M.El Hamma,R.Daher and A.Akhlidj
LaboratoireTAGMD,FacultédesSciencesA?nChock,UniversitéHassanII,B.P5366 Maarif,Casablanca,Maroc
Abstract.Two estimates useful in applications are proved for the Fourier transform in the space L2(X),where X a symmetric space,as applied to some classes of functions characterized by a generalized modulus of continuity.
Key Words:Fourier transform,generalized continuity modulus,symmetric space.
In[2],Abilov et al.proved two useful estimates for the Fourier transform classic in the space of square integrable functions on certain classes of functions characterized by the generalized continuity modulus,using a translation operator.
In this paper,we prove the analog of Abilov’s results see[2]in the Fourier transform on rank 1 symmetric space.
Let X=G/K where G is a connected noncompact semisimple Lie group with finite center and real rank one and K is a maximal compact subgroup.The form of Cartan decomposition is defined by g=e+p,where e is the Lie algebra of K.And g=e+a+n is Iwasawadecomposition,where a isamaximalabeliansubalgebraof p and n isanilpotent subalgebra of g.The rank one condition is that dima=1.the nilpotent subalgebra n has root space decomposition n=nγ+n2γ,where γ and 2γ are the positive roots.Let mγand m2γbe the respective root space dimensions and set(mγ+2m2γ).Choose H0∈a such that γ(H0)=1.This allows identifying a with R by the map t∈R←?tH0∈a,and denote a?the real dual space of a.
Let G=NAK be the Iwasawa decomposition of the group G,and g,e,a,and n the respective Lie algebras of the groups G,K,A,and N.Denote by M the centralizer of the subgroup A in K and put B=k/M.Let dx be a G-invariant measure on X,and let db and dk be the respective normed K-invariant measure on B and K.
The finiteWeylgroupWactson a?.Supposethat∑ isthesetofboundedroots(∑?a?),∑+is the set of positive bounded roots,and a+={h∈a;α(h)>0 for α∈∑+}is the positive Weyl chamber.Leth,ibe the Killing form on the Lie algebra g.For λ ∈a?we denote by Hλthe vector in a such that λ(H)=hHλ,Hifor all H in a.Let
The dimension of X is equal to
We return to the case in which X=G/K is an arbitrary symmetric space.Given g∈G,denote by A(g)∈a the unique element satisfying
where u∈K and n1∈N.For x=gK∈X and b=kM∈B=K/M,we put
In terms of this decomposition,the invariant measure dx on X has the form
where ?(t)= ?(α,β)(t)=(2sinht)2α+1(2cosht)2β+1,α =(mγ+m2γ?1)/2 and β =(m2γ?1)/2.The Laplacian on X is denoted L and its radial part is given by
The spherical function on X is the unique radial solution to the equation
The spherical function is defined by
where jα(t)is a normalized Bessel function of the first kind.
Proof.See[3],Lemma 9.
Lemma 1.2.The following inequalities are valid for a Jacobi function φλ(t)(λ,t∈R+):
1.|φλ(t)|≤1.
2.|1?φλ(t)|≤t2(λ2+ρ2).
Proof.See[6],Lemmas 3.1-3.2.
Consider in the Hilbert space L2(X)=L2(X,dx)with the norm
The Fourier transform on X was introduced by S.Helgason(see[4,5]),and defined for all f∈L2(X)by the formula
The inverse the Fourier transform
where|W|is the order of the Weyl group,and dμ(λ)=|c(λ)|?2dλ with dλ is the element of the Euclidean measure on a?,and c(λ)is the Harish-Chandra function.
The Fourier transform is an isomorphism of the Hilbert space L2(X)onto the Hilbert space L2(×B,dμ(λ)db).
The Palancherel formula
Introduce the translation operator on X.Denote by d(x,y)the distance between x,y∈X and let
be the sphere of radius h>0 on X centered at x.Let dσx(y)be the(dimX?1)-dimensional area element of the sphere σ(x,h)and let|σ(h)|be the area of the whole sphere σ(x,h).
Let f∈L2(X),the translation operator Shis defined by
From[7],we have
Lemma 1.3.Let f(x)∈L2(X),then
Proof.See[6],Lemma 1.4.
The finite differences of the first and higher orders are defined as follows:
is the kth-order generalized continuity modulus of f∈L2(X).
and ψ(t)is an arbitrary function defined on[0,∞).
From formula(1.1),we have
and
By Parseval identity,we obtain
To make the formulas concise,we introduce the notation
Using the Lemma 1.3 and it is easy to show tha for f∈L2(X)
Theorem 2.1.For functions f(x)∈L2(X)in the class
where m=0,1,···,k=1,2,···,;c>0 is a fixed constant,and ψ(t)is any function defined on the interval[0,∞).
Proof.In the terms of jp(x)a normalized Bessel function of the first kind we have(see[1])
where Jp(x)is Bessel function of the first kind,which is related to jp(x)by the formula
From the formula(1.2),we have
Then,
Combining this with formulas(2.1c)and(2.2),we have
Or
Then
This completes the proof of theorem.
Theorem 2.2.Let ψ(t)=tν,where(ν>0),then
m=0,1,2,···,k=1,2,···,0<ν<2.
Proof.We prove sufficiency by Theorem 2.1:letand ψ(t)=tνthen
To prove necessity let
It is easy to show,that there exists a function f∈L2(X)such that Lf∈L2(X)and
Hence,by the Plancherel equality,we have
This integral is divided into two
where N=[h?1].We estimate them separately
i.e.,
Now,we estimate I1,by(2)in Lemma 1.2,we obtain
i.e.,
Combining the estimates for I1and I2gives
Acknowledgements
The authors would like to thank the referee for his valuable comments and suggestions.
Analysis in Theory and Applications2018年2期