Let $u$ be function in $\mathbb{R}^n$ such that $u: U \rightarrow \mathbb{R}$ (here $U \subset \mathbb{R}^n$), $u \in C^{\infty}(\overline{U})$. Suppose $m(\partial U) = 0$ and $U$ has a $C^1$ boundary. Is it true that $m(\partial \{z: u(z) = 0 \}) = 0$. Here $m$ is the usual lebesgue measure in $\mathbb{R}^n$ i.e $m(A) = \int_{A}1_{A}(x) dx$ and $\partial A$ denotes the boundary of $A$.


For any closed subset $F\subseteq \mathbb{R}^n$, there exists a smooth function $u : \mathbb{R}^n \to \mathbb{R}$ such that $F$ is the zero-set of $u$.

Now pick a compact set $F \subset U$ so that $\partial F$ has positive measure and let $u$ be any smooth function having $F$ as zero-set of $u$.

Proof of the first claim: We know that the bump function

$$ \varphi(x) = \begin{cases} \exp\left(-1/(1-\|x\|^2)\right), & \|x\| < 1 \\ 0, & \|x\| \geq 1 \end{cases} $$

is smooth. Now fix a closed set $F \subseteq \mathbb{R}^n$ and let $V = \mathbb{R}^n \setminus F$ be its complement. Then there exists a countable family of open balls $\{ B(x_k, r_k) : k = 1, 2, \cdots \}$ such that $V = \bigcup_{k=1}^{\infty} B(x_k, r_k)$. Using this, for each multi-index $\alpha \in \mathbb{N}_0^{n}$ with $|\alpha| = \alpha_1 + \cdots + \alpha_n$, we define

$$ u_{\alpha}(x) = \sum_{k=1}^{\infty} \frac{(r_k \wedge 1)^k}{k!} \cdot \frac{1}{r_k^{|\alpha|}} (\partial^{\alpha}\varphi)\left(\frac{x - x_k}{r_k} \right), $$

where $a \wedge b := \min\{a, b\}$. Then we make the following observations.

  • The sum defining $u_{\alpha}$ converges uniformly for each $\alpha \in \mathbb{N}_0^{n}$. So it follows that $u = u_0$ is infinitely differentiable with $\partial^{\alpha} u = u_{\alpha}$.

  • For each $x \in B(x_k, r_k)$, we have $u(x) \geq \frac{(r_k\wedge 1)^k}{k!} \phi\left(\frac{x-x_k}{r_k}\right) > 0$.

  • If $x \notin U$, then $u(x) = 0$ since each summand vanishes.

Combining altogether, $u$ is a non-negative smooth function such that $u(x) > 0$ if and only if $x \in U$. Therefore $F$ is exactly the zero-set of $u$.

  • $\begingroup$ I agree with your second part but I don't see why the first statement is true $\endgroup$ – acreativename Sep 15 at 1:10
  • $\begingroup$ @acreativename, Although it is a well-known fact, I added a proof to show some of the basic ideas behind the claim. $\endgroup$ – Sangchul Lee Sep 15 at 1:33
  • 1
    $\begingroup$ Sigh...I thought that this result would have been true and I would have used this to prove another question $\endgroup$ – acreativename Sep 15 at 1:39

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.