Next Article in Journal / Special Issue
Wilson Bases and Ultradistributions
Previous Article in Journal
Revisiting a Classic Identity That Implies the Rogers–Ramanujan Identities II
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Characterization of Wave Fronts of Ultradistributions Using Directional Short-Time Fourier Transform

by
Sanja Atanasova
1,*,
Snježana Maksimović
2 and
Stevan Pilipović
3
1
Faculty of Electrical Engineering and Information Technologies, Ss. Cyril and Methodius University, Rugjer Boshkovik 18, 1000 Skopje, North Macedonia
2
Faculty of Architecture, Civil Engineering and Geodesy, University of Banja Luka, Stepe Stepanovića 77/3, 78000 Banja Luka, Bosnia and Herzegovina
3
Faculty of Sciences and Mathematics, University of Novi Sad, Trg Obradovića 4, 21000 Novi Sad, Serbia
*
Author to whom correspondence should be addressed.
Axioms 2021, 10(4), 240; https://doi.org/10.3390/axioms10040240
Submission received: 25 August 2021 / Revised: 14 September 2021 / Accepted: 18 September 2021 / Published: 28 September 2021
(This article belongs to the Special Issue Time-Frequency Analysis, Distributions, and Operators)

Abstract

:
In this paper we give a characterization of Sobolev k-directional wave front of order p [ 1 , ) of tempered ultradistributions via the directional short-time Fourier transform.

1. Introduction

The theory of directional sensitive kind of the short-time Fourier transform, in short STFT, was initially introduced and investigated in [1,2] as a blend of Radon transform and time–frequency analysis. It allows to gain information in time and frequency of a function along a certain direction or hyperplane. Following the concept of [1], in [3], the directional STFT was extended to the space of tempered distributions. Moreover, in [4], the k-directional short-time Fourier transform, in short k-DSTFT, was introduced and the results of [5] were extended to the spaces of tempered ultradistributions of Roumieu class.
Starting form [6], wave fronts have shown to be useful concepts when analyzing the propagation of different type of singularities in the theory of partial differential equations, which led to introducing various wave front sets [7,8,9,10].
Following the recent trend on studying integral transforms on the spaces of ultradistribution [11,12], authors in [4] introduce the k-directional regular sets to analyze the regularity properties of a tempered ultradistribution of Roumieu class. Furthermore, the wave front set using the k-DSTFT (k-directional wave front) of a tempered ultradistribution of Roumieu class and the partial wave front in terms of [6] are considered, and it is shown that this partial wave front is equivalent to the k-directional wave front.
This paper is a continuation of our work presented in [4] for both Beurling and Roumieu cases. The main result is established in Theorem 2 where we give characterization of the Sobolev wave front of order p [ 1 , ) via the k-DSTFT of tempered ultradistributions. We also consider partial wave fronts in terms of [6,13], and it is shown that these notions are equivalent with the k-directional Sobolev wave front.
The main novelty of this work is the proof of Theorem 2 where we follow the idea already proved in [4] but here with another decomposition and estimates of involved integrals. On the basis of this proof we introduce a new kind of wave front and make necessary analysis of it through our main theorem, Theorem 3.

1.1. Notation

For a given multi-index l = ( l 1 , , l n ) N 0 n and x = ( x 1 , , x n ) R n , we denote x l = x 1 l 1 x n l n and ( i ) | l | D l = x l = | l | x 1 l 1 x n l n , | l | = l 1 + + l n . Points in R k are denoted by x ˜ = ( x 1 , , x k ) . The notation Ω R n is used for an open set and K Ω for a compact set K which is contained in Ω . By F ( f ) ( x ) = f ^ ( x ) = R n f ( t ) e 2 π i t · x d t , x R n , we denote the Fourier transform of a function f. The inner product of f and g in L 2 is denoted by ( f , g ) and f , g means a dual paring. Thus, ( f , g ) = f , g ¯ . We also use the notation Γ ξ for a cone neighborhood of ξ , L r ( ξ ) and B r ( ξ ) for an open and a closed ball with a center ξ and radius r > 0 , respectively.

1.2. Ultradistribution Spaces

Let ( M l ) l N , M 0 = 1 be a sequence of positive numbers which monotonically increases to infinity and satisfies the following:
( M . 1 )
M l 2 M l 1 M l + 1 , l N ;
( M . 2 )
There exist constants A, H > 1 such that
M l A H l min 0 q l M q M l q , l , q N 0 ;
( M . 3 )
There exists a constant A such that l = q + 1 M l 1 / M l < A q M q + 1 / M q , q N ;
Sometimes we can replace properties ( M . 2 ) and ( M . 3 ) by the following weaker conditions:
( M . 2 )
There exist constants A, H > 1 such that
M l + 1 A H l M l , l N 0 ;
( M . 3 )
l = 1 M l 1 / M l < .
We will measure the decay properties of elements of Gelfand–Shilov spaces with respect to the Gevrey sequences M l = l ! α , α > 1 .
Let a > 0 . Following [14], we recall the definitions of some spaces of test functions:
E a α ( K ) : = { φ C Ω : sup t K , l N 0 n a | l | l ! α | D l φ ( t ) | < } ; D a α ( K ) : = E h α ( K ) { φ C Ω : supp φ K } ; E ( α ) ( K ) : = lim a E a α ( K ) ; E ( α ) ( Ω ) : = lim K Ω E ( α ) ( K ) ; D ( α ) ( K ) : = lim a D a α ( K ) ; D ( α ) ( Ω ) : = lim K Ω D ( α ) ( K ) . E { α } ( K ) : = lim a 0 E a α ( K ) ; E { α } ( Ω ) : = lim K Ω E { α } ( K ) ; D { α } ( K ) : = lim a 0 D a { α } ( K ) ; D { α } ( Ω ) : = lim K Ω D { α } ( K ) .
The elements of the space D ( α ) ( Ω ) (resp. D { α } ( Ω ) ) are called ultradifferentiable functions with compact support of Beurling class (resp. Roumieu class). Their strong duals are spaces of ultradistributions D ( α ) ( Ω ) (resp. D { α } ( Ω ) ). E ( α ) ( Ω ) (resp. E { α } ( Ω ) ) is a subspace of D ( α ) ( Ω ) (resp. D { α } ( Ω ) ) that consists of all compactly supported ultradistributions.
Following [11], we introduce the test spaces for spaces of Beurling and Roumieu tempered ultradistributions as a special case of ultradistributional spaces.
Let α , β > 0 , and α + β > 1 (resp. α + β 1 ). If α + β = 1 , then we presume also α , β 0 .
Let a > 0 . We denoted by ( S a ) β α ( R n ) the Banach space of all smooth functions φ on R n for which
σ a α , β ( φ ) = sup t R n , l , q N 0 n a | l | + | q | l ! β q ! α | t l φ ( q ) ( t ) | < .
The space Σ β α ( R n ) (resp. S β α ( R n ) ) is defined as a projective (resp. an inductive) limit of the space ( S a ) β α ( R n ) :
Σ β α ( R n ) = lim a ( S a ) β α ( R n ) ( resp . S β α ( R n ) = lim a 0 ( S a ) β α ( R n ) ) ,
and its strong dual Σ α β ( R n ) (resp. S α β ( R n ) ) is called the space of ultradistibutions of Beurling type (resp. Roumieu type). These spaces ( α + β > 1 , resp. α + β 1 ) are closed under translation, dilation, multiplication, differentiation, and under the action of specified infinite order differential operators (see Section 1.2.1). The Roumieu type spaces are the well-known spaces of Gelfand–Shilov.
When α = β , we use S ( α ) ( R n ) (resp. S { α } ( R n ) ) instead of Σ α α ( R n ) (resp. S α α ( R n ) ).

1.2.1. Ultradifferential Operators

It is said that P ( ξ ) = l N 0 n a l ξ l , ξ R n , is an ultrapolynomial of Beurling class (of Roumieu class), if the coefficients a l satisfy:
( a > 0 , C a > 0 ) ( resp . a > 0 , C a > 0 ) ( l N 0 n ) | a l | C a a | l | / M l .
The corresponding operator P ( D ) = l N 0 n a l D l is an ultradifferential operator of Beurling class (resp. Roumieu class). When M l = l ! α it is called ultradifferential operator of class ( α ) (resp. class { α } ). As M l satisfies ( M . 2 ) , they act continuously on E ( α ) and D ( α ) (resp. E { α } and D { α } ), and the corresponding spaces of ultradistributions.
The following representation theorem holds [11]:
For any f Σ β α ( R n ) (resp. f S β α ( R n ) ) there exist P 1 ( D ) -ultradifferential operator of class ( α ) (resp. class { α } ), an ultrapolynomial P 2 ( ξ ) of class ( β ) (resp. class { β } ) and an F L 2 ( R n ) such that
f ( ξ ) = P 1 ( D ) ( P 2 ( ξ ) F ( ξ ) ) .
We will deal only with elliptic operators for which the function P ( ξ ) satisfies [14] (Proposition 4.5): there exist a > 0 and C a > 0 (resp. for every a > 0 there exists C a > 0 ) such that
C a 1 e a | ξ | 1 / α | P ( ξ ) | C a e a | ξ | 1 / α , ξ R n .
In the quasi-analytic case (when ( M . 3 ) does not hold) we have [4]: Let r 1 there is C > 0 such that for all ξ R n , l N 0 n
| D ξ l 1 P ( ξ ) | C l ! r | l | | P ( ξ ) | .

1.3. The k-DSTFT and the k-Directional Synthesis Operator

We recall some definitions and assertions from [4], where only the Roumieu case was considered. Here we state also the Beurling case, since the results of [4] also hold for the Beurling-type spaces.
Let u k = ( u 1 , , u k ) , where u i , i = 1 , , k , are independent vectors of S n 1 . Let y ˜ = ( y 1 , , y k ) R k and g Σ β α ( R k ) { 0 } (resp. g S β α ( R k ) { 0 } ). The k-directional short-time Fourier transform of f L 2 ( R n ) is defined by [4]
D S g , u k f ( y ˜ , ξ ) = R n f ( t ) g u k , y ˜ , ξ ( t ) ¯ d t , ξ R n ,
and the k-directional synthesis operator of F L 2 ( R k + n ) is defined by [4]
D S g , u k * F ( t ) = R n R k F ( y ˜ , ξ ) g u k , y ˜ , ξ ( t ) d y ˜ d ξ , t R n ,
where g u k , y ˜ , ξ ( t ) = g ( ( u 1 · t , , u k · t ) ( y 1 , , y k ) ) e 2 π i ξ · t , t R n . It is shown in [4] (Proposition 2.4) that for f S β α ( R n ) the following reconstruction formula holds,
f ( t ) = 1 ( g , φ ) R n R k D S g , u k f ( y ˜ , ξ ) φ u k , y ˜ , ξ ( t ) d y ˜ d ξ , t R n ,
where φ S β α ( R k ) is the synthesis window for g S β α ( R k ) { 0 } . The same holds for f Σ β α ( R n ) when g , φ Σ β α ( R k ) . Thus, the relation (7) takes the form
( D S φ , u k * D S g , u k ) f = ( g , φ ) f .
For the sake of simplicity we transfer the STFT in direction of u k into the STFT in e k direction. Recall the procedure (see [4]): Let A = [ u i , j ] k × n be a matrix with rows u i , i = 1 , , k and I be the identity matrix of order n k . Let B be an n × n matrix determined by A and I so that B t = s , where s 1 = u 1 , 1 t 1 + + u 1 , n t n , , s k = u k , 1 t 1 + + u k , n t n ,   s k + 1 = t k + 1 , , s n = t n . The matrix B is regular, so put C = B 1 and e k = ( e 1 , , e k ) , where e 1 , , e k are unit vectors of the coordinate system of R k . If we change the variables t = C s , and η = C T ξ , then for f L 2 ( R n ) , g Σ β α ( R k ) (resp. g S β α ( R k ) ), the equality (5) is transformed into:
D S g , u k f ( y ˜ , ξ ) = ( D S g , e k h ( s ) ) ( y ˜ , η ) = R n h ( s ) g ( s ˜ y ˜ ) ¯ e 2 π i s · η d s ,
where h ( s ) = det ( C ) f ( C s ) and (6) is transformed, for F L 2 ( R k + n ) , g Σ β α ( R k ) (resp. g S β α ( R k ) ), into:
D S g , e k * F ( s ) = R n R k F ( y ˜ , η ) g ( s ˜ y ˜ ) e 2 π i s · η d y ˜ d η , s R n .
The function h ( s ) = det ( C ) f ( C s ) is in Σ β α ( R n ) (resp. S β α ( R n ) ) if f Σ β α ( R n ) (resp. f S β α ( R n ) ). Additionally, if g ( s 1 , , s k ) = g 1 ( s 1 ) g k ( s k ) ( Σ β α ( R ) ) k (resp. g ( s 1 , , s k ) = g 1 ( s 1 ) g k ( s k ) ( S β α ( R ) ) k ), then
D S g , u k f ( y ˜ , ξ ) : = R n f ( t ) g 1 ( u 1 · t y 1 ) ¯ g k ( u k · t y k ) ¯ e 2 π i t · ξ d t = R n h ( s ) g 1 ( s 1 y 1 ) ¯ g k ( s k y k ) ¯ e 2 π i s · μ d s ,
and it is referred to as the partial short-time Fourier transform.
We have D S g , e k : Σ β α ( R n ) × Σ β α ( R k ) Σ β α ( R k + n ) (resp. D S g , e k : S β α ( R n ) × S β α ( R k ) S β α ( R k + n ) ) is a continuous bilinear mapping. This is proved in Theorem 2.3 in [4] in the Roumieu case. With Theorem 2.5 and Corollary 2.7 in [4] in Roumieu case, and similarly in the Beurling case, it follows that D S g , e k * : Σ β α ( R k + n ) × Σ 0 α ( R k ) Σ β α ( R n ) (resp. D S g , e k * : S β α ( R k + n ) × S 0 α ( R k ) S β α ( R n ) ) is also continuous. This allows us to extend the definitions of the k-DSTFT and its synthesis operator to their duals (see [4] (Proposition 2.10)).
The relation of the k-DSTFTs with respect to different windows is presented with the following assertion. It is given in [4] (Theorem 2.11) for the Roumieu case:
Theorem 1.
Let u k = ( u 1 , , u k ) , where u i , i = 1 , , k are independent vectors of S n 1 . Let φ , g , γ 1 S ( α ) ( R k ) (resp. φ , g , γ 1 S { α } ( R k ) ) where γ 1 is the synthesis window for g and γ 0 S ( α ) ( R n k ) (resp. γ 0 S { α } ( R n k ) ) so that R n k γ 0 ( t n k + 1 , , t n ) d t n k + 1 d t n 0 . Put
γ ( t 1 , , t n ) = γ 1 ( t 1 , , t k ) γ 0 ( t n k + 1 , , t n ) .
Let f Σ β α ( R n ) (resp. f S β α ( R n ) ), then
D S φ , u k f ( x ˜ , η ) = ( D S g , u k f ( s ˜ , ζ ) ) ( D S φ , u k γ ( s ˜ , ζ ) ) ( x ˜ , η ) ,
x ˜ , s ˜ R k , η , ζ R n .

2. The Main Results

The STFT in the direction of u k can be used in the detection of singularities determined by the hyperplanes orthogonal to vectors u 1 , , u k . For this purpose, we introduce k-directional regular sets and wave front sets for the Beurling (resp. Roumieu)-tempered ultradistributions using the STFT in the direction of u k . To simplify our exposition we transfer the STFT in direction of u k into the STFT in e k direction by the use of (8).
As in [5], if k = 1 , we consider direction e 1 = e 1 while for 1 < k n , we consider direction e k = ( e 1 , , e k ) . Let k = 1 and y 0 = y 0 , 1 R , and let Π e 1 , y 0 , ε = Π y 0 , ε : = { t R n : | t 1 y 0 | < ε } . It is a part of R n between two hyperplanes orthogonal to e 1 , that is,
Π y 0 , ε = y ( y 0 ε , y 0 + ε ) P y , ( y 0 = ( y 0 , 0 , , 0 ) , y = ( y , 0 , , 0 ) ) ,
and P y denotes the hyperplane orthogonal to e 1 passing through y. Let
Π e k , y ˜ , ε = Π e 1 , y 1 , ε Π e k , y k , ε , Π e k , y ˜ = Π e 1 , y 1 Π e k , y k .
The set Π e k , y ˜ , ε is a parallelepiped in R k . In R n this parallelepiped is determined by 2 k finite edges while the other edges are infinite. The set Π e k , y ˜ equals R n k translated by vectors y 1 , , y k . We call it n k -dimensional element of R n and it is denoted by P e k , y ˜ R n k . When k = n , we have the point y = ( y 1 , , y n ) .
Definition 1.
Let f S ( α ) ( R n ) (resp. f S { α } ( R n ) ) , α > 1 and p [ 1 , ) . It is said that f is ( α ) -p-k-directionally microlocally regular (in short, ( α ) -p-k-d.m.r.) (resp. { α } -p-k-directionally microlocally regular (in short, { α } -p-k-d.m.r.)) at ( P e k , y ˜ 0 , ξ 0 ) R k × ( R n { 0 } ) , that is, at every point of the form ( y ˜ 0 , ξ 0 ) if there exist g D ( α ) ( R k ) (resp. g D { α } ( R k ) ), g ( 0 ˜ ) 0 , a product of open balls L r ( y ˜ 0 ) = L r ( y 0 , 1 ) × × L r ( y 0 , k ) R k , a cone Γ ξ 0 and for each N N (resp. for some N N ) there exists C N > 0 such that
sup y ˜ L r ( y ˜ 0 ) | | D S g , e k f ( y ˜ , ξ ) e N | ξ | 1 / α | | L p ( Γ ξ 0 ) = sup y ˜ L r ( y ˜ 0 ) Γ ξ 0 | F ( f ( t ) g ( t ˜ y ˜ ) ¯ ) ( ξ ) | p e p N | ξ | 1 / α d ξ 1 / p C N .
If k = n , Definition 1 gives the classical Hörmander’s regularity [6].
Remark 1.(a) If f is ( α ) -p-k-d.m.r. (resp. { α } -p-k-d.m.r.) at ( P e k , y ˜ 0 , ξ 0 ) , then there exist an open ball L r ( y ˜ 0 ) and an open cone Γ Γ ξ 0 so that f is ( α ) -p-k-d.m.r. (resp. { α } -p-k-d.m.r.) at ( P e k , z ˜ 0 , θ 0 ) for any z ˜ 0 L r ( y ˜ 0 ) and θ 0 Γ . This implies that the union of all ( α ) -p-k-d.m.r. (resp. { α } -p-k-d.m.r) points ( P e k , z ˜ 0 , θ 0 ) , ( z ˜ 0 , θ 0 ) L r ( y ˜ 0 ) × Γ is an open set of R k × ( R n { 0 } ) .
(b) Denote by P r k the projection of R n onto R k . Then, the ( α ) -p-k-d.m.r. (resp. { α } -p-k-d.m.r.) point ( P e k , y ˜ 0 , ξ 0 ) , considered in R n × ( R n { 0 } ) with respect to the first k variables, equals ( P r k 1 × I ξ ) ( P e k , y ˜ 0 , ξ 0 ) ( I ξ is the identity matrix on R n ).
A p-Sobolev k-directional wave front of Beurling (resp. Roumieu) type is defined as the complement in R k × ( R n { 0 } ) of all ( α ) -p-k-d.m.r. (resp. { α } -p-k-d.m.r.) points ( P e k , y ˜ 0 , ξ 0 ) , and we denoted as Σ W F e k ( f ) (resp. S W F e k ( f ) ).

2.1. Independence with Respect to a Window Function

One of the main results in [4] (Theorem 3.4) shows that the wave front set does not depend on the used window. Here, we prove the same assertion for the p-Sobolev k-directional wave fronts. The idea is similar to the one in [6,13] (see [4]) but here the decomposition of the involved integrals and the use of ultradifferential operators make the proof more complex.
Theorem 2.
If (12) holds for some g D ( α ) ( R k ) (resp. g D { α } ( R k ) ), g ( 0 ˜ ) 0 , then it holds for every h D ( α ) ( R k ) (resp. h D { α } ( R k ) ), h ( 0 ˜ ) 0 supported by a ball B ρ ( 0 ˜ ) , where ρ ρ 0 and ρ 0 depends on r in (12).
Proof. 
We will focus only on the Roumieu-type spaces. The proof in the Beurling case will follow similarly. We assume that φ , g , γ 1 , belong to S 0 α ( R k ) where γ 1 defined by (10) is the synthesis window for g and γ 0 S { α } ( R n k ) . Additionally, suppose that f is a continuous function which satisfies (3), since we can use the methods of oscillatory integral and transfer the differentiation from f on other factors in integral expressions. Using [11] (Theorem 3.2.2), we obtain that f = P 0 ( D ) F , where F is a continuous function which satisfies
a > 0 , C a > 0 , ξ R n | F ( ξ ) | C a e a | ξ | 1 / α
and P 0 ( D ) is a differential operator.
We use Theorem 1, that is, the form (11). Assume that (12) holds. The constructions of balls we repeat from [5]. The window function γ is chosen so that supp γ B ρ 1 ( 0 ) and ρ 1 < r r 0 . Let h D { α } ( R k ) and supp h B ρ ( 0 ˜ ) . The aim is to find ρ 0 such that (12) holds for D S h , e k f ( x ˜ , η ) , with x ˜ B r 0 ( y ˜ 0 ) , η Γ 1 Γ ξ 0 , for ρ ρ 0 ( Γ 1 Γ ξ 0 implies that Γ 1 S n 1 is a compact subset of Γ ξ 0 S n 1 ).
We choose ρ 0 such that ρ 0 + ρ 1 < r r 0 and
ρ + ρ 1 + r 0 < r h o l d s f o r ρ ρ 0 .
This implies that
| y ˜ y ˜ 0 | < r
as a consequence of
| q ˜ | ρ 1 , | x ˜ y ˜ 0 | r 0   and | q ˜ ( ( x ˜ y ˜ 0 ) ( y ˜ y ˜ 0 ) ) | ρ .
Let Γ 1 Γ ξ 0 . Then, there exists c ( 0 , 1 ) such that η Γ 1 , | η | > 1 and
| η ξ | c | η | η Γ ξ 0 ; | η ξ | c | η | | η | ( 1 c ) 1 | ξ | .
Let x ˜ B r 0 ( y ˜ 0 ) , η Γ 1 and K = e N | η | 1 / α ε | η | 1 / α | D S h , e k f ( x ˜ , η ) | . Then, by (8) and (11)
sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 | K | p d η = sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α e p N | η | 1 / α | R k d y ˜
| η ξ | c | η | + | η ξ | c | η | ( R t n f ( t ) g ( t ˜ y ˜ ) ¯ e 2 π i t · ξ d t
R q n γ ( q ) h ( q ˜ ( x ˜ y ˜ ) ) ¯ e 2 π i q · ( η ξ ) d q ) d ξ | p d η = I 1 + I 2 .
We continue to estimate I 1
I 1 C sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α | R k d y ˜ ( Γ ξ 0 e p N | η ξ | 1 / α ( R t n | f ( t ) g ( t ˜ y ˜ ) ¯ | d t ) p
e p N | ξ | 1 / α | R q n P ( D q ) ( γ ( q ) h ( q ˜ ( x ˜ y ˜ ) ) ¯ ) e 2 π i q · ( η ξ ) P ( 2 π ( η ξ ) ) d q | p d ξ ) | d η
C sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α | R k d y ˜ Γ ξ 0 e p N | ξ | 1 / α ( R t n | f ( t ) g ( t ˜ y ˜ ) ¯ | d t ) p d ξ
sup ξ R n e p N | η ξ | 1 / α P ( 2 π ( η ξ ) ) | R q n P ( D q ) ( γ ( q ) h ( q ˜ ( x ˜ y ˜ ) ) ¯ ) d q | p | d η
By the assumptions that g , h , γ are with compact support, integrals over R k and R q n are finite while the integral over Γ ξ 0 is finite because of the assumption (1).
Now we consider I 2 .
I 2 = sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α e p N | η | 1 / α | R k d y ˜ | η ξ | c | η | D S g , e k f ( y ˜ , η ξ )
D S h , e k γ ( x ˜ y ˜ , ξ ) d ξ | p d η .
Let Ω = { ξ : | η ξ | c | η | } . By κ d 0 , 0 < d < 1 , we denote the characteristic function of Ω d = ξ Ω L d ( ξ ) , where Ω d is an open d-neighborhood of Ω . Then, put
κ d = κ d 0 φ d ,
where φ d = 1 d n φ ( · / d ) , φ D { α } ( R n ) is non-negative, supported in the ball B 1 ( 0 ) and equals 1 on B 1 / 2 ( 0 ) . By the construction we have that κ d equals one on Ω , it is supported in Ω 2 d , and all the derivatives of κ d are bounded. We note that
| Ω d ξ | | Ω 2 d κ d ( ξ ) d ξ | + | Ω 2 d { ξ : | η ξ | c η } κ d ( ξ ) d ξ | .
Then,
I 2 sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α ( R k | R n κ d e N | η | 1 / α D S g , e k f ( y ˜ , η ξ )
D S h , e k γ ( x ˜ y ˜ , ξ ) d ξ | p d y ˜ ) d η
+ sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α ( R k | Ω 2 d { ξ : | η ξ | c | η | } κ d e N | η | 1 / α D S g , e k f ( y ˜ , η ξ )
D S h , e k γ ( x ˜ y ˜ , ξ ) d ξ | p d y ˜ ) d η = I 2 , 1 + I 2 , 2 .
We first estimate I 2 , 1 .
I 2 , 1 C sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α | R k d y ˜ ( R ξ d e p N | η ξ | 1 / α ( R t n | f ( t ) g ( t ˜ y ˜ ) ¯ P ( 2 π t ) | d t ) p ·
e p N | ξ | 1 / α | R q n P ( D q ) ( γ ( q ) h ( q ˜ ( x ˜ y ˜ ) ) ¯ ) ·
P ( D ξ ) ( κ d ( η ξ ) e 2 π i q · ( η ξ ) ) P ( 2 π ( η ξ ) ) d q | p d ξ ) | d η
C sup x ˜ L r 0 ( y ˜ 0 ) Γ 1 e ε p | η | 1 / α | R k d y ˜ R ξ d e p N | ξ | 1 / α ( R t n | f ( t ) g ( t ˜ y ˜ ) ¯ P ( 2 π t ) | d t ) p d ξ
sup ξ R n e p N | η ξ | 1 / α P ( 2 π ( η ξ ) ) | R q n P ( D q ) ( γ ( q ) h ( q ˜ ( x ˜ y ˜ ) ) ¯ ) d q | p ) | d η
is finite.
Since integration goes trough a subset of { ξ : | η ξ | c | η | } , we can conclude that I 2 , 2 can be estimated similarly as I 1 . Additional therm κ d does not cause any problems since it belongs to E { α } . □

2.2. Equivalent Definition

In this section we characterize the wave front sets given Definition 1 with the ones formulated in the next definition. We will use the Fourier transform as well as the cut-off function, and since we will show that these to definitions are equivalent but we need to distinguish them, we add the prefix “locally” in front of this notation in the Definition 2. We follow our ideas outlined in [10] and prove that both definitions determine the same sets. For the sake of completeness, we give all the details of the proof although it is the repetition of our proof of the theorem in [10] where we have considered distributions instead of ultradistributions.
Definition 2.
Let f S ( α ) ( R n ) (resp. f S { α } ( R n ) ), p [ 1 , ) and s R . The point ( y ˜ 0 , ξ 0 ) R k × ( R n { 0 } ) is locally ( α ) -p-k- microlocally regular, in short locally ( α ) -p-k- m.r., (resp. locally { α } -p-k-m.r.) for f if there exists χ D ( α ) ( R k ) (resp. χ D { α } ( R k ) ) so that χ ( y ˜ 0 ) 0 and a cone Γ ξ 0 such that
e s | ξ | 1 / α F ( χ ( y ˜ ) f ( y ) ) ( ξ ) L p ( Γ ξ 0 ) < ,
y = ( y ˜ , y k + 1 , , y n ) R k × R n k .
For the complements we use the notation L Σ W F e k ( f ) (resp. L S W F e k ( f ) )
Theorem 3.
Let f S ( α ) ( R n ) ( f S { α } ( R n ) ) and p [ 1 , ) . The following conditions are equivalent.
( i )
( y ˜ 0 , ξ 0 ) L Σ W F e k ( f ) (resp. ( y ˜ 0 , ξ 0 ) L S W F e k ( f ) ).
( i i )
There exist a compact neighborhood K ˜ of y ˜ 0 and a cone Γ ξ 0 such that for every s > 0 (resp. for some s > 0 ) the mapping χ e s | · | 1 / α F ( χ ( y ˜ ) f ( y ) ) , D ( α ) ( K ˜ ) L p ( Γ ξ 0 ) (resp. D { α } ( K ˜ ) L p ( Γ ξ 0 ) ), is well-defined and continuous.
( i i i )
There exist a compact neighborhood K ˜ of y ˜ 0 , a cone Γ ξ 0 and C , a > 0 such that for all χ D ( α ) ( K ˜ { y ˜ 0 } ) and s > 0 (resp. for all χ D { α } ( K ˜ { y ˜ 0 } ) and some s > 0 ) there holds
sup y ˜ K ˜ e s | ξ | 1 / α D S χ , e k f ( y ˜ , ξ ) L p ( Γ ξ 0 ) C sup l N 0 n a | l | l ! α D l χ L ( R k ) .
Here, the set K ˜ { y ˜ 0 } = { y ˜ R k | y ˜ + y ˜ 0 K ˜ } .
( i v )
There exist a compact neighborhood K ˜ of y ˜ 0 , a cone Γ ξ 0 , such that for all s > 0 and corresponding χ D ( α ) ( R k ) (resp. for some s > 0 and corresponding χ D { α } ( R k ) ) with χ ( 0 ˜ ) 0 there holds sup y ˜ K ˜ e s | ξ | 1 / α D S χ , e k f ( y ˜ , ξ ) L p ( Γ ξ 0 ) < .
Proof. 
( i ) ( i i ) We will prove only the Roumieu case. We have that f is locally { α } -p-k-m.r. at ( y ˜ 0 , ξ 0 ) , which means that there exist χ D { α } ( R k ) , χ ( y ˜ 0 ) 0 , and there exists a cone Γ ξ 0 such that
C χ = e s | ξ | 1 / α F ( χ ( y ˜ ) f ( y ) ) ( ξ ) L p ( Γ ξ 0 ) < .
There exists a compact neighborhood K ˜ of y ˜ 0 where χ never vanishes. Moreover, there are constants C 1 and r 1 such that
| F ( χ ( y ˜ ) f ( y ) ) ( ξ ) | C 1 e r | ξ | 1 / α , ξ R n .
Now let Γ ξ 0 be a cone such that Γ ¯ ξ 0 Γ ξ 0 { 0 } . One can find 0 < c < 1 such that
{ η R n | ξ Γ ξ 0 s u c h t h a t | ξ η | c | ξ | } Γ ξ 0 .
We take ψ D { α } ( K ˜ ) , then F ( ψ ( y ˜ ) χ ( y ˜ ) f ( y ) ) = F ( ψ ( y ˜ ) ) F ( χ ( y ˜ ) f ( y ) ) . By the Minkowski integral inequality we have
e s | ξ | 1 / α F ( ψ ( y ˜ ) χ ( y ˜ ) f ( y ) ) ( ξ ) L p ( Γ ξ 0 ) R η n Γ ξ 0 e p s | ξ | 1 / α | F ( ψ ( y ˜ ) ) ( η ) | p | F ( χ ( y ˜ ) f ( y ) ) ( ξ η ) | p d ξ 1 / p d η I 1 + I 2 ,
where
I 1 = R η n | F ( ψ ( y ˜ ) ) ( η ) | | ξ | | η | / c ξ Γ ξ 0 e p s | ξ | 1 / α | F ( χ ( y ˜ ) f ( y ) ) ( ξ η ) | p d ξ 1 / p d η ,
I 2 = R η n | F ( ψ ( y ˜ ) ) ( η ) | | ξ | < | η | / c ξ Γ ξ 0 e p s | ξ | 1 / α | F ( χ ( y ˜ ) f ( y ) ) ( ξ η ) | p d ξ 1 / p d η .
Using the change of variable ξ η in the inner integral in I 1 we have
I 1 = R η n | F ( ψ ( y ˜ ) ) ( η ) | | ξ + η | | η | / c ξ Γ ξ 0 { η } e p s | ξ + η | 1 / α | F ( χ ( y ˜ ) f ( y ) ) ( ξ ) | p d ξ 1 / p d η ( 1 c ) s R η n | F ψ ( η ) | Γ ξ 0 e p s | ξ | 1 / α | F ( χ ( y ˜ ) f ( y ) ) ( ξ ) | p d ξ 1 / p d η = C χ F ( ψ ( y ˜ ) ) L 1 ( R n ) ( 1 c ) s .
For the inequality above, we have used { ξ Γ ξ 0 { η } | | ξ + η | | η | / c } Γ ξ 0 , which follows from (16). For I 2 we have
I 2 C 1 R η n | F ( ψ ( y ˜ ) ) ( η ) | | ξ | < | η | / c ξ Γ ξ 0 e p s | ξ | 1 / α + p r | ξ η | 1 / α d ξ 1 / p d η C 1 ( 1 + c 1 ) r c s n 1 e ( n 1 ) | · | 1 / α L p ( R n ) · · e ( r + s + n + 1 ) | · | 1 / α F ( ψ ( y ˜ ) ) L 1 ( R n ) .
From the estimates of I 1 and I 2 , we conclude that there exists C χ > 0 such that
e s | · | 1 / α F ( ψ ( y ˜ ) χ ( y ˜ ) f ( y ) ) L p ( Γ ξ 0 ) C χ e ( s + r + n + 1 ) | · | 1 / α F ( ψ ( y ˜ ) ) L 1 ( R n ) ,
ψ D { α } ( K ˜ ) . Now, the claim in ( i i ) can be deduced since for ψ D { α } ( K ˜ ) , we have ψ f = ( ψ / χ ) χ f with ψ / χ D { α } ( K ˜ ) .
( i i ) ( i i i ) Let K ˜ 1 be a compact neighborhood of y ˜ 0 , and we choose a cone Γ ξ 0 such that the mapping
χ e s | · | 1 / α F ( χ ( y ˜ ) f ( y ) ) , D { α } ( K ˜ 1 ) L p ( Γ ξ 0 ) ,
is well-defined and continuous. Under the assumption that K ˜ 1 = B r ( y ˜ 0 ) , for some r > 0 , there exist C > 0 and a > 0 such that
e s | ξ | 1 / α F ( χ ( y ˜ ) f ( y ) ) ( ξ ) L p ( Γ ξ 0 ) C sup l N 0 n a | l | l ! α D l χ L ( K ˜ 1 ) ,
χ D { α } ( K ˜ 1 ) .
Let K ˜ = B r / 2 ( y ˜ 0 ) . For χ D { α } ( K ˜ { y ˜ 0 } ) and y ˜ K ˜ , the function t ˜ χ ( t ˜ y ˜ ) ¯ belongs to D { α } ( K ˜ 1 ) and, as F ( χ ( · ˜ y ˜ ) f ( y ) ) ( ξ ) = D S χ , e k f ( y ˜ , ξ ) , we have
sup y ˜ K e s | ξ | 1 / α D S χ , e k f ( y ˜ , ξ ) L p ( Γ ξ 0 ) C sup y ˜ K sup l N 0 n a | l | l ! α D l χ ( · ˜ y ˜ ) L ( K ˜ 1 ) = C sup l N 0 n a | l | l ! α D l χ L ( R k ) .
The implication ( i i i ) ( i v ) is trivial. If we let y ˜ = y ˜ 0 , the implication ( i v ) ( i ) follows immediately. □

Author Contributions

Individual contributions of the authors were equally distributed in writing the original draft of the manuscript, editing or revising it. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data is available from the authors, upon reasonable request.

Acknowledgments

This paper was supported by the project “Time-frequency methods”, No. 174024 financed by the Ministry of Science, Republic of Serbia, by the project “Localization in the phase space: theoretical, numerical and practical aspects”, No. 19.032/961-103/19. of the Republic of Srpska Ministry for Scientific and Technological Development, Higher Education and Information Society, and by the bilateral project “Microlocal analysis and applications†between the Macedonian and Serbian academies of sciences and arts.

Conflicts of Interest

The authors declare that there is no conflict of interest between them regarding the publishing in this journal, and no support from any organisation for the submitted work is obtained.

References

  1. Giv, H.H. Directional short-time Fourier transform. J. Math. Anal. Appl. 2013, 399, 100–107. [Google Scholar] [CrossRef]
  2. Grafakos, L.; Sansing, C. Gabor frames and directional time-frequency analysis. Appl. Comput. Harmon. Anal. 2008, 25, 47–67. [Google Scholar] [CrossRef] [Green Version]
  3. Saneva, K.; Atanasova, S. Directional short-time Fourier transform of distributions. J. Inequal. Appl. 2016, 124, 1–10. [Google Scholar]
  4. Atanasova, S.; Maksimović, S.; Pilipović, S. Directional Short-Time Fourier Transform of Ultradistributions. Bull. Malays. Math. Sci. Soc. 2021, 44, 3069–3087. [Google Scholar] [CrossRef]
  5. Atanasova, S.; Pilipović, S.; Saneva, K. Directional Time–Frequency Analysis and Directional Regularity. Bull. Malays. Math. Sci. Soc. 2019, 42, 2075–2090. [Google Scholar] [CrossRef]
  6. Hörmander, L. The Analysis of Linear Partial Differential Operators I, 2nd ed.; Springer: Berlin/Heidelberg, Germany; New York, NY, USA, 1990. [Google Scholar]
  7. Coriasco, S.; Johansson, K.; Toft, J. Local wave-front sets of Banach and Fréchet types, and pseudo-differential operators. Monatsh. Math. 2013, 169, 285–316. [Google Scholar] [CrossRef]
  8. Coriasco, S.; Johansson, K.; Toft, J. Global Wave-front Sets of Banach, Fréchet and Modulation Space Types, and Pseudo-differential Operators. J. Differ. Equ. 2013, 254, 3228–3258. [Google Scholar] [CrossRef]
  9. Dimovski, P.; Prangoski, B. Wave front sets with respect to Banach spaces of ultradistributions. Characterisation via the short-time fourier transform. Filomat 2019, 33, 5829–5836. [Google Scholar] [CrossRef] [Green Version]
  10. Pilipović, S.; Prangoski, B. On the characterization of wave front sets via the short-time Fourier transform. Math. Notes 2019, 105, 153–157. [Google Scholar] [CrossRef] [Green Version]
  11. Carmichael, R.; Kamiński, A.; Pilipović, S. Boundary Values and Convolution in Ultradistribution Spaces; World Scientific: Singapore, 2007. [Google Scholar]
  12. Teofanov, N. Gelfand-Shilov spaces and localization operators. Funct. Anal. Approx. Comput. 2015, 7, 135–158. [Google Scholar]
  13. Hörmander, L. Lectures on Nonlinear Hyperbolic Differential Equations; Mathématiques et Applications 26; Springer: Berlin/Heidelberg, Germany; New York, NY, USA, 1997. [Google Scholar]
  14. Komatsu, H. Ultradistributions, I: Structure theorems and a characterization. J. Fac. Sci. Univ. Tokyo Sect. IA Math. 1973, 20, 25–105. [Google Scholar]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Atanasova, S.; Maksimović, S.; Pilipović, S. Characterization of Wave Fronts of Ultradistributions Using Directional Short-Time Fourier Transform. Axioms 2021, 10, 240. https://doi.org/10.3390/axioms10040240

AMA Style

Atanasova S, Maksimović S, Pilipović S. Characterization of Wave Fronts of Ultradistributions Using Directional Short-Time Fourier Transform. Axioms. 2021; 10(4):240. https://doi.org/10.3390/axioms10040240

Chicago/Turabian Style

Atanasova, Sanja, Snježana Maksimović, and Stevan Pilipović. 2021. "Characterization of Wave Fronts of Ultradistributions Using Directional Short-Time Fourier Transform" Axioms 10, no. 4: 240. https://doi.org/10.3390/axioms10040240

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop