1
$\begingroup$

suppose we have a sequence of continuous functions $(f_n)_{n\geq 1}$ with $f_n:[0,\infty)\rightarrow \mathbb{R}$ such that $\int_{0}^{\infty}\vert f_n(x) \vert<\infty$ for all $n\in\mathbb{R}$. Further, let $\lim\limits_{n\rightarrow \infty}f_n(x) = 0$ for all $x\geq 0$.

Is it true that $$\lim\limits_{n\rightarrow \infty}\int_{0}^{\infty}f_n(x)\cdot g(x) dx = 0,$$

if $g:[0,\infty)\rightarrow \mathbb{R}$ is a continuous and absolutely integrable function (i.e. $\int_{0}^{\infty}\vert g(x) \vert<\infty$)? Can I interchange the order of integration and the limit? (If needed, we may also assume that $\int_{0}^{\infty}\vert f_n(x)\cdot g(x) \vert<\infty$ for all $n$).

Best regards

$\endgroup$
1
$\begingroup$

If $f_n\to 0$ uniformly then the answer is affirmative (see the comments below). In general the answer is no, however. Here is a counter example. Consider the sequence $$f_n(x)=e^{-(x-n)^2+x}\;,$$ and let $g(x)=e^{-x}$. Then $f_n\to 0$ pointwise but not uniformly on $[0,\infty)$. By translation invariance of $dx$, we have $$\lim_{n\to \infty}\int_0^{\infty}e^{-(x-n)^2+x}e^{-x}dx=\lim_{n\to \infty}\int_0^{\infty}e^{-(x-n)^2}dx =\lim_{n\to\infty}\int_{-n}^{\infty}e^{-y^2}dy=\sqrt{\pi}$$

$\endgroup$
  • 2
    $\begingroup$ You do not really need uniform convergence. It suffices to have $|f_n (x)| \leq C $ for all $n $ and (almost) all $x $. $\endgroup$ – PhoemueX Mar 17 '17 at 8:26
  • $\begingroup$ Yes, I did not want to say anything against your counterexample. I only objected to the sentence "unless $f_n \to 0$ uniformly, the answer is no". This sounds as if the claim was always false, unless we have uniform convergence. But in fact, it suffices to have $|f_n(x)| \leq C$ (which is implied by uniform convergence). (BTW: +1) $\endgroup$ – PhoemueX Mar 17 '17 at 9:53
  • 1
    $\begingroup$ But if $|f_n (x)| \leq C$ for all $x, n$, then $|f_n (x) \cdot g(x)| \leq C \cdot |g(x)|$, where the right-hand side is in $L^1$. Also, $f_n (x) \cdot g(x) \to 0$ pointwise, so that the dominated convergence theorem yields $\int f_n (x) \cdot g(x) \, dx \to 0$. $\endgroup$ – PhoemueX Mar 17 '17 at 12:12
  • $\begingroup$ @PhoemueX Yes you're right. I made a mistake. My $f_n$'s are not uniformly bounded...thanks for pointing that out. $\endgroup$ – user113529 Mar 17 '17 at 12:58

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.