• Aucun résultat trouvé

Tempered distributions and Sobolev spaces

N/A
N/A
Protected

Academic year: 2022

Partager "Tempered distributions and Sobolev spaces"

Copied!
2
0
0

Texte intégral

(1)

UPMC Basic functional analysis

Master 1, MM05E 2011-2012

Tempered distributions and Sobolev spaces

1) Topology of the Schwartz space.We recall that a functionf ∈ S(RN)if for any multi-indexesαandβ∈NN,

sup

x∈RN

|xαDβf(x)|<+∞.

For each n∈Nandf ∈ S(RN), define

pn(f) := sup

α,β∈NN,|α|≤n,|β|≤n

sup

x∈RN

|xαDβf(x)|.

For allf andg∈ S(RN), let

d(f, g) :=X

n∈N

1 2n

pn(f−g) 1 +pn(f−g).

Show thatdis a distance onS(RN)whose induced topology defines a complete metric space.

2) Compute the Fourier transform of the following tempered distributions : a)δ0.

b)1.

c)δ00, δ000, . . . , δ(k)0 . d)δa.

e)x7→e−2iπax.

3) Letµ∈ M(RN)be a bounded Radon measure inRN. Show that its Fourier transform is given by the function

Fµ(ξ) = Z

RN

e−2iπξ·xdµ(x) for allξ∈RN.

4) Cauchy principal value. For allx∈R, definef(x) := ln|x|.

a) Show thatf ∈L1loc(R).

b) Show that the mapping

T :ϕ∈ S(R)7→

Z

R

ϕ(x) ln|x|dx∈R

defines a tempered distribution.

c) Show that the derivative ofT is theprincipal valueof1/x, which is defined by the tempered distribution

pv 1

x

:ϕ∈ S(R)7→lim

ε→0

Z

{|x|>ε}

ϕ(x) x dx.

5) Hilbert transform.For allu∈ S(R), we denote byHu∈ S(R)theHilbert transformofudefined by

Hu:= 1 πu∗pv

1 x

.

a) Show that for everyx∈R,

(Hu)(x) = lim

ε→0

1 π

Z

{|x−y|>ε}

u(y) x−ydy.

1

(2)

b) Show that its Fourier transform is given by

Hu(ξ) =d −isign(ξ)ˆu(ξ) for allξ∈R. Deduce that for everyu∈ S(R),

kHukL2(R)=kukL2(R)

c) Deduce thatH extends to an isometry fromL2(R)toL2(R), and thatH2=−I, whereIis the identity mapping overL2(R).

6) Letχ∈ Cc (R)be such that χ= 1 in a neighborhood of0.

a) Show that any functionϕ∈ S(R)admits the following decomposition

ϕ(x) =ϕ(0)χ(x) +xψ(x) ∀x∈R,

for someψ∈ S(R).

b) SolvexT = 0inS0(R).

c) SolvexT = 1in S0(R).

7) We recall that a tempered distributionu∈ S0(RN)belongs to theSobolev spaceHs(RN)(withs∈R) if and only ifuˆ∈L2loc(RN)and

Z

RN

(1 +|ξ|2)s|ˆu(ξ)|2dξ <∞.

Show thatδ0∈Hs(RN)if and only ifs <−N/2.

8) The aim of this exercise is to show that for each s≥0, (Hs(RN))0 can be identified to H−s(RN). More precisely, for each T ∈(Hs(RN))0 there exists a uniqueu∈H−s(RN)such that

hT, vi(Hs(RN))0,Hs(RN)= Z

RN

ˆ

uˆv dξ for everyv∈Hs(RN),

and

kTk(Hs(RN))0 =kukH−s(RN). – Show that ifu∈H−s(RN)andv∈Hs(RN), thenuˆˆv∈L1(RN).

– Deduce that the mappingv7→R

RNuˆˆv dξ defines an element of(Hs(RN))0.

– Show that the Fourier transform F is an isometrical isomorphism from Hs(RN)to L2(RN, µ), where µis the absolutely continuous Radon measure(1 +|ξ|2)sLN.

– LetT ∈(Hs(RN))0, and define T˜:=T ◦ F−1. Deduce thatT˜ is a linear continuous map onL2(RN, µ).

– Conclude.

2

Références

Documents relatifs

UPMC Basic functional analysis.. Master 1,

UPMC Basic functional analysis. Master 1,

Namely, we characterize Sobolev and Hardy-Sobolev spaces in terms of maximal functions and provide an atomic decomposition for Hardy-Sobolev spaces.. We also investigate the

1.100 Lemma. Assume that Ω is smooth and bounded.. T is linear, and the above theorem implies that T is continuous and onto. Then the last conclusion in b) follows from the open

Montrer qu’un morphisme de revˆ etement est une application ouverte..

The measures usually considered are regu- lar Borel measures, those defined on the Borel sets and having the additional properties : (i) the measure of each compact set is finite;

The following Theorem 1.4 tells us then that Theorem 1.2 is still true for real Lie groups except for one list of classical homogeneous spaces and three exceptional

We prove that the regular representation of G in L 2 (G/H ) is tempered if and only if the orthogonal of h in g contains regular elements by showing simultaneously the equivalence