![MathJax Logo](/templates/jsp/_style2/_tandf/pb2/images/math-jax.gif)
ABSTRACT
We extend Paley–Wiener results in the Bargmann setting deduced in Nabizadeh et al. [Paley-Wiener properties for spaces of entire functions, (preprint), arXiv:1806.10752.] to larger classes of power series expansions. At the same time, we deduce characterizations of all Pilipović spaces and their distributions (and not only of low orders as in Nabizadeh et al. [Paley-Wiener properties for spaces of entire functions, (preprint), arXiv:1806.10752.]).
Keywords:
AMS SUBJECT CLASSIFICATIONS:
1. Introduction
Classical Paley–Wiener theorems characterize functions and distributions with certain restricted supports in terms of estimates of their Fourier-Laplace transforms. For example, let f be a distribution on and let
be the ball with centre at origin and radius
. Then f is supported in
if and only if
for some
.
In [Citation1,Citation2] a type of Paley–Wiener results were deduced for certain spaces of entire functions, where the usual Fourier transform were replaced by the reproducing kernel of the Bargmann transform (see [Citation3] and Section 2 for notations). This reproducing kernel is an analytic global Fourier-Laplace transform with respect to a suitable Gaussian measure. For example, in [Citation2] it is proved that if
is the set of all entire functions and
(
),
, is the set of all
such that
for some r>0 (for every r>0), then
(1)
(1) The latter equality in (Equation1
(1)
(1) ) is in [Citation1] refined into the relation
for some characteristic function χ of a polydisc D, centred at origin, and a function F which is defined and analytic in a neighbourhood of D. The space
is also characterized in [Citation1] by
when χ above is fixed.
In fact, in [Citation1], similar characterizations are deduced for all and
with
as well as for the set
when
which consists of all
such that
. For example, let χ be the characteristic function of a polydisc, centred at origin in
. Then it is proved in [Citation1] that
In Section 3, we extend the set of characterizations given in [Citation1] in different ways. First, we show that we can choose χ above in a larger class of functions and measures. Second, we characterize strictly larger spaces than . More specifically, we characterize all the spaces
(2)
(2) for any
, as well as the larger spaces
(3)
(3) the sets of all formal power series expansions
(4)
(4) such that
for every h, r>0 (for some h, r>0). We notice that the spaces (Equation2
(2)
(2) ) discussed above can also be described as the sets of all formal power series expansions (Equation4
(4)
(4) ) such that
for some h, r>0 (for every h, r>0). (Cf. [Citation2, Citation4] and Section 2.) We remark that for
,
and
, defined in this way, are still spaces of entire functions, but with other types of estimates, compared to the case
considered above (cf. [Citation2] and Section 2).
By the definitions we have , and by using the convention
it follows that
(cf. [Citation2]). In particular, if
, then
and
are contained in
, and in [Citation2], it is proved that the former spaces consist of all
such that
for every r>0 respective for some r>0.
In Section 3, we show among others that if is the map
with χ as above, then the mappings
when
,
when
, and
when
, are well-defined and bijective.
In Section 4, we apply these mapping properties to deduce characterizations of Pilipović spaces and their distribution spaces in terms of images of the adjoint of Gaussian windowed short-time Fourier transform, acting on suitable spaces which are strongly linked to the spaces in (Equation2(2)
(2) ) and (Equation3
(3)
(3) ). These spaces are obtained by imposing the same type of estimates on their Hermite coefficients as for the power series coefficients of the spaces in (Equation2
(2)
(2) ) and (Equation3
(3)
(3) ). It turns out that Pilipović spaces and their distribution spaces are exactly the counter images of the spaces in (Equation2
(2)
(2) ) and (Equation3
(3)
(3) ) under the Bargmann transform (cf. [Citation2]).
2. Preliminaries
In this section, we recall some basic facts. We start by discussing Pilipović spaces and some of their properties. Then we recall some facts on modulation spaces. Finally we recall the Bargmann transform and some of its mapping properties, and introduce suitable classes of entire functions on .
2.1. Spaces of sequences
The definitions of Pilipović spaces and spaces of power series expansions are based on certain spaces of sequences on , indexed by the extended set
of
. We extend the inequality relations on
to the set
, by letting
when
and
. (Cf. [Citation2].)
Definition 2.1
Let and
.
The set
consists of all formal sequences
, and
is the set of all
such that
for at most finite numbers of
;
If
and
, then the Banach spaces
consists of all
such that their corresponding norms
respectively, are finite;
The space
(
) is the inductive limit (projective limit) of
with respect to r>0, and
(
) is the projective limit (inductive limit) of
with respect to r>0.
We also let be the set of all
such that
when
. Then
and
is a Banach space under the norm
We equip
with the inductive limit topology of
, and supply
with Fréchet space topology through the semi-norms
.
In what follows, denotes the scalar product in the Hilbert space
.
Remark 2.2
For the spaces in Definition 2.1, the following is true. We leave the verifications for the reader.
If
then
The space
is dense in each space in Definition 2.1 (1) and (3);
If
and
, then the map
from
to
is uniquely extendable to continuous mappings from
to
. The duals of
and
can be identified by
respective
, through these extensions of the form
.
2.2. Pilipović spaces and spaces of power series expansions on ![](//:0)
![](//:0)
We recall that the Hermite function of order is defined by
It follows that
for some polynomial
of order α on
, called the Hermite polynomial of order α. The Hermite functions are eigenfunctions to the Fourier transform, and to the Harmonic oscillator
which acts on functions and (ultra-)distributions defined on
. More precisely, we have
It is well-known that the set of Hermite functions is a basis for
and an orthonormal basis for
(cf. [Citation5]). In particular, if
, then
where
(5)
(5) is the Hermite seriers expansion of f, and
(6)
(6) is the Hermite coefficient of f of order
.
We shall also consider formal power series expansions on , centred at origin. That is, we shall consider formal expressions of the form
(7)
(7)
Definition 2.3
The set consists of all formal Hermite series expansions (Equation5
(5)
(5) ), and
consists of all formal power series expansions (Equation7
(7)
(7) ). The sets
and
consist of all
respective
such that
and
for at most finite numbers of
.
If
and
, then
(8)
(8) are the sets of all Hermite series expansions (Equation5
(5)
(5) ) such that their coefficients
belong to
,
,
respective
;
If
and
, then
(9)
(9) are the sets of all power series expansions (Equation7
(7)
(7) ) such that their coefficients
belong to
,
,
respective
.
The spaces and
in Definition 2.3 are called Pilipović spaces of Roumieu respectively Beurling types of order s, and
and
are called Pilipović distribution spaces of Roumieu respectively Beurling types of order s.
Remark 2.4
Let be the map from
to
which takes the sequence
to the expansion (Equation5
(5)
(5) ), and let
be the map from
to
which takes the sequence
to the expansion (Equation7
(7)
(7) ). Then it is clear that
restricts to bijective mappings from
(10)
(10) to respective spaces in (Equation8
(8)
(8) ), and that
restricts to bijective mappings from the spaces in (Equation10
(10)
(10) ) to respecive spaces in (Equation9
(9)
(9) ).
We let the topologies of the spaces in (Equation8(8)
(8) ) and (Equation9
(9)
(9) ) be inherited from the topologies of respective spaces in (Equation10
(10)
(10) ), through the mappings
and
.
The following result shows that Pilipović spaces of order may in convenient ways be characterized by estimates of powers of harmonic oscillators applied on the involved functions. We omit the proof since the result follows in the case
from [Citation6] and from [Citation2] for general s.
Proposition 2.5
Let (s>0) be real,
be the harmonic oscillator on
and set
Then
(
), if and only if
for some r>0 (for every r>0). The topologies of
and
agree with the inductive and projective limit topologies, respectively, induced by the semi-norms
, r>0.
Remark 2.6
Let and
be the Fourier invariant Gelfand-Shilov spaces of order
and of Roumieu and Beurling types respectively (see [Citation2] for notations). Then it is proved in [Citation6,Citation7] that
and
2.3. Spaces of entire functions and the Bargmann transform
Let be open and let
be non-empty (but not necessary open). Then
is the set of all analytic functions in Ω, and
where the union is taken over all open sets
such that
. In particular, if
is fixed, then
is the set of all complex-valued functions which are defined and analytic near
.
We shall now consider the Bargmann transform. We set
and otherwise
denotes the duality between test function spaces and their corresponding duals. The Bargmann transform
of
is defined by the formula
(11)
(11) (cf. [Citation8]). We note that if
, then the Bargmann transform
of f is the entire function on
, given by
or
(12)
(12) where the Bargmann kernel
is given by
Evidently, the right-hand side in (Equation12
(12)
(12) ) makes sense when
and defines an element in
, since
can be interpreted as an element in
with values in
.
It was proved in [Citation8] that is a bijective and isometric map from
to the Hilbert space
, where
consists of all measurable functions F on
such that
(13)
(13) Here
, where
is the Lebesgue measure on
. We recall that
and
are Hilbert spaces, where the scalar product are given by
(14)
(14) If
, then we set
and
.
Furthermore, in [Citation8], Bargmann showed that there is a convenient reproducing formula on . More precisely, let
(15)
(15) when
belongs to
for every
. Then it is proved in [Citation8, Citation9] that
is the orthogonal projection of
onto
. In particular,
when
.
In [Citation8] it is also proved that
(16)
(16) In particular, the Bargmann transform maps the orthonormal basis
in
bijectively into the orthonormal basis
of monomials in
.
For general we now set
(17)
(17) where
and
are given by Remark 2.4. It follows from (Equation16
(16)
(16) ) that
in (Equation17
(17)
(17) ) agrees with
in (Equation11
(11)
(11) ) when
, and that this is the only way to continuously extend the Bargmann transform to the space
. It follows that
is a homeomorphism from
to
, which restricts to homeomorphisms from the spaces in (Equation8
(8)
(8) ) to the spaces in (Equation9
(9)
(9) ), respectively. If
and
are given by (Equation5
(5)
(5) ) and (Equation7
(7)
(7) ) with
for all
, then it follows that
.
It follows that if and
, then
(18)
(18) By the definitions we get the following proposition on duality for Pilipović spaces and their Bargmann images. The details are left for the reader.
Proposition 2.7
Let and
. Then the form
on
is uniquely extendable to sesqui-linear forms on
The duals of
and
are equal to
and
, respectively, through the form
.
The same holds true if the spaces in (Equation8(8)
(8) ) and the form
are replaced by corresponding spaces in (Equation9
(9)
(9) ) and the form
, at each occurrence.
If ,
,
,
and
, then
and
are defined by the formula (Equation18
(18)
(18) ). It follows that
(19)
(19) holds for such choices of f and g.
Remark 2.8
In [Citation2], the spaces in (Equation9(9)
(9) ), contained in
are identified as follows in terms of spaces of analytic functions:
if
, then
(
) is equal to
if
and
, then
(
) is equal to
and
(
) is equal to
Furthermore,
, and
;
if
, then
(
) is equal to
if
(
), then
(
) is equal to
and
(
) is equal to
Additionally to Remark 2.8 we have
Here and in what follows,
is the (open) polydisc
with centre and radii given by
2.4. A test function space introduced by Gröchenig
In this section, we recall some comparison results deduced in [Citation2], between a test function space, , introduced by Gröchenig in [Citation10] to handle modulation spaces with elements in spaces of ultra-distributions.
The definition of is given as follows.
Definition 2.9
The space consists of all
such that
(20)
(20) for some
.
Evidently, we could have included the factors and
in the function
in (Equation20
(20)
(20) ). The following reformulation of [Citation2, Lemma 4.9] justifies the separation. The result is essential when deducing the characterizations of Pilipović spaces in Section 4.
Lemma 2.10
Let . Then the Bargmann transform of f in (Equation20
(20)
(20) ) is given by
, where
(21)
(21)
The first part of the next result follows from [Citation2, Theorem 4.10] and the last part of the result follows [Citation1, Theorem 2.2]. The proof is therefore omitted.
Proposition 2.11
The following is true:
;
the image of
under the map
equals
.
3. Paley–Wiener properties for bargmann-Pilipović spaces
In this section, we show that if ν is a suitable compactly supported function or measure, then the composition between the reproducing kernel
and multiplication operator
maps spaces in (Equation9
(9)
(9) ) into other spaces in (Equation9
(9)
(9) ). In the first part we state the main results given in Theorems 3.7–3.8. They are straight-forward consequences of Propositions 3.11 and 3.12, where more detailed information concerning involved constants are given. Thereafter we deduce results which are needed for their proofs. Depending of the choices of
and
, there are several different situations for characterizing the spaces in (Equation9
(9)
(9) ). This gives rise to a quite large flora of main results, where each one takes care of a particular situation.
In order to present the main results, we need the following definition.
Definition 3.1
Let be such that
. Then the set
consists of all non-zero positive Borel measures ν on
such that the following is true:
is radial symmetric in each variable
;
the support of ν contains
and is contained in
The set of compactly supported, positive, bounded and radial symmetric measures is given by
Remark 3.2
Let be such that
,
,
be the set of all non-negative
such that (1) and (2) in Definition 3.1 holds with F in place of ν and let
Then it is clear that
and
decrease with p and are contained in
and
, respectively. In particular, these sets contain
and
, respectively, in [Citation1].
Remark 3.3
It is clear that the sets in Definition 3.1 and Remark 3.2 are invariant under multiplications with positive measurable, locally bounded functions on which are radial symmetric in each complex variable
in
. In particular, they are invariant under multiplications with
for every
.
Remark 3.4
Let be such that
and
. Here and in what follows we write
when
and
satisfy
for every
. By Riesz representation theorem it follows that
where
and some non-zero positive Borel measure
on
such that the support of
contains
and is contained in
3.1. Main results
Our main investigations concern mapping properties of operators of the form
(22)
(22) when acting on the spaces given in Definition 2.3 (2).
Before stating the main results, we need the following lemmas, which explain some properties of the map (Equation22(22)
(22) ) when acting on the monomials
.
Lemma 3.5
Let be such that
,
and let
be the same as in Remark 3.4. Then
(23)
(23) where
(24)
(24) satisfies
(25)
(25)
Proof.
By using polar coordinates in each complex variable when integrating we get
(26)
(26)
where
(27)
(27) with
By Taylor expansions, we get
where the second equality is justified by
and Weierstrass' theorem.
By inserting this into (Equation26(26)
(26) ) and (Equation27
(27)
(27) ) we get
and (Equation23
(23)
(23) ) follows.
The estimates in (Equation25(25)
(25) ) are straight-forward consequences of (Equation24
(24)
(24) ) and the support properties of
. The details are left for the reader.
By replacing ν in the previous lemma with suitable radial symmetric compactly supported distributions we get the following.
Lemma 3.6
Let s>1, be such that
,
be radial symmetric in each
such that
Then (Equation23
(23)
(23) ) holds with
(28)
(28) for some
with
Furthermore,
(29)
(29)
Proof.
By using polar coordinates in each complex variable, the pull-back formula [Citation3, Theorem 6.1.2] and Fubbini's theorem for distributions and ultra-distributions (cf. [Citation3, Section 5.1] and [Citation11, Section 2]), we get
(30)
(30)
for some
, where
By the same arguments as in the proof of Lemma 3.5 we get
and (Equation23
(23)
(23) ) follows with
given by (Equation28
(28)
(28) ), by combining the latter identity with (Equation30
(30)
(30) ).
The support assertions for follow from the support properties of ν, and the estimate (Equation29
(29)
(29) ) follows from the fact that
is a bounded set in
with respect to α in the support of
. This gives the result.
Due to Lemma 3.6 we let be the set of all
which are radial symmetric in each
and such that
.
Theorem 3.7
Let ,
and
(
). Then the map (Equation22
(22)
(22) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) on
Theorem 3.8
Let and
(
). Then the following is true:
If
, then the map (Equation22
(22)
(22) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
;
if
and
, then the map (Equation22
(22)
(22) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
;
if
and
, then the map (Equation22
(22)
(22) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
.
The limit cases for the situations in the previous theorem are treated in the next result. We observe that (3) and parts of (2) in the previous theorem are reached already in [Citation1].
Theorem 3.9
Let (
) and
. Then the following is true:
The map (Equation22
(22)
(22) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
;
The map (Equation22
(22)
(22) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
.
Remark 3.10
Since
Theorems 3.7–3.9 still hold true after
have been replaced by
.
The following result is an essential part of the proof of Theorem 3.7.
Proposition 3.11
Let ,
and
be fixed, and let
be as in (Equation24
(24)
(24) ). Then the following is true:
The map
from
to
is uniquely extendable to a homeomorphism on
, and
(31)
(31)
it holds
(32)
(32) for some
such that
, if and only if
(33)
(33) for some
such that
;
it holds
(34)
(34) for some
such that
, if and only if
(35)
(35) for some
such that
.
Here it is understood that the signs in the exponents in (Equation32(32)
(32) ) and (Equation33
(33)
(33) ) agree.
Proof.
The assertion (1) is an immediate consequence of (Equation23(23)
(23) ) in Lemma 3.5. In fact, by Lemma 3.5 and (Equation23
(23)
(23) ), the only possible extension of
is to let
(36)
(36) when
(37)
(37) which obviously defines a continuous map on
.
Since
it follows from (Equation31
(31)
(31) ) that
for every
with
. This gives (2) and (3).
In order to prove Theorem 3.8 we need the next proposition. The result is a straight-forward consequence of (Equation23(23)
(23) ), (Equation25
(25)
(25) ) in Lemma 3.5 and Proposition 3.11 (1). The details are left for the reader.
Proposition 3.12
Let ,
,
,
and
be as in (Equation24
(24)
(24) ). Then
(38)
(38)
and
(39)
(39)
Proof
Proof of Theorems 3.7, 3.8 and 3.9
Theorem 3.7 is a straight-forward consequence of Proposition 3.11. The details are left for the reader.
By letting ,
and
, then
. Hence (Equation38
(38)
(38) ) and (Equation39
(39)
(39) ) give
(40)
(40) Theorem 3.8 (1) now follows from Proposition 3.11 (1) and (Equation40
(40)
(40) ).
If instead ,
and
, then
. Hence (Equation38
(38)
(38) ) and (Equation39
(39)
(39) ) give
(41)
(41) Theorem 3.8 (2) now follows from Proposition 3.11 (1) and (Equation41
(41)
(41) ).
If instead ,
and
, then
. Hence (Equation38
(38)
(38) ) and (Equation39
(39)
(39) ) give
(42)
(42) Theorem 3.8 (3) now follows from Proposition 3.11 (1) and (Equation42
(42)
(42) ).
Finally, Theorem 3.9 follows by similar arguments, letting when proving (1) and letting
when proving (2) in Theorem 3.9. The details are left for the reader.
4. Characterizations of Pilipović spaces
In this section we combine Lemma 2.10 with Theorems 3.7–3.9 to get characterizations of Pilipović spaces.
We shall perform such characterizations by considering mapping properties of extensions of the map
from
to
, where
(43)
(43) Here we have identified
with the polydisc
in
when
. We notice that
equals f in (Equation20
(20)
(20) ), if in addition
.
We recall that the Bargmann transform is homeomorphic between the spaces in (Equation8(8)
(8) ) and (Equation9
(9)
(9) ) when
and
. The following results of Paley–Wiener types for Pilipović spaces follow from these facts and by some straight-forward combinations of Lemma 2.10 and Theorems 3.7–3.9. The details are left for the reader.
Theorem 4.1
Let ,
and
. Then the map (Equation43
(43)
(43) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from the spaces in (Equation8
(8)
(8) ) to corresponding spaces in (Equation9
(9)
(9) ).
Theorem 4.2
Let and
. Then the following is true:
If
, then the map (Equation43
(43)
(43) ) from
to
is uniquely extendable to homeomorphisms from
to
, and from
to
;
if
and
, then the map (Equation43
(43)
(43) ) from
to
is uniquely extendable to homeomorphisms from
to
, and from
to
;
if
and
, then the map (Equation43
(43)
(43) ) from
to
is uniquely extendable to homeomorphisms from
to
, and from
to
.
Theorem 4.3
Let and
. Then the following is true:
The map (Equation43
(43)
(43) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
;
The map (Equation43
(43)
(43) ) from
to
is uniquely extendable to homeomorphisms (continuous mappings) from
to
, and from
to
.
Disclosure statement
No potential conflict of interest was reported by the author.
References
- Nabizadeh E, Pfeuffer C, Toft J. Paley-Wiener properties for spaces of entire functions, (preprint), arXiv:1806.10752.
- Toft J. Images of function and distribution spaces under the Bargmann transform. J Pseudo-Differ Oper Appl. 2017;8:83–139. doi: 10.1007/s11868-016-0165-9
- Hörmander L. The analysis of linear partial differential operators, vol. I–III. Berlin: Springer-Verlag; 1983. 1985.
- Fernandez C, Galbis A, Toft J. The Bargmann transform and powers of harmonic oscillator on Gelfand-Shilov subspaces. RACSAM. 2017;111:1–13. doi: 10.1007/s13398-015-0273-z
- Reed M, Simon B. Methods of modern mathematical physics. London: Academic Press; 1979.
- Pilipović S. Tempered ultradistributions. Boll U.M.I.. 1988;7:235–251.
- Pilipović S. Generalization of Zemanian spaces of generalized functions which have orthonormal series expansions. SIAM J Math Anal. 1986;17:477–484. doi: 10.1137/0517037
- Bargmann V. On a Hilbert space of analytic functions and an associated integral transform. Commun Pure Appl Math. 1961;14:187–214. doi: 10.1002/cpa.3160140303
- Bargmann V. On a Hilbert space of analytic functions and an associated integral transform. part II. a family of related function spaces. application to distribution theory. Commun Pure Appl Math. 1967;20:1–101. doi: 10.1002/cpa.3160200102
- Gröchenig KH. Foundations of time-frequency analysis. Boston (MA): Birkhäuser; 2001.
- Toft J. Tensor products for Gelfand-Shilov and Pilipović distribution spaces. J Anal (appeared online 2019).
- Cordero AUGRP E, Pilipović S, Rodino L, et al. Quasianalytic Gelfand-Shilov spaces with applications to localization operators. Rocky Mt J Math. 2010;40:1123–1147. doi: 10.1216/RMJ-2010-40-4-1123
- Janssen AUGRP AMEM, Eijndhoven SJL. Spaces of type W, growth of Hermite coefficients, Wigner distribution, and Bargmann transform. J Math Anal Appl. 1990;152:368–390. doi: 10.1016/0022-247X(90)90071-M
- Toft AUGRPJ. The Bargmann transform on modulation and Gelfand-Shilov spaces, with applications to Toeplitz and pseudo-differential operators. J Pseudo-Differ Oper Appl. 2012;3:145–227. doi: 10.1007/s11868-011-0044-3