Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets

We prove a Calderón reproducing formula for the Dunkl continuous wavelet transform on R. We apply this result to derive new inversion formulas for the dual Dunkl-Sonine integral transform.

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Symmetry, Integrability and Geometry: Methods and Applications
Datum:2009
1. Verfasser: Mourou, M.A.
Format: Artikel
Sprache:English
Veröffentlicht: Інститут математики НАН України 2009
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/149135
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets / M.A. Mourou // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 16 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-149135
record_format dspace
spelling Mourou, M.A.
2019-02-19T17:36:42Z
2019-02-19T17:36:42Z
2009
Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets / M.A. Mourou // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 16 назв. — англ.
1815-0659
2000 Mathematics Subject Classification: 42B20; 42C15; 44A15; 44A35
https://nasplib.isofts.kiev.ua/handle/123456789/149135
We prove a Calderón reproducing formula for the Dunkl continuous wavelet transform on R. We apply this result to derive new inversion formulas for the dual Dunkl-Sonine integral transform.
The author is grateful to the referees and editors for careful reading and useful comments.
en
Інститут математики НАН України
Symmetry, Integrability and Geometry: Methods and Applications
Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
spellingShingle Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
Mourou, M.A.
title_short Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
title_full Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
title_fullStr Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
title_full_unstemmed Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets
title_sort inversion of the dual dunkl-sonine transform on r using dunkl wavelets
author Mourou, M.A.
author_facet Mourou, M.A.
publishDate 2009
language English
container_title Symmetry, Integrability and Geometry: Methods and Applications
publisher Інститут математики НАН України
format Article
description We prove a Calderón reproducing formula for the Dunkl continuous wavelet transform on R. We apply this result to derive new inversion formulas for the dual Dunkl-Sonine integral transform.
issn 1815-0659
url https://nasplib.isofts.kiev.ua/handle/123456789/149135
citation_txt Inversion of the Dual Dunkl-Sonine Transform on R Using Dunkl Wavelets / M.A. Mourou // Symmetry, Integrability and Geometry: Methods and Applications. — 2009. — Т. 5. — Бібліогр.: 16 назв. — англ.
work_keys_str_mv AT mourouma inversionofthedualdunklsoninetransformonrusingdunklwavelets
first_indexed 2025-11-25T21:08:28Z
last_indexed 2025-11-25T21:08:28Z
_version_ 1850551243388223488
fulltext Symmetry, Integrability and Geometry: Methods and Applications SIGMA 5 (2009), 071, 12 pages Inversion of the Dual Dunkl–Sonine Transform on R Using Dunkl Wavelets Mohamed Ali MOUROU Department of Mathematics, Faculty of Sciences, Al-Jouf University, P.O. Box 2014, Al-Jouf, Skaka, Saudi Arabia E-mail: mohamed ali.mourou@yahoo.fr Received March 02, 2009, in final form July 04, 2009; Published online July 14, 2009 doi:10.3842/SIGMA.2009.071 Abstract. We prove a Calderón reproducing formula for the Dunkl continuous wavelet transform on R. We apply this result to derive new inversion formulas for the dual Dunkl– Sonine integral transform. Key words: Dunkl continuous wavelet transform; Calderón reproducing formula; dual Dunkl–Sonine integral transform 2000 Mathematics Subject Classification: 42B20; 42C15; 44A15; 44A35 1 Introduction The one-dimensional Dunkl kernel eγ , γ > −1/2, is defined by eγ(z) = jγ(iz) + z 2(γ + 1) jγ+1(iz), z ∈ C, where jγ(z) = Γ(γ + 1) ∞∑ n=0 (−1)n (z/2)2n n! Γ(n+ γ + 1) is the normalized spherical Bessel function of index γ. It is well-known (see [3]) that the functions eγ(λ·), λ ∈ C, are solutions of the differential-difference equation Λγu = λu, u(0) = 1, where Λγf(x) = f ′(x) + ( γ + 1 2 ) f(x)− f(−x) x is the Dunkl operator with parameter γ+1/2 associated with the reflection grour Z2 on R. Those operators were introduced and studied by Dunkl [2, 3, 4] in connection with a generalization of the classical theory of spherical harmonics. Besides its mathematical interest, the Dunkl operator Λα has quantum-mechanical applications; it is naturally involved in the study of one- dimensional harmonic oscillators governed by Wigner’s commutation rules [6, 11, 16]. It is known, see for example [14, 15], that the Dunkl kernels on R possess the following Sonine type integral representation eβ(λx) = ∫ |x| −|x| Kα,β(x, y) eα(λy) |y|2α+1 dy, λ ∈ C, x 6= 0, (1.1) mailto:mohamed_ali.mourou@yahoo.fr http://dx.doi.org/10.3842/SIGMA.2009.071 2 M.A. Mourou where Kα,β(x, y) :=  aα,β sgn(x) (x+ y) ( x2 − y2 )β−α−1 |x|2β+1 if |y| < |x|, 0 if |y| ≥ |x|, (1.2) with β > α > −1/2, and aα,β := Γ(β + 1) Γ(α+ 1) Γ(β − α) . Define the Dunkl–Sonine integral transform Xα,β and its dual tXα,β, respectively, by Xα,βf(x) = ∫ |x| −|x| Kα,β(x, y) f(y) |y|2α+1 dy, tXα,βf(y) = ∫ |x|≥|y| Kα,β(x, y) f(x) |x|2β+1 dx. Soltani has showed in [14] that the dual Dunkl–Sonine integral transform tXα,β is a trans- mutation operator between Λα and Λβ on the Schwartz space S(R), i.e., it is an automorphism of S(R) satisfying the intertwining relation tXα,β Λβ f = Λα tXα,β f, f ∈ S(R). The same author [14] has obtained inversion formulas for the transform tXα,β involving pseudo- differential-difference operators and only valid on a restricted subspace of S(R). The purpose of this paper is to investigate the use of Dunkl wavelets (see [5]) to derive a new inversion of the dual Dunkl–Sonine transform on some Lebesgue spaces. For other applications of wavelet type transforms to inverse problems we refer the reader to [7, 8] and the references therein. The content of this article is as follows. In Section 2 we recall some basic harmonic analysis results related to the Dunkl operator. In Section 3 we list some basic properties of the Dunkl– Sonine integral trnsform and its dual. In Section 4 we give the definition of the Dunkl continuous wavelet transform and we establish for this transform a Calderón formula. By combining the results of the two previous sections, we obtain in Section 5 two new inversion formulas for the dual Dunkl–Sonine integral transform. 2 Preliminaries Note 2.1. Throughout this section assume γ > −1/2. Define Lp(R, |x|2γ+1dx), 1 ≤ p ≤ ∞, as the class of measurable functions f on R for which ||f ||p,γ <∞, where ||f ||p,γ = (∫ R |f(x)|p|x|2γ+1dx )1/p , if p <∞, and ||f ||∞,γ = ||f ||∞ = ess supx∈R|f(x)|. S(R) stands for the usual Schwartz space. The Dunkl transform of order γ+1/2 on R is defined for a function f in L1(R, |x|2γ+1dx) by Fγf(λ) = ∫ R f(x) eγ(−iλx) |x|2γ+1dx, λ ∈ R. (2.1) Inversion of the Dual Dunkl–Sonine Transform on R 3 Remark 2.2. It is known that the Dunkl transform Fγ maps continuously and injectively L1(R, |x|2γ+1dx) into the space C0(R) (of continuous functions on R vanishing at infinity). Two standard results about the Dunkl transform Fγ are as follows. Theorem 2.3 (see [1]). (i) For every f ∈ L1 ∩ L2(R, |x|2γ+1dx) we have the Plancherel formula∫ R |f(x)|2|x|2γ+1dx = mγ ∫ R |Fγf(λ)|2|λ|2γ+1dλ, where mγ = 1 22γ+2(Γ(γ + 1))2 . (2.2) (ii) The Dunkl transform Fα extends uniquely to an isometric isomorphism from L2(R, |x|2γ+1dx) onto L2(R,mγ |λ|2γ+1dλ). The inverse transform is given by F−1 γ g(x) = mγ ∫ R g(λ)eγ(iλx)|λ|2γ+1dλ, where the integral converges in L2(R, |x|2γ+1dx). Theorem 2.4 (see [1]). The Dunkl transform Fα is an automorphism of S(R). An outstanding result about Dunkl kernels on R (see [12]) is the product formula eγ(λx)eγ(λy) = T x γ (eγ(λ·)) (y), λ ∈ C, x, y ∈ R, where T x γ stand for the Dunkl translation operators defined by T x γ f(y) = 1 2 ∫ 1 −1 f (√ x2 + y2 − 2xyt )( 1 + x− y√ x2 + y2 − 2xyt ) Wγ(t)dt + 1 2 ∫ 1 −1 f ( − √ x2 + y2 − 2xyt )( 1− x− y√ x2 + y2 − 2xyt ) Wγ(t)dt, (2.3) with Wγ(t) = Γ(γ + 1)√ π Γ(γ + 1/2) (1 + t) ( 1− t2 )γ−1/2 . The Dunkl convolution of two functions f , g on R is defined by the relation f ∗γ g(x) = ∫ R T x γ f(−y)g(y)|y|2γ+1dy. (2.4) Proposition 2.5 (see [13]). (i) Let p, q, r ∈ [1,∞] such that 1 p + 1 q−1 = 1 r . If f ∈Lp(R, |x|2γ+1dx) and g ∈Lq(R, |x|2γ+1dx), then f ∗γ g ∈ Lr(R, |x|2γ+1dx) and ||f ∗γ g||r,γ ≤ 4||f ||p,γ ||g||q,γ . (2.5) (ii) For f ∈ L1(R, |x|2γ+1dx) and g ∈ Lp(R, |x|2γ+1dx), p = 1 or 2, we have Fγ(f ∗γ g) = FγfFγg. (2.6) For more details about harmonic analysis related to the Dunkl operator on R the reader is referred, for example, to [9, 10]. 4 M.A. Mourou 3 The dual Dunkl–Sonine integral transform Throughout this section assume β > α > −1/2. Definition 3.1 (see [14]). The dual Dunkl–Sonine integral transform tXα,β is defined for smooth functions on R by tXα,βf(y) := ∫ |x|≥|y| Kα,β(x, y)f(x)|x|2β+1 dx, y ∈ R, (3.1) where Kα,β is the kernel given by (1.2). Remark 3.2. Clearly, if supp (f) ⊂ [−a, a] then supp ( tXα,βf ) ⊂ [−a, a]. The next statement provides formulas relating harmonic analysis tools tied to Λα with those tied to Λβ, and involving the operator tXα,β. Proposition 3.3. (i) The dual Dunkl–Sonine integral transform tXα,β maps L1(R, |x|2β+1dx) continuously into L1(R, |x|2α+1dx). (ii) For every f ∈ L1(R, |x|2β+1dx) we have the identity Fβ(f) = Fα ◦ tXα,β(f). (3.2) (iii) Let f, g ∈ L1(R, |x|2β+1dx). Then tXα,β(f ∗β g) = tXα,βf ∗α tXα,βg. (3.3) Proof. Let f ∈ L1(R, |x|2β+1dx). By Fubini’s theorem we have∫ R tXα,β(|f |)(y)|y|2α+1dy = ∫ R (∫ |x|≥|y| Kα,β(x, y)|f(x)||x|2β+1 dx ) |y|2α+1dy = ∫ R |f(x)| (∫ |x| −|x| Kα,β(x, y)|y|2α+1dy ) |x|2β+1 dx. But by (1.1),∫ |x| −|x| Kα,β(x, y)|y|2α+1dy = eβ(0) = 1. (3.4) Hence, tXα,βf is almost everywhere defined on R, belongs to L1(R, |x|2α+1dx) and ||tXα,βf ||1,α ≤ ||f ||1,β , which proves (i). Identity (3.2) follows by using (1.1), (2.1), (3.1), and Fubini’s theorem. Identity (3.3) follows by applying the Dunkl transform Fα to both its sides and by using (2.6), (3.2) and Remark 2.2. � Remark 3.4. From (3.2) and Remark 2.2, we deduce that the transform tXα,β maps L1(R, |x|2β+1dx) injectively into L1(R, |x|2α+1dx). From [14] we have the following result. Inversion of the Dual Dunkl–Sonine Transform on R 5 Theorem 3.5. The dual Dunkl–Sonine integral transform tXα,β is an automorphism of S(R) satisfying the intertwining relation tXα,βΛβf = Λα tXα,βf, f ∈ S(R). Moreover tXα,β admits the factorization tXα,βf = tV −1 α ◦ tVβf for all f ∈ S(R), where for γ > −1/2, tVγ denotes the dual Dunkl intertwining operator given by tVγf(y) = Γ(γ + 1)√ π Γ(γ + 1/2) ∫ |x|≥|y| sgn(x) (x+ y) ( x2 − y2 )γ−1/2 f(x) dx. Definition 3.6 (see [14]). The Dunkl–Sonine integral transform Xα,β is defined for a locally bounded function f on R by Xα,βf(x) =  ∫ |x| −|x| Kα,β(x, y) f(y) |y|2α+1 dy if x 6= 0, f(0) if x = 0. (3.5) Remark 3.7. (i) Notice that by (3.4), ||Xα,βf ||∞ ≤ ||f ||∞ if f ∈ L∞(R). (ii) It follows from (1.1) that eβ(λx) = Xα,β(eα(λ·)(x) (3.6) for all λ ∈ C and x ∈ R. Proposition 3.8. (i) For any f ∈ L∞(R) and g ∈ L1(R, |x|2β+1dx) we have the duality relation∫ R Xα,βf(x)g(x)|x|2β+1dx = ∫ R f(y) tXα,βg(y)|y|2α+1dy. (3.7) (ii) Let f ∈ L1(R, |x|2β+1dx) and g ∈ L∞(R). Then Xα,β ( tXα,βf ∗α g ) = f ∗β Xα,βg. (3.8) Proof. Identity (3.7) follows by using (3.1), (3.5) and Fubini’s theorem. Let us check (3.8). Let ψ ∈ S(R). By using (3.3), (3.7) and Fubini’s theorem, we have∫ R f ∗β Xα,βg(x)ψ(x)|x|2β+1dx = ∫ R Xα,βg(x)ψ∗βf −(x) |x|2β+1dx = ∫ R g(y) tXα,β(ψ∗βf −)(y)|y|2α+1dy = ∫ R g(y) ( tXα,βψ ∗α tXα,βf −) (y)|y|2α+1dy, where f−(x) = f(−x), x ∈ R. But an easy computation shows that tXα,βf − = ( tXα,βf )−. Hence,∫ R f ∗β Xα,βg(x)ψ(x)|x|2β+1dx = ∫ R g(y) tXα,βψ ∗α ( tXα,βf )− (y)|y|2α+1dy = ∫ R tXα,βf ∗αg(y) tXα,βψ(y)|y|2α+1dy = ∫ R Xα,β ( tXα,βf ∗α g ) (x)ψ(x)|x|2β+1dx. This clearly yields the result. � 6 M.A. Mourou 4 Calderón’s formula for the Dunkl continuous wavelet transform Throughout this section assume γ > −1/2. Definition 4.1. We say that a function g ∈ L2(R, |x|2γ+1dx) is a Dunkl wavelet of order γ, if it satisfies the admissibility condition 0 < Cγ g := ∫ ∞ 0 |Fγg(λ)|2dλ λ = ∫ ∞ 0 |Fγg(−λ)|2dλ λ <∞. (4.1) Remark 4.2. (i) If g is real-valued we have Fγg(−λ) = Fγg(λ), so (4.1) reduces to 0 < Cγ g := ∫ ∞ 0 |Fγg(λ)|2dλ λ <∞. (ii) If 0 6= g ∈ L2(R, |x|2γ+1dx) is real-valued and satisfies ∃ η > 0 such that Fγg(λ)−Fγg(0) = O(λη) as λ→ 0+ then (4.1) is equivalent to Fγg(0) = 0. Note 4.3. For a function g in L2(R, |x|2γ+1dx) and for (a, b) ∈ (0,∞)× R we write gγ a,b(x) := 1 a2γ+2 T−b γ ga(x), where T−b γ are the generalized translation operators given by (2.3), and ga(x) := g(x/a), x ∈ R. Remark 4.4. Let g ∈ L2(R, |x|2γ+1dx) and a > 0. Then it is easily checked that ga ∈ L2(R, |x|2γ+1dx), ||ga||2,γ = aγ+1 ||g||2,γ , and Fγ(ga)(λ) = a2γ+2Fγ(g)(aλ). Definition 4.5. Let g ∈ L2(R, |x|2γ+1dx) be a Dunkl wavelet of order γ. We define for regular functions f on R, the Dunkl continuous wavelet transform by Φγ g (f)(a, b) := ∫ R f(x)gγ a,b(x)|x| 2γ+1dx which can also be written in the form Φγ g (f)(a, b) = 1 a2γ+2 f ∗γ g̃a(b), where ∗γ is the generalized convolution product given by (2.4), and g̃a(x) := g(−x/a), x ∈ R. The Dunkl continuous wavelet transform has been investigated in depth in [5] in which precise definitions, examples, and a more complete discussion of its properties can be found. We look here for a Calderón formula for this transform. We start with some technical lemmas. Lemma 4.6. For all f, g ∈ L2(R, |x|2γ+1dx) and all ψ ∈ S(R) we have the identity∫ R f ∗γ g(x)F−1 γ ψ(x)|x|2γ+1dx = mγ ∫ R Fγf(λ)Fγg(λ)ψ−(λ)|λ|2γ+1dλ, where mγ is given by (2.2). Inversion of the Dual Dunkl–Sonine Transform on R 7 Proof. Fix g ∈ L2(R, |x|2γ+1dx) and ψ ∈ S(R). For f ∈ L2(R, |x|2γ+1dx) put S1(f) := ∫ R f ∗γ g(x)F−1 γ ψ(x)|x|2γ+1dx and S2(f) := mγ ∫ R Fγf(λ)Fγg(λ)ψ−(λ)|λ|2γ+1dλ. By (2.5), (2.6) and Theorem 2.3, we see that S1(f) = S2(f) for each f ∈ L1 ∩L2(R, |x|2γ+1dx). Moreover, by using (2.5), Hölder’s inequality and Theorem 2.3 we have |S1(f)| ≤ ||f ∗γ g||∞||F−1 γ ψ||1,γ ≤ 4||f ||2,γ ||g||2,γ ||F−1 γ ψ||1,γ and |S2(f)| ≤ mγ ||FγfFγg||1,γ ||ψ||∞ ≤ mγ ||Fγf ||2,γ ||Fγg||2,γ ||ψ||∞ = ||f ||2,γ ||g||2,γ ||ψ||∞, which shows that the linear functionals S1 and S2 are bounded on L2(R, |x|2γ+1dx). Therefore S1 ≡ S2, and the lemma is proved. � Lemma 4.7. Let f1, f2 ∈ L2(R, |x|2γ+1dx). Then f1 ∗γ f2 ∈ L2(R, |x|2γ+1dx) if and only if Fγf1Fγf2 ∈ L2(R, |x|2γ+1dx) and we have Fγ(f1 ∗γ f2) = Fγf1Fγf2 in the L2-case. Proof. Suppose f1 ∗γ f2 ∈ L2(R, |x|2γ+1dx). By Lemma 4.6 and Theorem 2.3, we have for any ψ ∈ S(R), mγ ∫ R Fγf1(λ)Fγf2(λ)ψ(λ)|λ|2γ+1dλ = ∫ R f1 ∗γ f2(x)F−1 γ ψ−(x)|x|2γ+1dx = ∫ R f1 ∗γ f2(x)F−1 γ ψ(x)|x|2γ+1dx = mγ ∫ R Fγ(f1 ∗γ f2)(λ)ψ(λ)|λ|2γ+1dλ, which shows that Fγf1Fγf2 = Fγ(f1 ∗γ f2). Conversely, if Fγf1Fγf2 ∈ L2(R, |x|2γ+1dx), then by Lemma 4.6 and Theorem 2.3, we have for any ψ ∈ S(R),∫ R f1 ∗γ f2(x)F−1 γ ψ(x)|x|2γ+1dx = mγ ∫ R Fγf1(λ)Fγf2(λ)ψ̃(λ)|λ|2γ+1dλ = ∫ R F−1 γ (Fγf1Fγf2)(x)F−1 γ ψ(x)|x|2γ+1dx, which shows, in view of Theorem 2.4, that f1 ∗γ f2 = F−1 γ (Fγf1Fγf2). This achieves the proof of Lemma 4.7. � A combination of Lemma 4.7 and Theorem 2.3 gives us the following. Lemma 4.8. Let f1, f2 ∈ L2(R, |x|2γ+1dx). Then∫ R |f1 ∗γ f2(x)|2|x|2γ+1dx = mγ ∫ R |Fγf1(λ)|2|Fγf2(λ)|2|λ|2γ+1dλ, where both sides are finite or infinite. 8 M.A. Mourou Lemma 4.9. Let g ∈ L2(R, |x|2γ+1dx) be a Dunkl wavelet of order γ such that Fγg ∈ L∞(R). For 0 < ε < δ <∞ define Gε,δ(x) := 1 Cγ g ∫ δ ε ga ∗γ g̃a(x) da a4γ+5 (4.2) and Kε,δ(λ) := 1 Cγ g ∫ δ ε |Fγg(aλ)|2da a . (4.3) Then Gε,δ ∈ L2(R, |x|2γ+1dx), Kε,δ ∈ ( L1 ∩ L2 ) (R, |x|2γ+1dx), (4.4) and Fγ(Gε,δ) = Kε,δ. Proof. Using Schwarz inequality for the measure da a4γ+5 we obtain |Gε,δ(x)|2 ≤ 1 (Cγ g )2 (∫ δ ε da a4γ+5 )∫ δ ε |ga ∗γ g̃a(x)|2 da a4γ+5 , so ∫ R |Gε,δ(x)|2|x|2γ+1dx ≤ 1 (Cγ g )2 (∫ δ ε da a4γ+5 )∫ δ ε ∫ R |ga ∗γ g̃a(x)|2|x|2γ+1dx da a4γ+5 . By Theorem 2.3, Lemma 4.8, and Remark 4.4, we have∫ R |ga ∗γ g̃a(x)|2|x|2γ+1dx = mγ ∫ R |Fγ(ga)(λ)|4|λ|2γ+1dλ ≤ mγ ||Fγ(ga)||2∞ ∫ R |Fγ(ga)(λ)|2|λ|2γ+1dλ = ||Fγ(ga)||2∞ ||ga||22,γ = a6γ+6 ||Fγg||2∞ ||g|| 2 2,γ . Hence∫ R |Gε,δ(x)|2|x|2γ+1dx ≤ ||Fγg||2∞ ||g|| 2 2,γ (Cγ g )2 (∫ δ ε a2γ+1da )(∫ δ ε da a4γ+5 ) <∞. The second assertion in (4.4) is easily checked. Let us calculate Fγ(Gε,δ). Fix x ∈ R. From Theorem 2.3 and Lemma 4.7 we get ga ∗γ g̃a(x) = mγ ∫ R |Fγ(ga)(λ)|2eγ(iλx)|λ|2γ+1dλ, so Gε,δ(x) = mγ Cγ g ∫ δ ε (∫ R |Fγ(ga)(λ)|2eγ(iλx)|λ|2γ+1dλ ) da a4γ+5 . As |eγ(iz)| ≤ 1 for all z ∈ R (see [12]), we deduce by Theorem 2.3 that mγ ∫ δ ε ∫ R |Fγ(ga)(λ)|2|eγ(iλx)||λ|2γ+1dλ da a4γ+5 Inversion of the Dual Dunkl–Sonine Transform on R 9 ≤ ∫ δ ε ||ga||22,γ da a4γ+5 = ||g||22,γ ∫ δ ε da a2γ+3 <∞. Hence, applying Fubini’s theorem, we find that Gε,δ(x) = mγ ∫ R ( 1 Cγ g ∫ δ ε |Fγg(aλ)|2da a ) eγ(iλx)|λ|2γ+1dλ = mγ ∫ R Kε,δ(λ)eγ(iλx)|λ|2γ+1dλ which completes the proof. � We can now state the main result of this section. Theorem 4.10 (Calderón’s formula). Let g ∈ L2(R, |x|2γ+1dx) be a Dunkl wavelet of order γ such that Fγg ∈ L∞(R). Then for f ∈ L2(R, |x|2γ+1dx) and 0 < ε < δ <∞, the function fε,δ(x) := 1 Cγ g ∫ δ ε ∫ R Φγ g (f)(a, b)ga,b(x)|b|2γ+1db da a belongs to L2(R, |x|2γ+1dx) and satisfies lim ε→0, δ→∞ ∥∥fε,δ − f ∥∥ 2,γ = 0. (4.5) Proof. It is easily seen that fε,δ = f ∗γ Gε,δ, where Gε,δ is given by (4.2). It follows by Lemmas 4.7 and 4.9 that fε,δ ∈ L2(R, |x|2γ+1dx) and Fγ(fε,δ) = Fγ(f)Kε,δ, where Kε,δ is as in (4.3). From this and Theorem 2.3 we obtain ∥∥fε,δ − f ∥∥2 2,γ = mγ ∫ R |Fγ(fε,δ − f)(λ)|2|λ|2γ+1dλ = mγ ∫ R |Fγf(λ)|2(1−Kε,δ(λ))2|λ|2γ+1dλ. But by (4.1) we have lim ε→0, δ→∞ Kε,δ(λ) = 1, for almost all λ ∈ R. So (4.5) follows from the dominated convergence theorem. � Another pointwise inversion formula for the Dunkl wavelet transform, proved in [5], is as follows. Theorem 4.11. Let g ∈ L2(R, |x|2γ+1dx) be a Dunkl wavelet of order γ. If both f and Fγf are in L1(R, |x|2γ+1dx) then we have f(x) = 1 Cγ g ∫ ∞ 0 (∫ R Φγ g (f)(a, b)gγ a,b(x)|b| 2γ+1db ) da a , a.e., where, for each x ∈ R, both the inner integral and the outer integral are absolutely convergent, but possibly not the double integral. 10 M.A. Mourou 5 Inversion of the dual Dunkl–Sonine transform using Dunkl wavelets From now on assume β > α > −1/2. In order to invert the dual Dunkl–Sonine transform, we need the following two technical lemmas. Lemma 5.1. Let 0 6= g ∈ L1∩L2(R, |x|2α+1dx) such that Fαg ∈ L1(R, |x|2α+1dx) and satisfying ∃ η > β − 2α− 1 such that Fαg(λ) = O (|λ|η) as λ→ 0. (5.1) Then Xα,βg ∈ L2(R, |x|2β+1dx) and Fβ(Xα,βg)(λ) = mα mβ Fαg(λ) |λ|2(β−α) . Proof. By Theorem 2.3 we have g(x) = mα ∫ R Fαg(λ)eα(iλx)|λ|2α+1dλ, a.e. So using (3.6), we find that Xα,βg(x) = mβ ∫ R hα,β(λ)eβ(iλx)|λ|2β+1dλ, a.e. (5.2) with hα,β(λ) := mα mβ Fαg(λ) |λ|2(β−α) . Clearly, hα,β ∈ L1(R, |x|2β+1dx). So it suffices, in view of (5.2) and Theorem 2.3, to prove that hα,β belongs to L2(R, |x|2β+1dx). We have∫ R |hα,β(λ)|2|λ|2β+1dλ = ( mα mβ )2 ∫ R |λ|4α−2β+1|Fαg(λ)|2dλ = ( mα mβ )2 (∫ |λ|≤1 + ∫ |λ|≥1 ) |λ|4α−2β+1|Fαg(λ)|2dλ := I1 + I2. By (5.1) there is a positive constant k such that I1 ≤ k ∫ |λ|≤1 |λ|2η+4α−2β+1dλ = k η + 2α− β + 1 <∞. From Theorem 2.3, it follows that I2 = ( mα mβ )2 ∫ |λ|≥1 |λ|2(α−β)|Fαg(λ)|2|λ|2α+1dλ ≤ ( mα mβ )2 ∫ |λ|≥1 |Fαg(λ)|2|λ|2α+1dλ ≤ ( mα mβ )2 ||Fαg||22,α = mα (mβ)2 ||g||22,α <∞ which ends the proof. � Inversion of the Dual Dunkl–Sonine Transform on R 11 Lemma 5.2. Let 0 6= g ∈L1 ∩L2(R, |x|2α+1dx) be real-valued such that Fαg ∈L1(R, |x|2α+1dx) and satisfying ∃ η > 2(β − α) such that Fαg(λ) = O(λη) as λ→ 0+. (5.3) Then Xα,βg ∈ L2(R, |x|2β+1dx) is a Dunkl wavelet of order β and Fβ(Xα,βg) ∈ L∞(R). Proof. By combining (5.3) and Lemma 5.1 we see that Xα,βg ∈ L2(R, |x|2β+1dx), Fβ(Xα,βg) is bounded and Fβ(Xα,βg)(λ) = O ( λη−2(β−α) ) as λ→ 0+. Thus, in view of Remark 4.2, Xα,βg satisfies the admissibility condition (4.1) for γ = β. � Remark 5.3. In view of Remark 4.2, each function satisfying the conditions of Lemma 5.1 is a Dunkl wavelet of order α. Lemma 5.4. Let g be as in Lemma 5.2. Then for all f ∈ L1(R, |x|2β+1dx) we have Φβ Xα,βg(f)(a, b) = 1 a2(β−α) Xα,β [ Φα g ( tXα,βf ) (a, ·) ] (b). Proof. By Definition 4.5 we have Φβ Xα,βg(f)(a, b) = 1 a2β+2 f ∗β ˜(Xα,βg)a(b). But ˜(Xα,βg)a = Xα,β (g̃a) by virtue of (1.2) and (3.5). So using (3.8) we find that Φβ Xα,βg(f)(a, b) = 1 a2β+2 f ∗β [Xα,β (g̃a)] (b) = 1 a2β+2 Xα,β [ tXα,βf ∗α g̃a ] (b) = 1 a2(β−α) Xα,β [ Φα g ( tXα,βf ) (a, ·) ] (b), which gives the desired result. � Combining Theorems 4.10, 4.11 with Lemmas 5.2, 5.4 we get Theorem 5.5. Let g be as in Lemma 5.2. Then we have the following inversion formulas for the dual Dunkl–Sonine transform: (i) If both f and Fβf are in L1(R, |x|2β+1dx) then for almost all x ∈ R we have f(x) = 1 Cβ Xα,βg ∫ ∞ 0 (∫ R Xα,β [ Φα g ( tXα,βf ) (a, ·) ] (b) ( Xα,βg )β a,b (x)|b|2β+1db ) da a2(β−α)+1 . (ii) For f ∈ L1 ∩ L2(R, |x|2β+1dx) and 0 < ε < δ <∞, the function fε,δ(x) := 1 Cβ Xα,βg ∫ δ ε ∫ R Xα,β [ Φα g ( tXα,βf ) (a, ·) ] (b) ( Xα,βg )β a,b (x)|b|2β+1db da a2(β−α)+1 satisfies lim ε→0, δ→∞ ∥∥fε,δ − f ∥∥ 2,β = 0. Acknowledgements The author is grateful to the referees and editors for careful reading and useful comments. 12 M.A. Mourou References [1] de Jeu M.F.E., The Dunkl transform, Invent. Math. 113 (1993), 147–162. [2] Dunkl C.F., Differential-difference operators associated with reflections groups, Trans. Amer. Math. Soc. 311 (1989), 167–183. [3] Dunkl C.F., Integral kernels with reflection group invariance, Canad. J. Math. 43 (1991), 1213–1227. [4] Dunkl C.F., Hankel transforms associated to finite reflection groups, Contemp. Math. 138 (1992), 123–138. [5] Jouini A., Dunkl wavelets and applications to inversion of the Dunkl intertwining operator and its dual, Int. J. Math. Math. Sci. 6 (2004), 285–293. [6] Kamefuchi S., Ohnuki Y., Quantum field theory and parastatistics, Springer-Verlag, Berlin, 1982. [7] Mourou M.A., Trimèche K., Calderón’s formula associated with a differential operator on (0,∞) and inver- sion of the generalized Abel transform, J. Fourier Anal. Appl. 4 (1998), 229–245. [8] Mourou M.A., Trimèche K., Inversion of the Weyl integral transform and the Radon transform on Rn using generalized wavelets, Monatsh. Math. 126 (1998), 73–83. [9] Mourou M.A., Trimèche K., Calderon’s reproducing formula related to the Dunkl operator on the real line, Monatsh. Math. 136 (2002), 47–65. [10] Mourou M.A., Trimèche K., Transmutation operators and Paley–Wiener associated with a singular differential-difference operator on the real line, Anal. Appl. (Singap.) 1 (2003), 43–70. [11] Rosenblum M., Generalized Hermite polynomials and the Bose-like oscillator calculus, in Nonselfadjoint Operators and Related Topics (Beer Sheva, 1992), Oper. Theory Adv. Appl., Vol. 73, Birkhäuser, Basel, 1994, 369–396, math.CA/9307224. [12] Rösler M., Bessel-type signed hypergroups on R, in Probability Measures on Groups and Related Struc- tures, XI (Oberwolfach, 1994), Editors H. Heyer and A. Mukherjea, Oberwolfach, 1994, World Sci. Publ., River Edge, NJ, 1995, 292–304. [13] Soltani F., Lp-Fourier multipliers for the Dunkl operator on the real line, J. Funct. Anal. 209 (2004), 16–35. [14] Soltani F., Sonine transform associated to the Dunkl kernel on the real line, SIGMA 4 (2008), 092, 14 pages, arXiv:0812.4666. [15] Xu Y., An integral formula for generalized Gegenbauer polynomials and Jacobi polynomials, Adv. in Appl. Math. 29 (2002), 328–343. [16] Yang L.M., A note on the quantum rule of the harmonic oscillator, Phys. Rev. 84 (1951), 788–790. http://arxiv.org/abs/math.CA/9307224 http://arxiv.org/abs/0812.4666 1 Introduction 2 Preliminaries 3 The dual Dunkl-Sonine integral transform 4 Calderón's formula for the Dunkl continuous wavelet transform 5 Inversion of the dual Dunkl-Sonine transform using Dunkl wavelets References