1
$\begingroup$

Suppose $E\subset\mathbb{R}$ has finite Lebesgue measure and $\varphi\in L^1(\mathbb{R})$. Show that $\lim\limits_{t\rightarrow\infty}\int\limits_{E}\varphi(x+t)dx=0$.

I guess first I have to show $\lim\limits_{t\rightarrow\infty}\varphi(x+t)=0$ for $x\in\mathbb{R}$ and use the Lebesgue Dominated Convergence Theorem but I am not sure.

$\endgroup$
  • 1
    $\begingroup$ Maybe change of variables $x+t=\tau$ and use the integral convergence at infinity. $\endgroup$ – thing Dec 25 '19 at 1:36
  • 1
    $\begingroup$ Can you prove it for an interval first? $\endgroup$ – copper.hat Dec 25 '19 at 1:43
3
$\begingroup$

Given $\epsilon>0$, choose a $\psi\in C_{00}$ such that $\|\varphi-\psi\|_{L^{1}(\mathbb{R})}<\epsilon$, then \begin{align*} \int_{E}|\varphi(x+t)|dx&\leq\int_{E}|\varphi(x+t)-\psi(x+t)|dx+\int_{E}|\psi(x+t)|dx\\ &\leq\|\varphi-\psi\|_{L^{1}(\mathbb{R})}+\int_{E}|\psi(x+t)|dx. \end{align*} Now $|\psi(x+t)|\chi_{E}\leq\|\psi\|_{L^{\infty}}\chi_{E}\in L^{1}(\mathbb{R})$, apply Lebesgue Dominated Convergence Theorem to make the term $\displaystyle\int_{E}|\psi(x+t)|dx$ arbitrarily small as $t\rightarrow\infty$.

$\endgroup$
  • $\begingroup$ Does $C_{00}$ denote the set of continuous functions that vanish outside a bouned set? $\endgroup$ – user682705 Dec 25 '19 at 2:21
  • 1
    $\begingroup$ Yes, or you can say compact supported continuous functions. $\endgroup$ – user284331 Dec 25 '19 at 2:22
1
$\begingroup$

Note that $\int \phi(x+t)1_E(x) d x = \int \phi(x)1_E(x-t) d x$

If $E$ is a bounded set then $\lim_{t \to \infty} 1_E(x-t) = 0$ and so $\int \phi(x)1_E(x-t) d x = 0$.

Since $\phi$ is integrable, for any $\epsilon>0$, there is some $\delta>0$ such that if $mA < \delta$ then $\int_A|\phi| < \epsilon$ (absolute continuity).

Suppose $\epsilon>0$. Choose the $\delta>0$ as above.

If $E$ has bounded measure, then we can write $E = B \cup C$, where $B$ is bounded and $mC < \delta$.

Then $\int \phi(x+t)1_E(x) d x = \int \phi(x+t)1_C(x) d x + \int \phi(x)1_B(x-t) d x< \epsilon + \int \phi(x)1_B(x-t) d x$ and there is some $T$ such that if $t>T$ we have $\int \phi(x)1_B(x-t) d x < \epsilon$. Hence $\int \phi(x+t)1_E(x) d x \to 0$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.