Given two functions, $f, g \in C_c^{\infty}[-1,1]$ is the set of smooth functions that are compactly supported on $[-1,1]$, I want to evaluate: $$ \lim_{n \to \infty} \int_{-n}^n f(x/n)g(x) \text{d}x. $$

If we let $h_n(x) = f(x/n)g(x)$, then $h_n(x) \to f(0) g(x)$ pointwise, can I find a dominating function and use the dominated convergence theorem here? I'm a bit confused since the limits of integration are variable, but since the support of both functions is on the interval $[-1,1]$ can I just ignore this?

  • 1
    $\begingroup$ Looks like a typo in the question somewhere, $g$ needs to be defined outside the interval $[-1,1]$ for the integral to make sense. Of course, I don't really know what $C_c^{\infty}[-1,1]$ means (what is that $c$ subscript?) $\endgroup$ – Michael Oct 6 '18 at 7:34
  • $\begingroup$ @Michael I've clarified in the question $\endgroup$ – dimebucker Oct 6 '18 at 7:38
  • $\begingroup$ I don't think anyone would write a problem like that and assume we know that $g$ is supposed to be zero outside. $\endgroup$ – Michael Oct 6 '18 at 7:39
  • $\begingroup$ @Michael Sorry I thought the $c$ subscript notation was standard $\endgroup$ – dimebucker Oct 6 '18 at 7:40
  • $\begingroup$ Standard for what? $\endgroup$ – Michael Oct 6 '18 at 7:40

Under the hypothesis that $\text{supp}(g)\subseteq[-1,1]$, then for all $n \geq 1$ $$ \int_{-n}^n f(x/n)g(x) dx = \int_{-1}^{1} f(x/n)g(x) dx, $$ and the answer is somewhat trivial. For a matter of commodity, suppose that $f(x) \geq 0$ and $g(x)\geq 0$ for all $x \in \mathbb{R}$. Since $f$ is continuous on a compact set, we can define $m = \min_{x\in[-1,1]}\{f(x)\}$ and $M = \max_{x\in[-1,1]}\{f(x)\}$. Then for all $n>1$ $$ m\int_{-1}^{1} g(x) dx\leq \int_{-1}^{1} f(x/n)g(x) dx \leq M\int_{-1}^{1} g(x) dx. $$ A similar answer applies if $\text{supp}(g)\subseteq[-k,k]$ for some $k \in \mathbb{N}$.

Some trouble arises if $\text{supp}(g)$ is not compact. For instance, let $g(x)=1$ for all $x \in \mathbb{R}$ and let $f(x)$ be your favorite smooth function such that

  • $f(x) \geq 0$ for all $x \in\mathbb{R}$;
  • $f(x)=1$ for all $x\in\mathbb{R}$ with $|x|\leq1/2$ and
  • $f(x)=0$ for all $x\in\mathbb{R}$ with $|x|\geq0$.

Then $$ \int_{-n}^n f(x/n)g(x) dx \geq \int_{-n/2}^{n/2} f(x/n)g(x) dx = \int_{-n/2}^{n/2} 1 \cdot 1 dx = n. $$ As a consequence, the sequence $n \mapsto \int_{-n}^{n} f(x/n)g(x) dx$ diverges.

  • $\begingroup$ Your second line of inequalities need fixing if $g$ is allowed to be negative. Though, I think the spirit of this answer is similar to my above comment that $f$ is bounded. $\endgroup$ – Michael Oct 6 '18 at 23:03
  • $\begingroup$ Thanks, I added the hypotheses that $f, g \geq 0$. $\endgroup$ – Emanuele Bottazzi Oct 7 '18 at 16:59
  • $\begingroup$ Looks good (+1). $\endgroup$ – Michael Oct 8 '18 at 2:29

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.