I want to prove that $\mathbb{1}_{B_1(0)}$ doesn't have a $\partial_{x_i}$weak derivative in $\mathbb{R}^n$. So I tried a proof by contradiction. So assume that $v$ is its weak derivative w.r.t $x_i$. Then we have that for any $\phi \in C_c^{\infty}(\mathbb{R}^n) $

$-\int_{\mathbb{R}^n}v \phi = \int \mathbb{1}_{B_1(0)} \phi_{x_i} = \int_{B_1(0)} \phi_{x_i}$.

So I think now I should construct a sequence of functions in $C_c^{\infty}(\mathbb{R}^n)$ so that they are uniformly bounded (by, say 1) and they converge to $0$ point-wise a.e. And try to get a contradiction. But I don't know what to do with $\int_{B_1(0)} \phi_{x_i}$ for such functions.


There is a formula for integration by parts in $\mathbb R^n$: $$\int_\Omega f_{x_i}g = \int_{\partial \Omega} fg \, \nu_i - \int_\Omega fg_{x_i}$$ where $\nu_i$ is the $i^{\text{th}}$ component of the outward facing unit normal. Using this with $f = \phi$ and $g = 1$ shows that $$\int_{B_1(0)} \phi_{x_i} = 0$$ for all $\phi$ supported in $B_1(0)$. Supposing such $v$ exists, we would then have $$\int_{\mathbb R^n} v \phi = \int_{B_1(0)} v \phi = 0$$ for any smooth $\phi$ with support in $B_1(0)$ and thus $v = 0$ a.e. in $B_1(0)$. Now taking $\phi$ with compact support in $\mathbb R^n \setminus B_1(0)$, we will also have $$\int_{\mathbb R^n \setminus B_1(0)} v\phi = \int_{\mathbb R^n} v\phi = \int_{B_1(0)} \phi_{x_i} = 0$$ and thus $v = 0$ a.e. in $\mathbb R^n \setminus B_1(0)$ as well. Hence $v = 0$ a.e. in $\mathbb R^n$. Thus we must have $$\int_{B_1(0)} \phi_{x_i} =0$$ for all $\phi \in C_c^\infty(\mathbb R^n)$ which is clearly absurd. Hence such $v$ does not exist.

Here I have twice used that fact that for open $\Omega \subseteq \mathbb R^n$, if $v \in L^2(\Omega)$ and $$\int_{\Omega} v\phi = 0$$ for all $\phi \in C_c^\infty(\Omega)$, then $v =0$ a.e. in $\Omega$.


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.