2
$\begingroup$

Let $B_0(1)$ be the unit ball in $\mathbb R^n$, $n\geq2$. $h\in W_0^{1,2}(B_0(1))$. For $r\in (0,1)$, define a function $f_r(x):[0,1]\rightarrow \mathbb R$ by \begin{equation} f_r(x):= \begin{cases} 1,&\text{when} \ x\in(0,r],\\ \frac{1-x}{1-r},& \text{when} \ x\in(r,1]. \end{cases} \end{equation} Let $g_r(x):B_0(1)\rightarrow \mathbb R$ to be $g_r(x):=f_r(|x|)\cdot h(x)$.

Question: Can we prove that $\int_{B_0(1)\backslash B_0(r)}|\nabla g_r|^2dx\to0$, as $r\to1^-$?

$\endgroup$
2
$\begingroup$

Yes, this works. By computing the gradient with the product rule, this boils down to showing that $$ \frac{1}{\delta^2} \int_{1-\delta\le |x|\le 1} |h(x)|^2 \to 0 $$ as $\delta\to 0$ for $h\in W^{1,2}_0$. If $h$ is also smooth, then, since $h(y)=0$ on $|y|=1$, $$ h(x) = -\frac{x}{|x|}\cdot\int_{|x|}^1 \nabla h(tx/|x|)\, dt . $$ So $|h(x)|^2\le (1-|x|)\int_{|x|}^1 |\nabla h|^2\, dt$, and if we now integrate this over $1-\delta\le |x|\le 1$ (in spherical coordinates), we find that $$ \int_{1-\delta\le |x|\le 1} |h(x)|^2\, dx \lesssim \int_{1-\delta}^1 (1-r)\, dr \int_{1-\delta\le |x|\le 1} |\nabla h|^2\, dx , $$ and this gives the claim because the first integral on the RHS is $O(\delta^2)$ and the second one is $o(1)$ as $\delta\to 0$. The general case now follows by approximation.

$\endgroup$
  • $\begingroup$ The second quantity $q_h(\delta)$ is $o_h(1)$. Is it true that $q_h(\delta)\to 0$ uniformly in $h$? $\endgroup$ – Liviu Nicolaescu Dec 24 '15 at 17:55
  • $\begingroup$ @LiviuNicolaescu: The $W^{1,2}$ norm controls $\|\nabla h\|_2$, that is enough to be able to carry out the approximation at the end. $\endgroup$ – Christian Remling Dec 24 '15 at 18:04
  • $\begingroup$ I realized that after I wrote the comment but I was not able to delete it $\endgroup$ – Liviu Nicolaescu Dec 24 '15 at 20:27
  • $\begingroup$ Small detail: shouldn't the integral in the second equation be from $1$ to $|x|^{-1}$ ? $\endgroup$ – Pietro Majer Dec 25 '15 at 20:00
  • $\begingroup$ @PietroMajer: Yes, that's right, corrected, thanks. $\endgroup$ – Christian Remling Dec 25 '15 at 21:41

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.