I wonder if the following definitions of the Gaussian surface measure are equivalent.

First, let $\mathbb{R}^n$ be the Euclidean space and $A \subseteq \mathbb{R}^n$ be a sufficiently regular set, e.g., has smooth boundary or is convex. We define the $\epsilon$-neighbor of $A$ in the $\ell_p$ sense as \begin{align} A^{\epsilon}_p = \{ y \in \mathbb{R}^n \colon \text{there exists}~ x \in A ~\text{such that}~ \| x - y \|_{p} \leq \epsilon \}, \end{align} where $1\leq p\ \leq + \infty$. We denote $\gamma$ as the standard $n$-dimensional Gaussian measure $N(0, I_n)$, then the Gaussian surface measure is defined as \begin{align} \tau (A) = \lim_{\epsilon \rightarrow 0} \frac{ \gamma( A^{\epsilon}_p \setminus A) }{ \epsilon}. \end{align}

Second, another definition of Gaussian surface measure is \begin{align} \tau (A) = \int _{\partial A} \varphi(x) \mathrm{d}\sigma(x), \end{align} where $\varphi$ is the density of $N(0, I_n)$ and $\mathrm{d}\sigma(x)$ is the standard surface measure in $\mathbb{R}^n$.

In ``On the maximal perimeter of a convex set in R n with respect to a Gaussian measure'' it is shown that these two notions are equivalent for $p =2$, that is, the neighbors are considered in the sense of $\ell_2$-balls.

I was wondering whether such an equivalence holds for all $p\in [1, +\infty]$ in general? If not, for what $p$ and what set $A$ can we establish such an equivalence?


No: unless $p = 2$, these definitions are not equivalent. The value that you define in the question for a rectangle $A$ whose sides are parallel to $y = \pm x$ is easily found to be $2^{1/2-1/p} \int_{\partial A} \varphi(x) d\sigma(x)$.

| cite | improve this answer | |

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.