Scribd
Upload
95 views2 pages

Math 202B Solutions: Assignment 7 D. Sarason

This document contains solutions to several problems regarding properties of convolutions and functions in Lp spaces. It proves that: 1) The convolution of functions f in Lp and g in Lp is bounded and continuous. 2) The convolution of a C∞ function with compact support and a function in Lp results in a C∞ function. 3) C∞ functions with compact support that are dense in Lp. 4) Sets of finite positive measure subtracted from themselves contain neighborhoods of the origin. 5) The algebra L1 has no identity function.

Uploaded by

Super Nezh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
95 views2 pages

Math 202B Solutions: Assignment 7 D. Sarason

This document contains solutions to several problems regarding properties of convolutions and functions in Lp spaces. It proves that: 1) The convolution of functions f in Lp and g in Lp is bounded and continuous. 2) The convolution of a C∞ function with compact support and a function in Lp results in a C∞ function. 3) C∞ functions with compact support that are dense in Lp. 4) Sets of finite positive measure subtracted from themselves contain neighborhoods of the origin. 5) The algebra L1 has no identity function.

Uploaded by

Super Nezh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Math 202B Solutions

Assignment 7
D. Sarason
0
25. Let f be in Lp (λN ) and g in Lp (λN ). Prove that f ∗ g is bounded and continuous.
R
Proof: Clearly, for any x, |(f ∗ g)(x)| = RN f (y)g(x − y) dλN (y) ≤ kf kp kgkp0 , so f ∗ g is bounded. Also,
if Ta represents the operation of translation by a (i.e. Ta f (x) = f (x − a)), then Ta (f ∗ g) = (Ta f ) ∗ g. Proof:
Z Z
(f ∗ g)(x − a) = f (y)g(x − a − y) dy = f (y − a)g(x − y) dy = ((Ta f ) ∗ g)(x).
RN RN

Thus, if 1 ≤ p < ∞, then kTa (f ∗ g) − f ∗ gk∞ ≤ kTa f − f kp kgkp0 → 0 as a → 0, which implies that f ∗ g is
uniformly continuous. If p = ∞, use symmetry to reverse the roles of f and g.
26. (a) Let the function ψ : R → R be of class C ∞ and have compact support. Let the function g be in Lp (λ)
(1 ≤ p ≤ ∞). Prove that ψ ∗ g is of class C ∞ , with (ψ ∗ g)(n) = ψ (n) ∗ g for all positive integers n.
Proof: It will be enough to show that ψ ∗g is differentiable with derivative ψ 0 ∗g; the desired conclusion
will then follow by induction.
For δ 6= 0 define the function Dδ ψ by
ψ(x + δ) − ψ(x)
Dδ ψ(x) = .
δ
One easily verifies that
(ψ ∗ g)(x + δ) − (ψ ∗ g)(x)
= (Dδ ψ) ∗ g(x).
δ
We now have Dδ ψ → ψ 0 pointwise; also, by the mean value theorem, Dδ ψ(x) = ψ 0 (c) for some c
between x and x + δ. However, since ψ 0 is continuous and has compact support, it is uniformly
continuous. This implies that kDδ ψ − ψ 0 k∞ → 0 as δ → 0. Also, if ψ is supported on [−R, R]
and |δ| < 1, then Dδ ψ is supported on [−R − 1, R + 1]. If we let K = [−R − 1, R + 1], then
0
kDδ ψ − ψ 0 kp0 ≤ kDδ ψ − ψ 0 k∞ m(K)1/p → 0 for p0 < ∞ also.
Therefore,

(ψ ∗ g)(x + δ) − (ψ ∗ g)(x)
− (ψ ∗ g)(x) ≤ kDδ ψ − ψ 0 kp0 kgkp → 0 as δ → 0,
0


δ
which gives the desired conclusion.
(b) Use Part (a) to prove that C ∞ (R) ∩ Lp (λ) is dense in Lp (λ) for 1 ≤ p < ∞.
Proof: Take a nonnegative C ∞ function ψ of compact support such that kψk1 = 1. For example, let
(  
C exp x21−1 , |x| < 1
ψ(x) =
0, |x| ≥ 1,

where the constant C is chosen so that kψk1 = 1. For  > 0 let ψ (x) = 1 ψ( x ).
Let g be in Lp (λ) (1 ≤ p < ∞). By Theorem 15.1 in lecture we have ψ ∗ g → g in Lp norm. By part
(a), each function ψ ∗ g is of class C ∞ , giving the desired conclusion.
27. Let E be a Lebesgue measurable subset of RN of finite positive measure. Prove that the set

E − E = {x − y : x, y ∈ E}

contains a neighborhood of the origin.


Proof: Consider the convolution χE ∗ χ−E ; we calculate
Z Z
(χE ∗ χ−E )(a) = χE (t)χ−E (a − t) dλN (t) = χE (t)χE+a (t) dλN (t) = λN (E ∩ (E + a)).
RN RN

1
By problem 25, since χE ∈ L1 (λN ) and χ−E ∈ L∞ (λN ), this means that λN (E ∩ (E + a)) is a continuous
function of a. By hypothesis this function has value λN (E) > 0 at a = 0, so the function is positive for a
in some neighborhood of 0. In particular, for any such a, E ∩ (E + a) 6= ∅, so a ∈ E − E.
28. Prove that the algebra L1 (λ) has no identity (λ = Lebesgue measure on R).
Proof 1: Suppose, to the contrary, that there were a function e ∈ L1 (λ) such that e ∗ g = g for each
g ∈ L1 (λ). Let ψ = χ[0,1] , and for  > 0 let ψ (x) = 1 ψ( x ). Then by Theorem 15.1 in lecture, we have
ψ = e ∗ ψ → e in L1 (λ) as  → 0. However, this is impossible, since a straightforward calculation shows
that kψ2−n − ψ2−(n+1) k1 = 1, so the sequence (ψ2−n )∞ 1
n=1 is not Cauchy in L (λ).
Proof 2: If e ∈ L1 (λ) were an identity, then by problem 25 we would have e ∗ χ[0,1] = χ[0,1] is equal
to some continuous function almost everywhere. However, this is impossible since χ[0,1] clearly remains
discontinuous at 0 and 1 after any modification on a set of measure zero.

You might also like