposed concepts of Schur-geometrically convex function
which is extension of Schur-convex function and established corresponding decision
theorem 15. Since then, Schur-geometric convexity has evoked the interest of many
researchers and numerous applications and extensions have appeared in the literature cf.
1619.
In order to prove the main result of this paper we need the following denitions and
auxiliary lemmas.
Denition 2.1 see 4, 20. Let x x
1
, . . . , x
n
R
n
and y y
1
, . . . , y
n
R
n
.
i x is said to be majorized by y in symbols x y if
k
i1
x
i
k
i1
y
i
for k
1, 2, . . . , n 1 and
n
i1
x
i
n
i1
y
i
, where x
1
x
n
and y
1
y
n
are
rearrangements of x and y in a descending order.
ii R
n
is called a convex set if x
1
y
1
, . . . , x
n
y
n
for every x and y ,
where and 0, 1 with 1.
iii Let R
n
. The function : R is said to be a Schur-convex function on if
x y on implies x y. is said to be a Schur-concave function on if and
only if is Schur-convex.
Denition 2.2 see 15. Let x x
1
, . . . , x
n
R
n
and y y
1
, . . . , y
n
R
n
.
i R
n
1
y
1
, . . . , x
n
y
n
for all x,y
and , 0, 1 such that 1.
ii Let R
n
. The function : R
_
x
1
x
1
x
2
x
2
_
0 0 2.1
holds for any x x
1
, . . . , x
n
0
, then is a Schur-geometrically convex Schur-geometrically
concave function.
Lemma 2.5. For a, b R
2
0, 0, one has
1
a b
_
2a
2
b
2
1
2
2ab
a b
2
,
2.2
a b
_
2a
2
b
2
ab
a b
2
3
4
,
2.3
3
2
a b
2
_
b
_
a b
5
4
ab
a b
2
.
2.4
Proof. It is easy to see that the left-hand inequality in 2.2 is equivalent to a b
2
0, and
the right-hand inequality in 2.2 is equivalent to
_
2a
2
b
2
a b
_
2a
2
b
2
a b
2
4ab
2a b
2
, 2.5
that is,
a b
2
2a
2
b
2
_
2a
2
b
2
a b
a b
2
2a b
2
. 2.6
Indeed, from the left-hand inequality in 2.2 we have
2
_
a
2
b
2
_
_
2a
2
b
2
a b 2
_
a
2
b
2
_
a b
2
2a b
2
, 2.7
so the right-hand inequality in 2.2 holds.
The inequality in 2.3 is equivalent to
_
2a
2
b
2
a b
_
2a
2
b
2
a b
2
4a b
2
. 2.8
6 International Journal of Mathematics and Mathematical Sciences
Since
_
2a
2
b
2
a b
_
2a
2
b
2
2
_
a
2
b
2
_
a b
2
_
2a
2
b
2
_
_
2a
2
b
2
a b
_
a b
2
2a
2
b
2
a b
_
2a
2
b
2
,
2.9
so it is sucient prove that
2
_
a
2
b
2
_
a b
_
2a
2
b
2
4a b
2
, 2.10
that is,
a b
_
2a
2
b
2
2
_
a
2
b
2
4ab
_
, 2.11
and, from the left-hand inequalities in 2.2, we have
a b
_
2a
2
b
2
2
_
a
2
b
2
_
2
_
a
2
b
2
4ab
_
, 2.12
so the inequality in 2.3 holds.
Notice that the functions in the inequalities 2.4 are homogeneous. So, without loss of
generality, we may assume
a
b 1, and set t
ab. Then 0 < t 1/4 and 2.4 reduces
to
3
2
1 2t
2
1
1 2t
5
4
t
2
1 2t
2
. 2.13
Squaring every side in the above inequalities yields
9
4
1 2t
2
1
2 4t
1
25
16
t
4
1 2t
4
5t
2
21 2t
2
. 2.14
Reducing to common denominator and rearranging, the right-hand inequality in 2.14
reduces to
1 2t
_
16t
2
2t 1
2
1/816t 7
2
7/8
_
162t 1
4
0,
2.15
and the left-hand inequality in 2.14 reduces to
International Journal of Mathematics and Mathematical Sciences 7
21 2t
2
2 51 2t
21 2t
1 2t
2
0,
2.16
so two inequalities in 2.4 hold.
Lemma 2.6 see 16. Let a b, ut ta 1 tb, vt tb 1 ta. If 1/2 t
2
t
1
1 or
0 t
1
t
2
1/2, then
_
a b
2
,
a b
2
_
ut
2
, vt
2
ut
1
, vt
1
a, b. 2.17
3. Proof of Main Result
Proof of Theorem I. Let a, b R
2
.
1 For
D
SHSA
a, b
1
3
M
SH
a, b M
SA
a, b
a b
2
2ab
3a b
2
3
_
a
2
b
2
2
,
3.1
we have
D
SHSA
a, b
a
1
2
2b
2
3a b
2
2
3
a
_
2a
2
b
2
,
D
SHSA
a, b
b
1
2
2a
2
3a b
2
2
3
b
_
2a
2
b
2
,
3.2
whence
: lna lnb
_
a
D
SHSA
a, b
a
b
D
SHSA
a, b
b
_
a blna lnb
_
1
2
2ab
3a b
2
2
3
a b
_
2a
2
b
2
_
.
3.3
From 2.3 we have
1
2
2ab
3a b
2
2
3
a b
_
2a
2
b
2
0,
3.4
which implies 0 and, by Lemma 2.4, it follows that D
SHSA
is Schur-geometrically convex
in R
2
.
2 For
D
AHSH
a, b
1
2
M
AH
a, b
1
3
M
SH
a, b
a b
4
ab
3a b
1
3
_
a
2
b
2
2
.
3.5
8 International Journal of Mathematics and Mathematical Sciences
To prove that the function D
AHSH
is Schur-geometrically convex in R
2
it is enough to
notice that D
AHSH
a, b 1/2D
SHSA
a, b.
3 For
D
SGAH
a, b M
SG
a, b M
AH
a, b
_
a
2
b
2
2
_
ab
a b
2
2ab
a b
,
3.6
we have
D
SGAH
a, b
a
a
_
2a
2
b
2
b
2
ab
1
2
2b
2
a b
2
,
D
SGAH
a, b
b
b
_
2a
2
b
2
a
2
ab
1
2
2a
2
a b
2
,
3.7
and then
: lna lnb
_
a
D
SHSA
a, b
a
b
D
SHSA
a, b
b
_
a blna lnb
_
a b
_
2a
2
b
2
1
2
2ab
a b
2
_
.
3.8
From 2.2 we have 0, so by Lemma 2.4, it follows that D
SHSA
is Schur-
geometrically convex in R
2
.
4 For
D
AGSG
a, b M
AG
a, b
1
2
M
SG
a, b
1
2
_
_
a b
_
ab
_
a
2
b
2
2
_
_
, 3.9
we have
D
AGSG
a, b
a
1
2
_
1
b
2
ab
a
_
2a
2
b
2
_
,
D
AGSG
a, b
b
1
2
_
1
a
2
ab
b
_
2a
2
b
2
_
,
3.10
and then
: lna lnb
_
a
D
SHSA
a, b
a
b
D
SHSA
a, b
b
_
a blna lnb
_
1
a b
_
2a
2
b
2
_
.
3.11
International Journal of Mathematics and Mathematical Sciences 9
By 2.2 we infer that
1
a b
_
2a
2
b
2
0,
3.12
so 0. By Lemma 2.4, we get that D
AGSG
is Schur-geometrically convex in R
2
.
5 For
D
N
2
N
1
AH
a, b M
N
2
N
1
a, b
1
8
M
AH
a, b
b
2
_
_
_
_
a b
2
_
_
1
4
a b
1
2
_
ab
1
8
_
a b
2
2ab
a b
_
,
3.13
we have
D
N
2
N
1
AH
a, b
a
1
4
a
_
a b
2
1
4
_
b
2
_
_
a b
2
_
1/2
1
4
b
4
ab
1
8
_
1
2
2b
2
a b
2
_
,
D
N
2
N
1
AH
a, b
b
1
4
b
_
a b
2
1
4
_
b
2
_
_
a b
2
_
1/2
1
4
a
4
ab
1
8
_
1
2
2a
2
a b
2
_
,
3.14
and then
lna lnb
_
a
D
N
2
N
1
AH
a, b
a
b
D
N
2
N
1
AH
a, b
b
_
1
4
a blna lnb
_
_
_
a b
2
_
b
_
a b
5
4
ab
a b
2
_
_
_.
3.15
From 2.4 we have
a b
2
_
b
_
a b
5
4
ab
a b
2
0,
3.16
so 0; it follows that D
N
2
N
1
AH
is Schur-geometrically convex in R
2
.
10 International Journal of Mathematics and Mathematical Sciences
6 For
D
N
2
GN
2
N
1
a, b
1
3
M
N
2
G
a, b M
N
2
N
1
a, b
a b
4
ab
6
2
3
_
b
2
_
_
_
_
a b
2
_
_
,
3.17
we have
D
N
2
GN
2
N
1
a, b
a
1
4
b
12
ab
1
6
a
_
a b
2
1
6
_
b
2
_
_
a b
2
_
1/2
,
D
N
2
GN
2
N
1
a, b
b
1
4
a
12
ab
1
6
b
_
a b
2
1
6
_
b
2
_
_
a b
2
_
1/2
,
3.18
and then
lna lnb
_
a
D
N
2
GN
2
N
1
a, b
a
b
D
N
2
GN
2
N
1
a, b
b
_
lna lnb
_
_
_
1
4
a b
b
6
_
a b
2
a b
_
b
_
12
_
a b
2
_
1/2
_
_
_
1
6
a blna lnb
_
_
_
3
2
a b
2
_
b
_
a b
_
_
_.
3.19
By 2.4 we infer that 0, which proves that D
N
2
GN
2
N
1
is Schur-geometrically convex in
R
2
.
7 For
D
AGN
2
G
a, b
1
4
M
AG
a, b
1
3
M
N
2
G
a, b
a b
8
1
12
_
ab
1
3
_
b
2
_
_
_
_
a b
2
_
_
,
3.20
we have
D
AGN
2
G
a, b
a
1
8
b
24
ab
a b
12
2a
b
12
_
2a b
,
D
AGN
2
G
a, b
b
1
8
a
24
ab
a b
12
2b
b
12
_
2a b
,
3.21
International Journal of Mathematics and Mathematical Sciences 11
and then
lna lnb
_
a
D
AGN
2
G
a, b
a
b
D
AGN
2
G
a, b
b
_
lna lnb
_
_
_
a b
8
a b
_
b
_
12
2
a b
_
b
_
12
_
2a b
_
_
_
a blna lnb
8
_
_
_1
2
3
_
_
_
a b
2
_
b
_
a b
_
_
_
_
_
_.
3.22
From 2.4 we have 0, and, consequently, by Lemma 2.4, we obtain that D
AGN
2
G
is
Schur-geometrically convex in R
2
.
8 In order to prove that the function D
AN
2
AG
a, b is Schur-geometrically convex in
R
2
a b
2
1
5
_
a
2
b
2
2
1
5
_
a
_
b
_
_
2a b,
3.24
we have
D
SN
2
SA
a, b
a
1
2
a
5
_
2a
2
b
2
1
5
_
a b
2a
b
5
_
2a b
,
D
SN
2
SA
a, b
b
1
2
b
5
_
2a
2
b
2
1
5
_
a b
2b
b
5
_
2a b
,
3.25
12 International Journal of Mathematics and Mathematical Sciences
and then
lna lnb
_
D
SN
2
SA
a, b
a
D
SN
2
SA
a, b
b
_
lna lnb
_
_
_
a b
2
a
2
b
2
5
_
2a
2
b
2
1
5
_
_
_
aa b
2
_
ba b
2
_
_
b
_
a b
5
_
2a b
_
_
_
a blna lnb
5
2
_
5
2
a b
a
2
b
2
a b
a b
_
.
3.26
From 2.2 and 2.4 we obtain that
5
2
a b
a
2
b
2
a b
a b
2
3
2
0, 3.27
so 0, which proves that the function D
SN
2
SA
a, b is Schur-geometrically convex in R
2
.
10 One can easily check that
D
AN
A
N
2
SN
2
a, b 4D
SN
2
SA
a, b, 3.28
and, consequently, the function D
AN
2
SN
2
is Schur-geometrically convex in R
2
.
11 To prove that the function
D
SN
1
SH
a, b 2M
SN
1
a, b M
SH
a, b
_
a
2
b
2
2
a b
2
_
ab
2ab
a b
3.29
is Schur-geometrically convex in R
2
1
2
_
_
a b
_
ab
_
a
2
b
2
2
_
_
,
3.31
International Journal of Mathematics and Mathematical Sciences 13
we have
D
SGSN
1
a, b
a
1
2
_
1
b
2
ab
a
_
2a
2
b
2
_
,
D
SGSN
1
a, b
b
1
2
_
1
a
2
ab
b
_
2a
2
b
2
_
,
3.32
and then
lna lnb
_
a
D
SGSN
1
a, b
a
b
D
SGSN
1
a, b
b
_
a blna lnb
2
_
1
a b
_
2a
2
b
2
_
.
3.33
By the inequality 2.2 we get that 0, which proves that D
SGSN
1
is Schur-
geometrically convex in R
2
.
13 It is easy to check that
D
SN
3
SA
a, b
1
2
D
AGSG
a, b, 3.34
which means that the function D
SN
3
SA
is Schur-geometrically convex in R
2
.
14 To prove that the function D
SN
1
SN
3
is Schur-geometrically convex in R
2
it is
enough to notice that
D
SN
1
SN
3
a, b
1
6
D
AGSG
a, b. 3.35
The proof of Theorem I is complete.
4. Applications
Applying Theorem I, Lemma 2.6, and Denition 2.2 one can easily prove the following.
14 International Journal of Mathematics and Mathematical Sciences
Theorem II. Let 0 < a b. 1/2 t 1 or 0 t 1/2, u a
t
b
1t
and v b
t
a
1t
. Then
M
SA
a, b
1
3
M
SH
a, b
_
1
3
M
SH
u, v M
SA
u, v
_
1
3
M
SH
a, b
1
2
M
AH
a, b
_
1
2
M
AH
u, v
1
3
M
SH
u, v
_
1
2
M
AH
a, b
1
2
M
SG
a, b
_
1
2
M
SG
u, v
1
2
M
AH
u, v
_
1
2
M
SG
a, b
M
AG
a, b
_
M
AG
u, v
1
2
M
SG
u, v
_
M
AG
a, b,
4.1
1
8
M
AH
a, b M
N
2
N
1
a, b
_
M
N
2
N
1
u, v
1
8
M
AH
u, v
_
M
N
2
N
1
a, b
1
3
M
N
2
G
a, b
_
1
3
M
N
2
G
u, v M
N
2
N
1
u, v
_
1
3
M
N
2
G
a, b
1
4
M
AG
a, b
_
1
4
M
AG
u, v
1
3
M
N
2
G
u, v
_
1
4
M
AG
a, b
M
AN
2
a, b
_
M
AN
2
u, v
1
4
M
AG
u, v
_
M
AN
2
a, b,
4.2
M
SA
a, b
4
5
M
SN
2
a, b
_
4
5
M
SN
2
u, v
4
5
M
SN
2
u, v
_
4
5
M
SN
2
a, b
4M
AN
2
a, b
_
4M
AN
2
u, v
4
5
M
SN
2
u, v
_
4M
AN
2
a, b,
4.3
M
SH
a, b 2M
SN
1
a, b 2M
SN
1
u, v M
SH
u, v 2M
SN
1
a, b
3
2
M
SG
a, b
_
3
2
M
SG
u, v
3
2
M
SG
u, v
_
3
2
M
SG
a, b,
4.4
M
SA
a, b
3
4
M
SN
3
a, b
_
3
4
M
SN
3
u, v M
SA
u, v
_
3
4
M
SN
3
a, b
2
3
M
SN
1
a, b
_
2
3
M
SN
1
u, v
3
4
M
SN
3
u, v
_
2
3
M
SN
1
a, b.
4.5
Remark 4.1. Equation 4.1, 4.2, 4.3, 4.4, and 4.5 are a renement of 1.5, 1.6, 1.7,
1.8, and 1.9, respectively.
Acknowledgments
The authors are grateful to the referees for their helpful comments and suggestions. The
rst author was supported in part by the Scientic Research Common Program of Beijing
Municipal Commission of Education KM201011417013.
International Journal of Mathematics and Mathematical Sciences 15
References
1 I. J. Taneja, Renement of inequalities among means, Journal of Combinatorics, Information & System
Sciences, vol. 31, no. 14, pp. 343364, 2006.
2 I. J. Taneja, On a Dierence of Jensen Inequality and its Applications to Mean Divergence Measures,
RGMIA Research Report Collection, vol. 7, article 16, no. 4, 2004, http://rgmia.vu.edu.au/.
3 I. J. Taneja, On symmetric and nonsymmetric divergence measures and their generalizations,
Advances in Imaging and Electron Physics, vol. 138, pp. 177250, 2005.
4 A. W. Marshall and I. Olkin, Inequalities: Theory of Majorization and Its Applications, vol. 143 of
Mathematics in Science and Engineering, Academic Press, New York, NY, USA, 1979.
5 X. Zhang and Y. Chu, The Schur geometrical convexity of integral arithmetic mean, International
Journal of Pure and Applied Mathematics, vol. 41, no. 7, pp. 919925, 2007.
6 K. Guan, Schur-convexity of the complete symmetric function, Mathematical Inequalities &
Applications, vol. 9, no. 4, pp. 567576, 2006.
7 K. Guan, Some properties of a class of symmetric functions, Journal of Mathematical Analysis and
Applications, vol. 336, no. 1, pp. 7080, 2007.
8 C. Stepniak, An eective characterization of Schur-convex functions with applications, Journal of
Convex Analysis, vol. 14, no. 1, pp. 103108, 2007.
9 H.-N. Shi, Schur-convex functions related to Hadamard-type inequalities, Journal of Mathematical
Inequalities, vol. 1, no. 1, pp. 127136, 2007.
10 H.-N. Shi, D.-M. Li, and C. Gu, The Schur-convexity of the mean of a convex function, Applied
Mathematics Letters, vol. 22, no. 6, pp. 932937, 2009.
11 Y. Chu and X. Zhang, Necessary and sucient conditions such that extended mean values are Schur-
convex or Schur-concave, Journal of Mathematics of Kyoto University, vol. 48, no. 1, pp. 229238, 2008.
12 N. Elezovi c and J. Pe cari c, A note on Schur-convex functions, The Rocky Mountain Journal of
Mathematics, vol. 30, no. 3, pp. 853856, 2000.
13 J. S andor, The Schur-convexity of Stolarsky and Gini means, Banach Journal of Mathematical Analysis,
vol. 1, no. 2, pp. 212215, 2007.
14 H.-N. Shi, S.-H. Wu, and F. Qi, An alternative note on the Schur-convexity of the extended mean
values, Mathematical Inequalities & Applications, vol. 9, no. 2, pp. 219224, 2006.
15 X. M. Zhang, Geometrically Convex Functions, Anhui University Press, Hefei, China, 2004.
16 H.-N. Shi, Y.-M. Jiang, and W.-D. Jiang, Schur-convexity and Schur-geometrically concavity of Gini
means, Computers & Mathematics with Applications, vol. 57, no. 2, pp. 266274, 2009.
17 Y. Chu, X. Zhang, and G. Wang, The Schur geometrical convexity of the extended mean values,
Journal of Convex Analysis, vol. 15, no. 4, pp. 707718, 2008.
18 K. Guan, A class of symmetric functions for multiplicatively convex function, Mathematical
Inequalities & Applications, vol. 10, no. 4, pp. 745753, 2007.
19 H.-N. Shi, M. Bencze, S.-H. Wu, and D.-M. Li, Schur convexity of generalized Heronian means
involving two parameters, Journal of Inequalities and Applications, vol. 2008, Article ID879273, 9 pages,
2008.
20 B. Y. Wang, Foundations of Majorization Inequalities, Beijing Normal University Press, Beijing, China,
1990.
Hindawi Publishing Corporation
International Journal of Mathematics and Mathematical Sciences
Volume 2012, Article ID 730962, 13 pages
doi:10.1155/2012/730962
Research Article
Complete Moment Convergence of Weighted Sums
for Arrays of Rowwise -Mixing Random Variables
Ming Le Guo
School of Mathematics and Computer Science, Anhui Normal University, Wuhu 241003, China
Correspondence should be addressed to Ming Le Guo, mleguo@163.com
Received 5 June 2012; Accepted 20 August 2012
Academic Editor: Mowaaq Hajja
Copyright q 2012 Ming Le Guo. This is an open access article distributed under the Creative
Commons Attribution License, which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
The complete moment convergence of weighted sums for arrays of rowwise -mixing random
variables is investigated. By using moment inequality and truncation method, the sucient
conditions for complete moment convergence of weighted sums for arrays of rowwise -mixing
random variables are obtained. The results of Ahmed et al. 2002 are complemented. As an
application, the complete moment convergence of moving average processes based on a -mixing
random sequence is obtained, which improves the result of Kim et al. 2008.
1. Introduction
Hsu and Robbins 1 introduced the concept of complete convergence of {X
n
}. A sequence
{X
n
, n 1, 2, . . .} is said to converge completely to a constant C if
n1
P|X
n
C| > < , > 0. 1.1
Moreover, they proved that the sequence of arithmetic means of independent identically
distributed i.i.d. randomvariables converge completely to the expected value if the variance
of the summands is nite. The converse theorem was proved by Erd os 2. This result has
been generalized and extended in several directions, see Baum and Katz 3, Chow 4, Gut
5, Taylor et al. 6, and Cai and Xu 7. In particular, Ahmed et al. 8 obtained the following
result in Banach space.
2 International Journal of Mathematics and Mathematical Sciences
Theorem A. Let {X
ni
; i 1, n 1} be an array of rowwise independent random elements in a
separable real Banach space B, . Let PX
ni
> x CP|X| > x for some random variable X,
constant C and all n, i and x > 0. Suppose that {a
ni
, i 1, n 1} is an array of constants such that
sup
i1
|a
ni
| O
_
n
r
_
, for some r > 0,
i1
|a
ni
| On
, for some 0, r.
1.2
Let be such that
/
1 and x > 0 such that 1 /r < 2. Denote s max1
1/r, . If E|X|
s
< and S
n
i1
a
ni
X
ni
0 in probability, then
n1
n
PS
n
> <
for all > 0.
Chow 4 established the following renement which is a complete moment
convergence result for sums of i.i.d. random variables.
Theorem B. Let EX
1
0, 1 p < 2 and r p. Suppose that E|X
1
|
r
|X
1
| log1 |X
1
| < .
Then
n1
n
r/p21/p
E
__
_
_
_
_
n
i1
X
i
_
_
_
_
_
n
1/p
_
km
, PA > 0
_
0, as m ,
1.4
where
k
j
{X
i
; j i k}, 1 j k .
Denition 1.2. A sequence {X
n
, n 1} of random variables is said to be stochastically
dominated by a random variable X write {X
i
} X if there exists a constant C, such that
P{|X
n
| > x} CP{|X| > x} for all x 0 and n 1.
International Journal of Mathematics and Mathematical Sciences 3
The following lemma is a well-known result.
Lemma 1.3. Let the sequence {X
n
, n 1} of random variables be stochastically dominated by a
random variable X. Then for any p > 0, x > 0
E|X
n
|
p
I|X
n
| x C
_
E|X|
p
I|X| x x
p
P{|X| > x}
_
, 1.5
E|X
n
|
p
I|X
n
| > x CE|X|
p
I|X| > x. 1.6
Denition 1.4. Areal-valued function lx, positive and measurable on A, for some A > 0,
is said to be slowly varying if lim
x
lx/lx 1 for each > 0.
By the properties of slowly varying function, we can easily prove the following lemma.
Here we omit the details of the proof.
Lemma 1.5. Let lx > 0 be a slowly varying function as x , then there exists C (depends only
on r) such that
i Ck
r1
lk
k
n1
n
r
ln Ck
r1
lk for any r > 1 and positive integer k,
ii Ck
r1
lk
nk
n
r
ln Ck
r1
lk for any r < 1 and positive integer k.
The following lemma will play an important role in the proof of our main results. The
proof is due to Shao 10.
Lemma 1.6. Let {X
i
, 1 i n} be a sequence of -mixing random variables with mean zero. Suppose
that there exists a sequence {C
n
} of positive numbers such that E
km
ik1
X
i
2
C
n
for any k 0, n
1, m n. Then for any q 2, there exists C Cq, such that
Emax
1jn
_
_
_
_
_
_
kj
ik1
X
i
_
_
_
_
_
_
q
C
_
C
q/2
n
E max
k1ikn
|X
i
|
q
_
. 1.7
Lemma 1.7. Let {X
i
, 1 i n} be a sequence of -mixing random variables with
i1
1/2
i < ,
then there exists C such that for any k 0 and n 1
E
_
kn
ik1
X
i
_
2
C
kn
ik1
EX
2
i
.
1.8
4 International Journal of Mathematics and Mathematical Sciences
Proof. By Lemma 5.4.4 in 11 and H olders inequality, we have
E
_
kn
ik1
X
i
_
2
kn
ik1
EX
2
i
2
k1i<jkn
EX
i
X
j
kn
ik1
EX
2
i
4
k1i<jkn
1/2
_
j i
_
_
EX
2
i
_
1/2
_
EX
2
j
_
1/2
kn
ik1
EX
2
i
2
kn1
ik1
kn
ji1
1/2
_
j i
_
_
EX
2
i
EX
2
j
_
_
1 4
n
i1
1/2
i
_
kn
ik1
EX
2
i
.
1.9
Therefore, 1.8 holds.
2. Main Results
Now we state our main results. The proofs will be given in Section 3.
Theorem 2.1. Let {X
ni
, i 1, n 1} be an array of rowwise -mixing random variables with EX
ni
0, {X
ni
} X and
m1
1/2
m < . Let lx > 0 be a slowing varying function, and {a
ni
, i 1, n
1} be an array of constants such that
sup
i1
|a
ni
| O
_
n
r
_
, for some r > 0,
i1
|a
ni
| On
, for some 0, r.
2.1
a If 1 > 0 and there exists some > 0 such that /r 1 < 2, and s
max1 1/r, , then E|X|
s
l|X|
1/r
< implies
n1
n
lnE
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
_
n1
n
1
lnE
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
_
n1
n
lnE
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
_
n1
n
ln
i1
|a
ni
|E|X
ni
|
n1
n
ln
C
n1
n
lnE|X|
n1
n
ln < .
2.4
Remark 2.3. Note that
>
n1
n
lnE
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
_
n1
n
ln
_
0
P
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
> x
_
dx
n1
n
lnP
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
> x
_
dx.
2.5
Therefore, from2.5, we obtain that the complete moment convergence implies the complete
convergence, that is, under the conditions of Theorem 2.1, result 2.2 implies
n1
n
lnP
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
>
_
< , 2.6
and 2.3 implies
n1
n
1
lnP
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
>
_
< . 2.7
Corollary 2.4. Under the conditions of Theorem 2.1,
1 if 1 > 0 and there exists some > 0 such that /r 1 < 2, and s
max1 1/r, , then E|X|
s
l|X|
1/r
< implies
n1
n
lnE
__
_
_
_
_
i1
a
ni
X
ni
_
_
_
_
_
_
n1
n
1
lnE
__
_
_
_
_
i1
a
ni
X
ni
_
_
_
_
_
_
m1
1/2
m < . Suppose that lx > 0 is a slowly varying function.
6 International Journal of Mathematics and Mathematical Sciences
1 Let p > 1 and 1 t < 2. If E|X|
pt
l|X|
t
< , then
n1
n
p21/t
lnE
_
max
1kn
_
_
_
_
_
k
i1
X
ni
_
_
_
_
_
n
1/t
_
n1
n
11/t
lnE
_
max
1kn
_
_
_
_
_
k
i1
X
ni
_
_
_
_
_
n
1/t
_
i
a
in
Y
i
, n 1, where {a
i
, < i < } is a sequence of
real numbers with
|a
i
| < , and {Y
i
, < i < } is a sequence of -mixing random variables
with EY
i
0, {Y
i
} Y and
m1
1/2
m < . Let lx be a slowly varying function.
1 Let 1 t < 2, r 1 t/2. If E|Y|
r
l|Y|
t
< , then
n1
n
r/t21/t
lnE
__
_
_
_
_
n
i1
X
i
_
_
_
_
_
n
1/t
_
n1
n
11/t
lnE
__
_
_
_
_
n
i1
X
i
_
_
_
_
_
n
1/t
_
i1
|a
ni
| n
. 3.1
International Journal of Mathematics and Mathematical Sciences 7
Let S
nk
x
k
i1
a
ni
X
ni
I|a
ni
X
ni
| n
r
x for any k 1, n 1, and x 0. First note that
E|X|
s
l|X|
1/r
< implies E|X|
t
< for any 0 < t < s. Therefore, for x > n
r
,
x
1
n
r
sup
k1
E|S
nk
x| x
1
n
r
sup
k1
E
_
_
_
_
_
k
i1
a
ni
X
ni
I
_
|a
ni
X
ni
| > n
r
x
_
_
_
_
_
_
EX
ni
0
i1
E|a
ni
X
ni
|I
_
|a
ni
X
ni
| > n
r
x
_
i1
E|a
ni
X|I
_
|a
ni
X| > n
r
x
_
i1
|a
ni
|E|X|I|X| > x n
E|X|I|X| > x
x
/r
E|X|I|X| > x E|X|
1/r
I|X| > n
r
0 as n .
3.2
Hence, for n large enough we have sup
k1
E|S
nk
x| < /2n
r
x. Then
n1
n
lnE
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
_
n1
n
ln
_
P
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
x
_
dx
n1
n
r
ln
_
n
r
P
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
n
r
x
_
dx
C
n1
n
r
ln
_
n
r
P
_
sup
i
|a
ni
X
ni
| > n
r
x
_
dx
C
n1
n
r
ln
_
n
r
P
_
sup
k1
|S
nk
x ES
nk
x| n
r
x
2
_
dx : I
1
I
2
.
3.3
Noting that > 1, by Lemma 1.5, Markov inequality, 1.6, and 3.1, we have
I
1
C
n1
n
r
ln
_
n
r
i1
P
_
|a
ni
X
ni
| > n
r
x
_
dx
C
n1
n
r
ln
_
n
r
n
r
x
1
i1
E|a
ni
X
ni
|I
_
|a
ni
X
ni
| > n
r
x
_
dx
C
n1
n
ln
_
n
r
x
1
E|X|I|X| > xdx
C
n1
n
ln
kn
_
k
r1
k
r
x
1
E|X|I|X| > xdx
8 International Journal of Mathematics and Mathematical Sciences
C
n1
n
ln
kn
k
1
E|X|I|X| > k
r
C
k1
k
1
E|X|I|X| > k
r
n1
n
ln
C
k1
k
lkE|X|I|X| > k
r
CE|X|
11/r
l
_
|X|
1/r
_
< .
3.4
Now we estimate I
2
, noting that
m1
1/2
m < , by Lemma 1.7, we have
sup
1m<
E
_
m
i1
a
ni
X
ni
I
_
|a
ni
X
ni
| n
r
x
_
E
m
i1
a
ni
X
ni
I
_
|a
ni
X
ni
| n
r
x
_
_
2
C
i1
Ea
2
ni
X
2
ni
I
_
|a
ni
X
ni
| n
r
x
_
.
3.5
By Lemma 1.6, Markov inequality, C
r
inequality, and 1.5, for any q 2, we have
P
_
sup
k1
|S
nk
x ES
nk
x| n
r
x
2
_
Cn
rq
x
q
Esup
k1
|S
nk
x ES
nk
x|
q
Cn
rq
x
q
_
_
_
i1
Ea
2
ni
X
2
ni
I
_
|a
ni
X
ni
| n
r
x
_
_
q/2
i1
E|a
ni
X
ni
|
q
I
_
|a
ni
X
ni
| n
r
x
_
_
_
Cn
rq
x
q
_
i1
Ea
2
ni
X
2
I
_
|a
ni
X| n
r
x
_
_
q/2
Cn
rq
x
q
i1
E|a
ni
X|
q
I
_
|a
ni
X| n
r
x
_
C
_
i1
P
_
|a
ni
X| > n
r
x
_
_
q/2
C
i1
P
_
|a
ni
X| > n
r
x
_
: J
1
J
2
J
3
J
4
.
3.6
So,
I
2
n1
n
r
ln
_
n
r
J
1
J
2
J
3
J
4
dx. 3.7
From 3.4, we have
n1
n
r
ln
_
n
r
J
4
dx < .
For J
1
, we consider the following two cases.
International Journal of Mathematics and Mathematical Sciences 9
If s > 2, then EX
2
< . Taking q 2 such that q r/2 < 1, we have
n1
n
r
ln
_
n
r
J
1
dx
C
n1
n
rrq
ln
_
n
r
x
q
_
i1
a
2
ni
_
q/2
dx
C
n1
n
rrq
lnn
qr/2
n
rq1
C
n1
n
qr/2
ln < .
3.8
If s 2, we choose s
< 1, we have
n1
n
r
ln
_
n
r
J
1
dx
C
n1
n
rrq
ln
_
n
r
x
q
_
i1
|a
ni
||a
ni
|
s
1
E|a
ni
X|
2s
|X|
s
I
_
|a
ni
X| n
r
x
_
_
q/2
dx
C
n1
n
rrq
lnn
q/2
n
qr/2s
1
_
n
r
x
q
_
n
r
x
_
q/22s
dx
C
n1
n
qr/21/rs
ln < .
3.9
So,
n1
n
r
ln
_
n
r
J
1
dx < .
Now, we estimate J
2
. Set I
nj
{i 1 | nj 1
r
< |a
ni
| nj
r
}, j 1, 2, . . .. Then
j1
I
nj
N, where N is the set of positive integers. Note also that for all k 1, n 1,
n
i1
|a
ni
|
j1
iI
nj
|a
ni
|
j1
_
I
nj
__
n
_
j 1
__
r
n
r
jk
_
I
nj
__
j 1
_
rq
k 1
rqr
.
3.10
Hence, we have
jk
_
I
nj
_
j
rq
Cn
r
k
rrq
.
3.11
10 International Journal of Mathematics and Mathematical Sciences
Note that
n1
n
r
ln
_
n
r
J
2
dx
C
n1
n
rrq
ln
_
n
r
x
q
j1
iI
nj
E|a
ni
X|
q
I
_
|a
ni
X| n
r
x
_
dx
C
n1
n
rrq
ln
j1
_
I
nj
__
nj
_
rq
kn
_
k1
r
k
r
x
q
E|X|
q
I
_
|X| x
_
j 1
_
r
_
dx
C
n1
n
rrq
ln
j1
_
I
nj
__
nj
_
rq
kn
k
rq11
E|X|
q
I
_
|X| k 1
r
_
j 1
_
r
_
C
n1
n
r
ln
kn
k
rq11
j1
_
I
nj
_
j
rq
k1j11
i0
E|X|
q
I
_
i
r
< |X| i 1
r
_
C
n1
n
r
ln
kn
k
rq11
j1
_
I
nj
_
j
rq
2k11
i0
E|X|
q
I
_
i
r
< |X| i 1
r
_
C
n1
n
r
ln
kn
k
rq11
j1
_
I
nj
_
j
rq
k1j1
i2k1
E|X|
q
I
_
i
r
< |X| i 1
r
_
: J
2
J
2
.
3.12
Taking q 2 large enough such that rq r < 1, for J
2
, by Lemma 1.6 and 3.11, we
get
J
2
C
n1
n
r
ln
kn
k
rq11
n
r
2k11
i0
E|X|
q
I
_
i
r
< |X| i 1
r
_
C
k1
k
rq11
2k11
i0
E|X|
q
I
_
i
r
< |X| i 1
r
_
k
n1
n
ln
C
k1
k
rqr
lk
2k11
i0
E|X|
q
I
_
i
r
< |X| i 1
r
_
C C
i3
E|X|
q
I
_
i
r
< |X| i 1
r
_
ki/2
k
rqr
lk
C C
i3
i
rqr1
liE|X|
q
I
_
i
r
< |X| i 1
r
_
C CE|X|
11/r
l
_
|X|
1/r
_
< .
3.13
International Journal of Mathematics and Mathematical Sciences 11
For J
2
, we obtain
J
2
C
n1
n
r
ln
kn
k
rq11
j1
_
I
nj
_
j
rq
j1k1
i2k1
E|X|
q
I
_
i
r
< |X| i 1
r
_
C
n1
n
r
ln
kn
k
rq11
i2k1
E|X|
q
I
_
i
r
< |X| i 1
r
_
jik1
1
1
_
I
nj
_
j
rq
C
n1
n
r
ln
kn
k
rq11
i2k1
n
r
i
r1q
k
r1q
E|X|
q
I
_
i
r
< |X| i 1
r
_
C
k1
k
1
i2k1
i
r1q
E|X|
q
I
_
i
r
< |X| i 1
r
_
k
n1
n
ln
C
k1
k
lk
i2k1
i
r1q
E|X|
q
I
_
i
r
< |X| i 1
r
_
C
i4
i
1rrq
E|X|
q
I
_
i
r
< |X| i 1
r
_
CE|X|
11/r
l
_
|X|
1/r
_
< .
3.14
So
n1
n
r
ln
_
n
r
J
2
dx < . Finally, we prove
n1
n
r
ln
_
n
r
J
3
dx < . In fact, noting
1 a/r < s
n1
n
r
ln
_
n
r
J
3
dx C
n1
n
r
ln
_
n
r
_
i1
n
rs
x
s
E|a
ni
X|
s
_
q/2
dx
C
n1
n
r
lnn
qrs
/2
n
rs
1q/2
n
q/2
_
n
r
x
s
q/2
dx
C
n1
n
rrq/2q/2
lnn
rs
q/21
C
n1
n
qr/21/2s
ln < .
3.15
Thus, we complete the proof in a. Next, we prove b. Note that E|X|
1/r
< implies that
3.2 holds. Therefore, from the proof in a, to complete the proof of b, we only need to
prove
I
2
C
n1
n
1r
ln
_
n
r
P
_
sup
k1
|S
nk
x ES
nk
x| n
r
x
2
_
dx < . 3.16
12 International Journal of Mathematics and Mathematical Sciences
In fact, noting 1, 1 > 0, r < 1 and E|X|
1/r
l|X|
1/r
< . By taking q 2
in the proof of 3.12, 3.13, and 3.14, we get
C
n1
n
1r
ln
_
n
r
x
2
i1
Ea
2
ni
X
2
I
_
|a
ni
X| n
r
x
_
dx C CE|X|
1/r
l
_
|X|
1/r
_
< . 3.17
Then, by 3.17, we have
I
2
C
n1
n
1r
ln
_
n
r
n
2r
x
2
E|S
xn
ES
xn
|
2
dx
C
n1
n
1r
ln
_
n
r
x
2
i1
Ea
2
ni
X
2
ni
I
_
|a
ni
X
ni
| n
r
x
_
dx
C
n1
n
1r
ln
_
n
r
x
2
i1
Ea
2
ni
X
2
I
_
|a
ni
X| n
r
x
_
dx
C
n1
n
1r
ln
_
n
r
i1
P
_
|a
ni
X| > n
r
x
_
dx
C
n1
n
1r
ln
_
n
r
x
2
i1
Ea
2
ni
X
2
I
_
|a
ni
X| n
r
x
_
dx C < .
3.18
The proof of Theorem 2.1 is completed.
Proof of Corollary 2.4. Note that
__
_
_
_
_
i1
a
ni
X
ni
_
_
_
_
_
_
_
sup
k1
_
_
_
_
_
k
i1
a
ni
X
ni
_
_
_
_
_
_
. 3.19
Therefore, 2.8 and 2.9 hold by Theorem 2.1.
Proof of Corollary 2.5. By applying Theorem 2.1, taking p 2, a
ni
n
1/t
for 1 i n,
and a
ni
0 for i > n, then we obtain 2.10. Similarly, taking 1, a
ni
n
1/t
for 1 i n,
and a
ni
0 for i > n, we obtain 2.11 by Theorem 2.1.
Proof of Corollary 2.6. Let X
ni
Y
i
and a
ni
n
1/t
n
j1
a
ij
for all n 1, < i < .
Since
|a
i
| < , we have sup
i
|a
ni
| On
1/t
and
i
|a
ni
| On
11/t
. By applying
Corollary 2.4, taking r/t 2, r 1/t, 1 1/t, we obtain
n1
n
r/t21/t
lnE
__
_
_
_
_
n
i1
X
i
_
_
_
_
_
n
1/t
_
n1
n
lnE
__
_
_
_
_
i
a
ni
X
ni
_
_
_
_
_
_
< , > 0.
3.20
Therefore, 2.12 and 2.13 hold.
International Journal of Mathematics and Mathematical Sciences 13
Acknowledgment
The paper is supported by the National Natural Science Foundation of China no. 11271020
and 11201004, the Key Project of Chinese Ministry of Education no. 211077, the Natural
Science Foundation of Education Department of Anhui Province KJ2012ZD01, and the
Anhui Provincial Natural Science Foundation no. 10040606Q30 and 1208085MA11.
References
1 P. L. Hsu and H. Robbins, Complete convergence and the law of large numbers, Proceedings of the
National Academy of Sciences of the United States of America, vol. 33, pp. 2531, 1947.
2 P. Erd os, On a theorem of hsu and robbins, Annals of Mathematical Statistics, vol. 20, pp. 286291,
1949.
3 L. E. Baum and M. Katz, Convergence rates in the law of large numbers, Transactions of the American
Mathematical Society, vol. 120, pp. 108123, 1965.
4 Y. S. Chow, On the rate of moment convergence of sample sums and extremes, Bulletin of the Institute
of Mathematics, vol. 16, no. 3, pp. 177201, 1988.
5 A. Gut, Complete convergence and ces` aro summation for i.i.d. random variables, Probability Theory
and Related Fields, vol. 97, no. 1-2, pp. 169178, 1993.
6 R. L. Taylor, R. F. Patterson, and A. Bozorgnia, A strong law of large numbers for arrays of rowwise
negatively dependent random variables, Stochastic Analysis and Applications, vol. 20, no. 3, pp. 643
656, 2002.
7 G. H. Cai and B. Xu, Complete convergence for weighted sums of -mixing sequences and its
application, Journal of Mathematics, vol. 26, no. 4, pp. 419422, 2006.
8 S. E. Ahmed, R. G. Antonini, and A. Volodin, On the rate of complete convergence for weighted sums
of arrays of banach space valued random elements with application to moving average processes,
Statistics & Probability Letters, vol. 58, no. 2, pp. 185194, 2002.
9 T. S. Kim and M. H. Ko, Complete moment convergence of moving average processes under
dependence assumptions, Statistics & Probability Letters, vol. 78, no. 7, pp. 839846, 2008.
10 Q. M. Shao, A moment inequality and its applications, Acta Mathematica Sinica, vol. 31, no. 6, pp.
736747, 1988.
11 W. F. Stout, Almost Sure Convergence, Academic Press, New York, NY, USA, 1974.
12 P. Y. Chen, T. C. Hu, and A. Volodin, Limiting behaviour of moving average processes under -
mixing assumption, Statistics & Probability Letters, vol. 79, no. 1, pp. 105111, 2009.
Hindawi Publishing Corporation
International Journal of Mathematics and Mathematical Sciences
Volume 2012, Article ID 597490, 9 pages
doi:10.1155/2012/597490
Research Article
On Huygens Inequalities and the Theory of Means
J ozsef S andor
Department of Mathematics, Babes-Bolyai University, Strada Kog alniceanu no. 1,
400084 Cluj-Napoca, Romania
Correspondence should be addressed to J ozsef S andor, jsandor@math.ubbcluj.ro
Received 24 March 2012; Accepted 20 August 2012
Academic Editor: Mowaaq Hajja
Copyright q 2012 J ozsef S andor. This is an open access article distributed under the Creative
Commons Attribution License, which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
By using the theory of means, various renements of Huygens trigonometric and hyperbolic
inequalities will be proved. New Huygens type inequalities will be provided, too.
1. Introduction
The famous Huygens trigonometric inequality see e.g., 13 states that for all x 0, /2
one has
2 sinx tanx > 3x. 1.1
The hyperbolic version of inequality 1.1 has been established recently by Neuman
and S andor 3:
2 sinhx tanhx > 3x, for x > 0. 1.2
Let a, b > 0 be two positive real numbers. The logarithmic and identric means of a and
b are dened by
L La, b :
b a
lnb lna
for a
/
b; La, a a,
I Ia, b :
1
e
_
b
b
a
a
_
1/ba
for a
/
b; Ia, a a,
1.3
2 International Journal of Mathematics and Mathematical Sciences
respectively. Seierts mean P is dened by
P Pa, b :
a b
2 arcsina b/a b
for a
/
b, Pa, a a. 1.4
Let
A Aa, b :
a b
2
, G Ga, b
_
ab,
H Ha, b 2
_
_
1
a
1
b
_
1.5
denote the arithmetic, geometric, and harmonic means of a and b, respectively. These means
have been also in the focus of many research papers in the last decades. For a survey of
results, see, for example, 46. In what follows, we will assume a
/
b.
Now, by remarking that letting a 1 sinx, b 1 sinx, where x 0, /2, in
P, G, and A, we nd that
P
sinx
x
, G cos x, A 1, 1.6
so Huygens inequality 1.1 may be written also as
P >
3AG
2G A
3
_
_
2
A
1
G
_
HA, A, G. 1.7
Here Ha, b, c denotes the harmonic mean of the numbers a, b, c:
Ha, b, c 3
_
_
1
a
1
b
1
c
_
. 1.8
On the other hand, by letting a e
x
, b e
x
in L, G, and A, we nd that
L
sinhx
x
, G 1, A coshx,
1.9
so Huygens hyperbolic inequality 1.2 may be written also as
L >
3AG
2A G
3
_
_
2
G
1
A
_
HG, G, A. 1.10
2. First Improvements
Suppose a, b > 0, a
/
b.
International Journal of Mathematics and Mathematical Sciences 3
Theorem 2.1. One has
P > HL, A >
3AG
2G A
HA, A, G, 2.1
L > HP, G >
3AG
2A G
HG, G, A. 2.2
Proof. The inequalities P > HL, A and L > HP, G have been proved in paper 7 see
Corollary 3.2. In fact, stronger relations are valid, as we will see in what follows.
Now, the interesting fact is that the second inequality of 2.1, that is, 2LA/L A >
3AG/2GA becomes, after elementary transformations, exactly inequality 1.10, while the
second inequality of 2.2, that is, 2PG/P G > 3AG/2AG becomes inequality 1.7.
Another improvements of 1.7, respectively, 1.10 are provided by
Theorem 2.2. One has the inequalities:
P >
3
_
A
2
G >
3AG
2G A
, 2.3
L >
3
_
G
2
A >
3AG
2A G
. 2.4
Proof. The rst inequality of 2.3 is proved in 6, while the rst inequality of 2.8 is a well-
known inequality due to Leach and Sholander 8 see 4 for many related references. The
second inequalities of 2.3 and 2.4 are immediate consequences of the arithmetic-geometric
inequality applied for A, A, G and A, G, G, respectively.
Remark 2.3. By 2.3 and 1.6, we can deduce the following improvement of the Huygens
inequality 1.1:
sinx
x
>
3
cos x >
3 cos x
2 cos x 1
, x
_
0,
2
_
. 2.5
From 2.1 and 1.6, we get
sinx
x
>
2L
1
>
3 cos x
2 cos x 1
, x
_
0,
2
_
.
_
2.5
_
Similarly, by 2.4 and 1.9, we get
sinhx
x
>
3
_
coshx >
3 coshx
2 coshx 1
, x > 0.
2.6
From 2.2 and 1.9, we get
sinhx
x
>
2P
1
>
3 coshx
2 coshx 1
, x > 0.
_
2.6
_
Here, L
L1 sinx, 1 sinx, P
Pe
x
, e
x
.
4 International Journal of Mathematics and Mathematical Sciences
We note that the rst inequality of 2.5 has been discovered by Adamovi c and
Mitrinovi c see 3, while the rst inequality of 2.6 by Lazarevi c see 3.
Now, we will prove that inequalities 2.2 of Theorem 2.1 and 2.4 of Theorem 2.2
may be compared in the following way.
Theorem 2.4. One has
L >
3
_
G
2
A > HP, G >
3AG
2A G
. 2.7
Proof. We must prove the second inequality of 2.7. For this purpose, we will use the inequal-
ity see 6:
P <
2A G
3
. 2.8
This implies G/P > 3G/G 2A, so 1/21 G/P > 2G A/G 2A.
Now, we will prove that
2G A
G 2A
>
3
_
G
A
.
2.9
By letting x G/A 0, 1, inequality 2.9 becomes
2x 1
x 2
>
3
x. 2.10
Put x a
3
, where a 0, 1. After elementary transformations, inequality 2.10
becomes a 1a 1
3
< 0, which is true.
Note. The Referee suggested the following alternative proof: since P < 2A G/3 and the
harmonic mean increases in both variables, it suces to prove stronger inequality
3
A
2
G >
H2A G/3, G which can be written as 2.9.
Remark 2.5. The following renement of inequalities
_
2.6
_
is true:
sinhx
x
>
3
_
coshx >
2P
1
>
3 coshx
2 coshx 1
, x > 0.
2.11
Unfortunately, a similar renement to 2.7 for the mean P is not possible, as by numeri-
cal examples one can deduce that generally HL, A and
3
A
2
Gare not comparable. However,
in a particular case, the following result holds true.
Theorem 2.6. Assume that A/G 4. Then one has
P > HL, A >
3
_
A
2
G >
3AG
2G A
. 2.12
First, prove one the following auxiliary results.
International Journal of Mathematics and Mathematical Sciences 5
Lemma 2.7. For any x 4, one has
3
_
x 1
2
_
2
3
x 1
_
> x
3
4.
2.13
Proof. A computer computation shows that 2.13 is true for x 4. Now put x a
3
in 2.13.
By taking logarithms, the inequality becomes
fa 2 ln
_
a
3
1
2
_
9 lna 3 ln2a 1 > 0. 2.14
An easy computation implies
a2a 1
_
a
3
1
_
f
a 3a 1
_
a
2
a 3
_
. 2.15
As
3
4
2
4 3 2
3
2
3
2
2
3
3
2 1
3
a > 0 for
a
3
4.
Proof of the theorem. We will apply the inequality:
L >
3
_
G
_
A G
2
_
2
,
2.16
due to the author 9. This implies
1
2
_
1
A
L
_
<
1
2
_
1
3
_
4A
3
GA G
2
_
N. 2.17
By letting x A/G in 2.13, we can deduce
N <
3
_
A
G
.
2.18
So
1
2
_
1
A
L
_
<
3
_
A
G
.
2.19
This immediately gives HL, A >
3
A
2
G.
Remark 2.8. If cos x 1/4, x 0, /2, then
sinx
x
>
2L
1
>
3
cos x >
3 cos x
2 cos x 1
, 2.20
which is a renement, in this case, of inequality
_
2.5
_
.
6 International Journal of Mathematics and Mathematical Sciences
3. Further Improvements
Theorem 3.1. One has
P >
LA >
3
_
A
2
G >
AG
L
>
3AG
2G A
, 3.1
L >
GP >
3
_
G
2
A >
AG
P
>
3AG
2A G
. 3.2
Proof. The inequalities P >
LA and L >
>
3
cos x >
cos x
L
>
3 cos x
2 cos x 1
, x
_
0,
2
_
, 3.3
sinhx
x
>
>
3
_
coshx >
coshx
P
>
3 coshx
2 coshx 1
, x > 0,
3.4
where L
and P
_
and
_
2.5
_
.
Theorem 3.3. One has
P >
IG >
PG >
3
_
G
2
A >
3AG
2A G
. 3.6
Proof. The rst two inequalities of 3.5 one followed by the rst inequality of 3.1 and the
fact that Gx, y > Hx, y with x L, y A.
Now, the inequality HA, L > AL/I may be written also as
I >
A L
2
, 3.7
which has been proved in 4 see also 12.
Further, by Alzers inequality L
2
> GI see 13 one has
L
I
>
G
L
3.8
and by Carlsons inequality L < 2G A/3 see 11, we get
AL
I
>
AG
L
>
3AG
2G A
, 3.9
so 3.5 is proved.
International Journal of Mathematics and Mathematical Sciences 7
The rst two inequalities of 3.6 have been proved by the author in 5. Since I > P
see 14 and by 3.2, inequalities 3.6 are completely proved.
Remark 3.4. One has the following inequalities:
sinx
x
>
>
2L
1
>
L
>
cos x
L
>
3 cos x
2 cos x 1
, x
_
0,
2
_
, 3.10
where I
I1 sinx, 1 sinx;
sinhx
x
>
sinhx
x
_
e
x cothx1
1
coshx sinhx/x
_
> e
x cothx1/2
>
>
3
_
coshx >
3 coshx
2 coshx 1
.
3.11
Theorem 3.5. One has
P >
3
_
A
_
A G
2
_
2
>
_
A
_
A 2G
3
_
>
IG >
_
G
_
2A G
3
_
>
PG >
3
_
G
2
A >
3AG
2A G
.
3.13
Proof. In 3.12, we have to prove the rst three inequalities, the rest are contained in 3.5.
The rst inequality of 3.12 is proved in 6. For the second inequality, put A/G t > 1
By taking logarithms, we have to prove that
gt 4 ln
_
t 1
2
_
3 ln
_
t 2
3
_
lnt > 0. 3.14
As g
3
_
x y y >
3
_
1
x
1
y
1
y
_
3
_
1
x
2
y
_,
4.3
for x A, y A G/2 in order to deduce the second inequality of 4.1. The last two
inequalities become, after certain transformation,
A G
2
> 0. 4.4
The proof of 4.2 follows on the same lines, and we omit the details.
Theorem 4.2. For all x
_
0,
2
_
, one has
sinx 4 tan
x
2
> 3x. 4.5
For all x > 0, one has
sinhx 4 tanh
x
2
> 3x. 4.6
Proof. Apply 1.6 for P > 3AA G/5A G of 4.1.
As cos x 1 2cos
2
x/2 and sinx 2 sinx/2 cosx/2, we get inequality 4.5. A
similar argument applied to 4.6, by an application of 4.2 and the formulae coshx 1
2cosh
2
x/2 and sinhx 2 sinhx/2 coshx/2.
Remarks 4.3. By 4.1, inequality 4.5 is a renement of the classical Huygens inequality 1.1:
2 sinx tanx > sinx 4 tan
x
2
> 3x.
_
4.3
_
International Journal of Mathematics and Mathematical Sciences 9
Similarly, 4.6 is a renement of the hyperbolic Huygens inequality 1.2:
2 sinhx tanhx > sinhx 4 tanh
x
2
> 3x.
_
4.4
_
We will call 4.5 as the second Huygens inequality, while 4.6 as the second hyper-
bolic Huygens inequality.
In fact, by 4.1 and 4.2 renements of these inequalities may be stated, too.
The inequality P > A2G A/2A G gives
sinx
x
>
2 cos x 1
cos x 2
, 4.7
or written equivalently:
sinx
x
3
cos x 2
> 2, x
_
0,
2
_
. 4.8
Acknowledgments
The author is indebted to Professor Edward Neuman for his support and discussions on this
topic. He also thanks the Referee for a careful reading of the paper and a new proof of
Theorem 2.4.
References
1 C. Huygens, Oeuvres Completes 18881940, Soci ete Hollondaise des Science, Haga, Gothenburg.
2 J. S. S andor and M. Bencze, On Huygens trigonometric inequality, RGMIA Research Report
Collection, vol. 8, no. 3, article 14, 2005.
3 E. Neuman and J. S andor, On some inequalities involving trigonometric and hyperbolic functions
with emphasis on the Cusa-Huygens, Wilker, and Huygens inequalities, Mathematical Inequalities &
Applications, vol. 13, no. 4, pp. 715723, 2010.
4 J. S andor, On the identric and logarithmic means, Aequationes Mathematicae, vol. 40, no. 2-3, pp.
261270, 1990.
5 J. S andor, On renements of certain inequalities for means, Archivum Mathematicum, vol. 31, no. 4,
pp. 279282, 1995.
6 J. S andor, On certain inequalities for means. III, Archiv der Mathematik, vol. 76, no. 1, pp. 3440, 2001.
7 E. Neuman and J. S andor, On the Schwab-Borchardt mean, Mathematica Pannonica, vol. 14, no. 2,
pp. 253266, 2003.
8 E. B. Leach and M. C. Sholander, Extended mean values. II, Journal of Mathematical Analysis and
Applications, vol. 92, no. 1, pp. 207223, 1983.
9 J. S andor, On certain inequalities for means. II, Journal of Mathematical Analysis and Applications, vol.
199, no. 2, pp. 629635, 1996.
10 E. Neuman and J. S andor, On the Schwab-Borchardt mean. II, Mathematica Pannonica, vol. 17, no. 1,
pp. 4959, 2006.
11 B. C. Carlson, The logarithmic mean, The American Mathematical Monthly, vol. 79, pp. 615618, 1972.
12 J. S andor, A note on some inequalities for means, Archiv der Mathematik, vol. 56, no. 5, pp. 471473,
1991.
13 H. Alzer, Two inequalities for means, La Soci et e Royale du Canada, vol. 9, no. 1, pp. 1116, 1987.
14 H. J. Seiert, Ungleichungen f ur einen bestimmten Mittelwert, Nieuw Arch, Wisk, vol. 13, no. 42,
pp. 195198, 1995.
15 J. S andor, New renements of two inequalities for means, submitted.
Hindawi Publishing Corporation
International Journal of Mathematics and Mathematical Sciences
Volume 2012, Article ID 430692, 6 pages
doi:10.1155/2012/430692
Research Article
A Nice Separation of Some Seiffert-Type Means by
Power Means
Iulia Costin
1
and Gheorghe Toader
2
1
Department of Computer Sciences, Technical University of Cluj-Napoca, 400114 Cluj-Napoca, Romania
2
Department of Mathematics, Technical University of Cluj-Napoca, 400114 Cluj-Napoca, Romania
Correspondence should be addressed to Gheorghe Toader, gheorghe.toader@math.utcluj.ro
Received 21 March 2012; Accepted 30 April 2012
Academic Editor: Edward Neuman
Copyright q 2012 I. Costin and G. Toader. This is an open access article distributed under the
Creative Commons Attribution License, which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Seiert has dened two well-known trigonometric means denoted by P and T. In a similar
way it was dened by Carlson the logarithmic mean L as a hyperbolic mean. Neuman and
S andor completed the list of such means by another hyperbolic mean M. There are more known
inequalities between the means P, T, and L and some power means A
p
. We add to these
inequalities two new results obtaining the following nice chain of inequalities A
0
< L < A
1/2
<
P < A
1
< M < A
3/2
< T < A
2
, where the power means are evenly spaced with respect to their
order.
1. Means
A mean is a function M : R
2
_
_
qa
p
b
p
pa
q
b
q
_
1/pq
, pq
_
p q
_
/
0
1
e
p
_
a
a
p
b
b
p
_
1/a
p
b
p
, p q
/
0
_
a
p
b
p
plna lnb
_
1/p
, p
/
0, q 0
ab, p q 0.
1.8
The mean A
1
A is the arithmetic mean and the mean S
1,0
L is the logarithmic
mean. As Carlson remarked in 5, the logarithmic mean can be represented also by
La, b
a b
2 tanh
1
a b/a b
;
1.9
International Journal of Mathematics and Mathematical Sciences 3
thus the means P, T, M, and L are very similar. In 3 it is also proven that these means can
be dened using the nonsymmetric Schwab-Borchardt mean SB given by
SBa, b
_
b
2
a
2
cos
1
a/b
, if a < b
a
2
b
2
cosh
1
a/b
, if a > b
1.10
see 6, 7. It has been established in 3 that
L SBA, G, P SBG, A, T SBA, A
2
, M SBA
2
, A. 1.11
2. Interlacing Property of Power Means
Given two means M and N, we will write M < N if
Ma, b < Na, b, for a
/
b. 2.1
It is known that the family of power means is an increasing family of means, thus
A
p
< A
q
, if p < q. 2.2
Of course, it is more dicult to compare two Stolarsky means, each depending on two
parameters. To present the comparison theoremgiven in 8, 9, we have to give the denitions
of the following two auxiliary functions:
k
_
x, y
_
_
|x|
_
_
y
_
_
x y
, x
/
y
signx, x y,
l
_
x, y
_
_
L
_
x, y
_
, x > 0, y > 0
0, x 0, y 0, xy 0.
2.3
Theorem 2.1. Let p, q, r, s R. Then the comparison inequality
S
p,q
S
r,s
2.4
holds true if and only if pq r s, and (1) lp, q lr, s if 0 minp, q, r, s, (2) kp, q kr, s
if minp, q, r, s < 0 < maxp, q, r, s, or (3) lp, q lr, s if maxp, q, r, s 0.
We need also in what follows an important double-sided inequality proved in 3 for
the Schwab-Borchardt mean:
3
_
ab
2
< SBa, b <
a 2b
3
, a
/
b.
2.5
4 International Journal of Mathematics and Mathematical Sciences
Being rather complicated, the Seiert-type means were evaluated by simpler means,
rst of all by power means. The evaluation of a given mean M by power means assumes the
determination of some real indices p and q such that A
p
< M < A
q
. The evaluation is optimal
if p is the the greatest and q is the smallest index with this property. This means that M cannot
be compared with A
r
if p < r < q.
For the logarithmic mean in 10, it was determined the optimal evaluation
A
0
< L < A
1/3
. 2.6
For the Seiert means, there are known the evaluations
A
1/3
< P < A
2/3
, 2.7
proved in 11 and
A
1
< T < A
2
, 2.8
given in 2. It is also known that
A
1
< M < T, 2.9
as it was shown in 3. Moreover in 12 it was determined the optimal evaluation
A
ln2/ln
< P < A
2/3
. 2.10
Using these results we deduce the following chain of inequalities:
A
0
< L < A
1/2
< P < A
1
< M < T < A
2
. 2.11
To prove the full interlacing property of power means, our aimis to showthat A
3/2
can be put
between M and T. We thus obtain a nice separation of these Seiert-type means by power
means which are evenly spaced with respect to their order.
3. Main Results
We add to the inequalities 2.11 the next results.
Theorem 3.1. The following inequalities
M < A
3/2
< T 3.1
are satised.
International Journal of Mathematics and Mathematical Sciences 5
Proof. First of all, let us remark that A
3/2
S
3,3/2
. So, for the rst inequality in 3.1, it is
sucient to prove that the following chain of inequalities
M <
A
2
2A
3
< S
3,1
< S
3,3/2
3.2
is valid. The rst inequality in 3.2 is a simple consequence of the property of the mean M
given in 1.11 and the second inequality from 2.5. The second inequality can be proved by
direct computation or by taking a 1 t, b 1 t, 0 < t < 1 which gives
1 t
2
2
3
<
_
3 t
2
3
,
3.3
which is easy to prove. The last inequality in 3.2 is given by the comparison theorem of the
Stolarsky means. In a similar way, the second inequality in 3.1 is given by the relations
S
3,3/2
< S
4,1
3
_
AA
2
2
< T.
3.4
The rst inequality is again given by the comparison theorem of the Stolarsky means. The
equality in 3.4 is shown by elementary computations, and the last inequality is a simple
consequence of the property of the mean T given in 1.11 and the rst inequality from 2.5.
Corollary 3.2. The following two-sided inequality
x
sinh
1
x
< A
3/2
1 x, 1 x <
x
tan
1
x
,
3.5
is valid for all 0 < x < 1.
Acknowledgment
The authors wish to thank the anonymous referee for oering them a simpler proof for their
results.
References
1 H.-J. Seiert, Problem 887, Nieuw Archief voor Wiskunde, vol. 11, no. 2, p. 176, 1993.
2 H.-J. Seiert, Aufgabe 16, Die Wurzel, vol. 29, pp. 221222, 1995.
3 E. Neuman and J. S andor, On the Schwab-Borchardt mean, Mathematica Pannonica, vol. 14, no. 2,
pp. 253266, 2003.
4 K. B. Stolarsky, Generalizations of the logarithmic mean, Mathematics Magazine, vol. 48, no. 2, pp.
8792, 1975.
5 B. C. Carlson, The logarithmic mean, The American Mathematical Monthly, vol. 79, pp. 615618, 1972.
6 J. M. Borwein and P. B. Borwein, Pi and the AGM: A Study in Analytic Number Theory and Computational
Complexity, John Wiley & Sons, New York, NY, USA, 1987.
6 International Journal of Mathematics and Mathematical Sciences
7 B. C. Carlson, Algorithms involving arithmetic and geometric means, The American Mathematical
Monthly, vol. 78, pp. 496505, 1971.
8 Zs. P ales, Inequalities for dierences of powers, Journal of Mathematical Analysis and Applications,
vol. 131, no. 1, pp. 271281, 1988.
9 E. B. Leach and M. C. Sholander, Multi-variable extended mean values, Journal of Mathematical
Analysis and Applications, vol. 104, no. 2, pp. 390407, 1984.
10 T. P. Lin, The power mean and the logarithmic mean, The American Mathematical Monthly, vol. 81,
pp. 879883, 1974.
11 A. A. Jagers, Solution of problem 887, Nieuw Archief voor Wiskunde, vol. 12, pp. 230231, 1994.
12 P. A. H ast o, Optimal inequalities between Seierts means and power means, Mathematical
Inequalities & Applications, vol. 7, no. 1, pp. 4753, 2004.