• Aucun résultat trouvé

Lebesgue spaces

N/A
N/A
Protected

Academic year: 2022

Partager "Lebesgue spaces"

Copied!
2
0
0

Texte intégral

(1)

UPMC Basic functional analysis

Master 1, MM05E 2011-2012

Lebesgue spaces

In the sequel,(X,M, µ)stands for a measure space.

1) Interpolation inequality. Let1≤p≤q≤+∞.

a) Show that if f ∈Lp(X, µ)∩Lq(X, µ), thenf ∈Lr(X, µ)for allr∈[p, q], and that

kfkr≤ kfkαpkfk1−αq ,

whereα∈[0,1]is defined by 1r =αp +1−αq .

b) Show that ifµ(X)<+∞andf ∈Lp(X, µ), thenf ∈Lr(X, µ)for allr∈[1, p], and that there exists a constantC >0 (independent off) such that

kfkr≤Ckfkp.

2) Generalized Hölder inequality. Letf1∈Lp1(X, µ), . . . , fk∈Lpk(X, µ)be such that 1

p1 +· · ·+ 1 pk =: 1

r ≤1.

Show that the productQk

i=1fi belongs toLr(X, µ)and

k

Y

i=1

fi r

k

Y

i=1

kfikpi.

3) Let1≤p0<+∞.

a) Show that if f ∈Lp0(X, µ)∩L(X, µ), then

p→∞lim kfkp=kfk.

b) Let f ∈Lp(X, µ)for allp∈[p0,+∞)such thatkfkp→ ∞asp→ ∞. Show thatf /∈L(X, µ).

c) Letf ∈Lp(X, µ)for allp∈[p0,+∞)such thatf /∈L(X, µ). Show thatkfkp→ ∞asp→ ∞.

4) Continuity of the translation inLp(RN). LetX =RN,M=L(RN)be theσ-algebra of all Lebesgue measurable subsets ofRN, andµ=LN be the Lebesgue measure. Let1≤p <∞andf ∈Lp(RN). For each h∈RN, we define the translation off by

τhf(x) :=f(x−h) ∀x∈RN.

Show that

lim

|h|→0hf−fkp= 0.

5) Assume thatµis a probability measure, i.e.,µ(X) = 1. Letf :X →[0,+∞)be a function inL1(X, µ).

1

(2)

a) Using Hölder’s inequality, show that ifµ({f >0})<1, thenkfkp→0 asp→0.

b) Show that

p→0lim Z

X

fpdµ=µ({f >0}).

c) Show that for allp∈(0,1), and ally∈(0,+∞), then

|yp−1|

p ≤y+|logy|.

d) From now on, we assume thatf >0 onX, and thatlogf ∈L1(X, µ). Show that

p→0lim Z

X

fp−1 p dµ=

Z

X

logf dµ.

e) Show that

p→0limkfkp= exp Z

X

logf dµ

.

6) Jensen’s inequality. Assume thatµ is a probability measure,i.e., µ(X) = 1. Letϕ: (a, b)→Rbe a convex function (with−∞ ≤a < b≤+∞).

a) Show that

ϕ(t)−ϕ(s)

t−s ≤ ϕ(u)−ϕ(t) u−t whenevera < s < t < u < b.

b) Deduce thatϕis continuous, and that for eachs∈(a, b), there existsβs∈Rsuch that

ϕ(t)≥ϕ(s) +βs(t−s)

for everyt∈(a, b).

c) Letf :X →(a, b)such thatf ∈L1(X, µ). Show thatϕ◦f is measurable and that

ϕ Z

X

f dµ

≤ Z

X

ϕ◦f dµ.

7) Let1≤p <∞andp0=p/(p−1). Show that for everyu∈Lp(X, µ), then

kukp= sup Z

X

|uv|dµ:v∈Lp0(X, µ),kvkp0 ≤1

.

Ifp=∞andX isσ-finite, show that for everyu∈L(X, µ), then

kuk= sup Z

X

|uv|dµ:v∈L1(X, µ),kvk1≤1

.

We recall thatX isσ-finite is there exists an increasing sequence of measurable sets (Xn)n∈N⊂Msuch that X =S

n∈N andµ(Xn)<∞for each n∈N.

2

Références

Documents relatifs

Linear algebra: vector spaces, linear maps, inner products, dual space 0.3.. Measure theory: null sets, equality “almost everywhere”, integrable

UPMC Basic functional analysis.. Master 1,

Duality in Lebesgue spaces and bounded Radon measures. 1) Let Ω be an open subset of

UPMC Basic functional analysis. Master 1,

Quand t0 varie, on peut assimiler ce segment à une échelle qui glisse le long de AO et qui est alors toujours tangente à l'astroïde.. a Le champ U est déni sur le domaine D, borné

A now classical result in the theory of variable Lebesgue spaces due to Lerner [24] is that a modular inequality for the Hardy-Littlewood maximal function in L p(·) ( R n ) holds if

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

Inversement, ce théorème permet de retrouver que tout sous-espace vectoriel strict de R N a une mesure (de Lebesgue N -dimensionnelle) nulle: il suffit de prendre pour T la