1. Introduction and Preliminaries
Let
X be an infinite dimensional complex Banach space. A continuous function
is said to be an
entire analytic function (or just an
entire function) if its restriction on any finite dimensional subspace is analytic. If an entire function
f satisfies
for every
and
then
f is called an
n-homogeneous polynomial. It is well known that for an
n-homogeneous polynomial
f there exists a unique symmetric
n-linear mapping
associated with f such that
Each zero-homogeneous polynomial is a constant. A finite sum of homogeneous polynomials is a
polynomial. The space of all entire analytic functions on
X is denoted by
the space of all polynomials on
X is denoted by
and the space of all
n-homogeneous polynomials on
X is denoted by
For every entire function
f there exists a sequence of continuous
n-homogeneous polynomials
(so-called
Taylor polynomials) such that
and the series converges for every
The Taylor series expansion (
1) uniformly converges on the open ball
centered at zero with radius
where
The radius is called the radius of uniform convergence of f at zero or the radius of boundedness of f at zero because the ball is the largest open ball at zero such that f is bounded on every closed subset of it. If then f is bounded on all bounded subsets of X and is called a function of bounded type. The algebra of all functions of bounded type on X is denoted by Functions in are called entire functions of unbounded type. Note that for every
It is well-known that every infinite dimensional Banach space
X admits entire functions of unbounded type. For example, for a given weak*-null sequence
, which always exists (see p. 157 [
1]), the function
is an entire function of unbounded type on
We say that a sequence of functions
on
X (not necessary linear) is
weak*-null if
as
for every
Entire functions of unbounded type were studied by many authors. In [
2] Aron constructed an entire function
f on a Banach space
X such that for every
there is a point
such that
f is unbounded on the ball of radius
centered at
In [
3,
4] Ansemil, Aron, and Ponte constructed entire functions
f on a Banach space which are bounded on any given finite collection of balls and unbounded on another given finite collection of balls. The set
is not linear and is not closed under multiplication of functions. However, Lopez–Salazar Codes in [
5] show that for every infinite-dimensional Banach space
X the set
contains an infinite-dimensional linear space (without zero) and even an infinite-dimensional algebra (without zero).
Let
be a sequence of polynomials on
We denote by
the smallest unital algebra containing all polynomials in
Let
be the closure of
in
with respect to the metrizable topology of the uniform convergence on bounded subsets of
and
is the subalgebra of all entire functions
f on
X such that their Taylor polynomials
are in
The algebras
and
were investigated in [
6,
7]. A typical example of
is the algebra of symmetric polynomials. Let
S be a group of isometric operators on a Banach space
A function
f on
X is
S-
symmetric if it is invariant with respect to the action of
Symmetric polynomials and analytic functions on Banach spaces with respect to various groups were studied in [
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18].
Symmetric entire functions of unbounded type on
were studied in [
19]. In [
20] the authors considered the question:
Let be subspaces of Under which conditions is there a function such that ? In this paper we show that some natural subspaces
do not support entire functions of unbounded type. In particular, we show that there are no symmetric entire functions of unbounded type on
In
Section 2 we propose some conditions under which
supports entire functions of unbounded type and construct some counterexamples. In
Section 3 we show that if
X is an infinite dimensional Banach space, then
can be represented as a union of infinite dimensional linear subspaces (without the origin). Furthermore, we show that for some cases
contains infinite dimensional algebras (without the origin). Some results of this paper were announced in [
21].
We refer the reader to the books of Dineen [
1] and Mujica [
22] for extensive studies of analytic functions on Banach spaces.
2. Conditions of the Unboundedness
Proposition 1. Let be a weak*-null sequence of polynomials on X such that and Then for every strictly increasing sequence of positive integers the functionis a function of unbounded type. Proof. Let
Since
as
there exists a number
m such that
for some
all
Hence the series
converges. Thus
is well-defined for every
On the other hand,
and so
Therefore,
f is an entire function of unbounded type. □
Let us notice the condition “ is a weak*-null sequence of polynomials on X such that ” is not sufficient to claim that is a function of unbounded type. For example, if for all then the series may be divergent.
Throughout the paper we will use the notations for the principal root of a and for the multi-valued root function of
Proposition 2. Let be a sequence of polynomials on X such that and The functionis of unbounded type if and only if for every Proof. Let us suppose that
satisfies (
3). Then for every
there is a number
such that for every
Thus,
Hence, is well-defined for every On the other hand, and so f is an entire function of unbounded type.
Conversely, let
Then for every
the series
converges for every
This implies
and so
Since, every vector can be represented by the proposition is proved. □
In [
20] the following theorem was proved.
Theorem 1. Let us suppose that there is a dense subset and a sequence of polynomials such that for every there exists with the property that for every for all and where is the symmetric n-linear mapping associated with Then It is not difficult to show that if a sequence of polynomials satisfies the conditions of Theorem 1, then it is weak*-null. From Proposition 2 it follows that if satisfies the conditions of Theorem 1, then is weak*-null.
For a given sequence of polynomials
on
we denote by
a multi-valued map from
X to
defined by
Let
be polynomials on
defined by
The algebra, generated by polynomials
was considered in [
7]. Let us fix some evident properties of
Proposition 3. For every sequence of polynomials on the following statements hold:
- 1.
The range of X under is in
- 2.
maps the ball into the ball
- 3.
If z is in the range of then
Proof. Since for every So if then and In addition, □
Lemma 1. Let Then the restriction of f to belongs to
Proof. According to the Aron–Berner result [
23], a function
can be extended to an analytic function
f on
if and only if
□
We say that the polynomial algebra supports analytic functions of unbounded type if there exists a function
Theorem 2. Let be as in Proposition 3.
- 1.
If maps X onto then the algebra does not support analytic functions of unbounded type.
- 2.
If maps X to then - 3.
If there is such that then
Proof. (1) Let us prove first that
does not support analytic functions of unbounded type on
where
is
for
Suppose that
Since each
is an algebraic combination of
and
we have that
depends of finitely many coordinates
Thus, there is
such that
In other words, the norm of
in
is equal to the norm of the restriction of
on
Let
be the restriction of
f on
Hence we have that
So if f is a function of unbounded type, then is a function of unbounded type. However, it contradicts Lemma 1.
Let
f now be an arbitrary function of unbounded type in
Since all polynomials
belong to
there are polynomials
of
n complex variables,
such that
for every
Clearly every
is an
n-homogeneous continuous polynomial. Since
maps bounded sets to bounded sets,
must be a function of unbounded type. However, it is impossible because of the first part of the proof.
(2) If
then
Thus f is well defined on X and so belongs to Since Hence,
(3) If
for some fixed
then there exists a constant
and a subsequence
such that
for all
Let us consider the following function of one complex variable
The radius of convergence of this series satisfies
so if
then the series
diverges. Thus it does not belong to
□
Remark 1. Formally, we do not assume in Theorem 2 that X is infinite dimensional. However, any finite dimensional space does not admit entire functions of unbounded type. Thus, if then X must be infinite dimensional.
Example 1. In [19] (see also [20]) it is shown that if wherethenis an entire function of unbounded type on It is known [13] that Thus, Theorem 2 implies that for every The algebra coincides with the algebra of all symmetric polynomials on (see e.g., [12]) and admits another algebraic basis of homogeneous polynomials It is easy to see that and for every where Hence, However, as we observed, the algebra of symmetric polynomials supports entire functions of unbounded type. Therefore, if then but may still support entire functions of unbounded type.
Note that the existence of an isomorphism of and does not imply that does not support analytic functions of unbounded type.
Example 2. It is known [7] that there is an isomorphism such that but supports analytic functions of unbounded type. For example In other words, the isomorphism J can not be extended to an isomorphism between and
Let us recall that a function on
is
symmetric if it is invariant with respect to measuring and measure preserving automorphisms of the interval
The polynomials
form an algebraic basis in the space of all symmetric polynomials on
Thus, the algebra of symmetric polynomials
is a partial case of
if
and
In [
20] the authors asked:
Does an entire symmetric analytic function of unbounded type exist on ? Now we have a negative answer to this question.
Corollary 1. All entire symmetric functions on are functions of bounded type.
Proof. Let
be an entire symmetric function on
Then each Taylor’s polynomial
must be symmetric and so can be represented as an algebraic combination of polynomials
In [
14] is proved that the map
is onto
Thus, by Theorem 2, the algebra of symmetric polynomials on
does not support entire functions of unbounded type. Hence
is of bounded type. □
3. Lineability of
Theorem 3. If supports analytic functions of unbounded type, then for every there exists an infinite dimensional linear subspace in which consists (excepting zero) of analytic functions of unbounded type and contains
Proof. Then
for some
Let
and
Denote by
the subset of all nonnegative integers
such that
In other words, if
then
that is, for every subsequence
Let
be a partition of the set
into infinite many disjoint subsets
so that
for
and
is the finite or infinite set defined above. Let
We denote
Thus is a partition of into infinitely many disjoint subsets of infinite cardinality.
For any bounded sequence of numbers
we assign a function
For every
the function
is well defined on
X and is of unbounded type. Indeed, for the Taylor polynomials
of
we have
Moreover, since
there is a number
j such that
Thus
and so
Since
for every
On the other hand, the set which depends on the choice of
and
is a linear space because
for all
Clearly
for
□
Note that the subspace is not maximal. Indeed, if is a subpartition of then It is easy to deduce by the Zorn Lemma that there is a maximal linear subspace in containing a given function of unbounded type. So we have the following corollary.
Corollary 2. Let X be an infinite dimensional Banach space. The set can be represented as a union of infinite dimensional linear subspaces (without the origin).
It is known (see [
5]) that for every infinite dimensional Banach space
X there are sequences
and
such that
for every
where and is the Kronecker delta.
In Theorem 2 of [
5], actually it was proved that if the functionals
are as above, then for every strictly increasing sequence of prime numbers
the following functions
generate an infinite dimensional algebra
such that every nonzero element
h in
is an entire function of unbounded type and
In particular, it is so if
is the basis in
and
is the sequence of coordinate functionals.
Theorem 4. Let be a sequence of n-homogeneous polynomials on a Banach space X such that and there exists a sequence in X such that and Then for every strictly increasing sequence of prime numbers the functionsgenerate an infinite-dimensional algebra such that every nonzero element in is an entire function of unbounded type and Proof. Let us consider the algebra
generated by functions
Note that
for every evaluation of
From Theorem 2 of [
5] mentioned above, it follows that if
and
then
Every function
can be represented as a finite algebraic combination of functions
where
and
for all
In other words,
Hence, and so all functions in are unbounded on some bounded subsets. On the other hand, since all functions by Theorem 2 are well-defined on X and so their finite algebraic combinations are well-defined on X too. Thus, □
Corollary 3. Let be the basis of symmetric polynomials on as in Example 1. Then for every strictly increasing sequence of prime numbers the functionsgenerate an infinite-dimensional subalgebra in the algebra of symmetric analytic functions comprising (excepting zero) of analytic functions of unbounded type. Proof. We need to construct a sequence
biorthogonal to
Let us define
where
are the roots of the unity. From the Vieta formulas it follows that
and
If
then
because
has only
n nonzero coordinates. Thus
In addition, using the Stirling formula, we can estimate
Thus, we can apply Theorem 4 for the sequence of polynomials □