Let $u(x,y),v(x,y)\in L^2$. What can we say about $\int_{-\infty}^\infty \int_{-\infty}^\infty \frac{d}{dy}(uv) \, dy \, dx$? Does it equal zero? If so, why?

  • 1
    $\begingroup$ It's possible that the limit $\lim_{y\to +\infty}u(x,y)$ doesn't exist. $\endgroup$ Jun 20, 2012 at 12:35
  • $\begingroup$ Plus: your notation is confusing, I do not understand it. $\endgroup$
    – GEdgar
    Jun 20, 2012 at 13:41
  • $\begingroup$ I've changed the notation. $\endgroup$ Jun 20, 2012 at 17:07

2 Answers 2


The question as stated is not well-defined. This is true even in the case where $u = v$, when you are differentiating $(u(x,y))^2$ with respect to to $y$. There are plenty of examples of nonnegative $L^1$ functions $f(y)$ which are nowhere differentiable, so if you take $u(x,y)$ to be of the form $\chi_{[0,1]}(x)\sqrt{f(y)}$ for example, you won't be able to take the $y$ derivative.

If you are taking distributional derivatives, there is a way of making sense out of the question, but I'm not sure that's what you mean here.


I think your integral might not make sense. First of all, I would write


rather than

$$\left(u\left(x,y\right)v\left(x,y\right)\right)|_{-y=\infty}^{y=\infty}$$ but that's just style. More importantly, this limit might not exist. The fact that a function is in $L^1$ (like $uv$ in your case) does not imply that it has a limit (which, if any, has to be $0$) at infinity. Indeed, consider the function

$$f(x) = \begin{cases}n\qquad \text{ if }n-\dfrac{1}{n^3}\leq x\leq n\\0\qquad\text{else}\end{cases}$$ for $n\in \mathbb{N}\setminus\{0\}$. This function is in $L^1(\mathbb{R})$, since

$$\int_{\infty}^{+\infty}f(x)dx=\int_0^{+\infty}f(x)dx = \lim_{m\to\infty}\int_0^m f(x)dx =$$ $$=\lim_{m\to\infty}\sum_{n=1}^{m}\int_{n-\frac{1}{n^3}}^n ndx = \lim_{m\to\infty}\sum_{n=1}^{m}\dfrac{1}{n^2}=\sum_{n=1}^{\infty}\dfrac{1}{n^2}=\dfrac{\pi^2}{6}$$ However, $f$ does not have a limit as $x$ goes to infinity. All you can say for an integrable (over $\mathbb{R}$) function is that

$$\liminf_{|x|\to\infty}f(x)=0$$ which is not enough to compute the limit that appears in your question.

If your functions have more regularity, namely they are not just $L^2$ but also $H^1$, then your integral exists and it's equal to zero, since the derivative cannot grow at infinity, so there can't be spikes whose amplitude does not go to zero. As a consequence, also the $\limsup$ is zero, which tells you that the limit exists and it is zero.


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.