On the Maximal Gaussian Perimeter of Convex Sets, Revisited
Shivam Nadimpalli, Caleb Pascale · 2025
Read On the Maximal Gaussian Perimeter of Convex Sets, Revisited by Shivam Nadimpalli, Caleb Pascale (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-20079.txt.
Introduction
Shivam Nadimpalli and Caleb Pascale revisit a known construction in a short 2025 mathematics paper. The topic is the Gaussian perimeter of convex sets, a measure of surface area weighted by the normal bell-shaped distribution. An earlier construction by Nazarov gave a convex set whose weighted surface area comes close to the maximum possible.
The authors give a fresh analysis of that construction. Their argument rests on a notion called convex influence, which tracks how much single coordinates sway a function. The new viewpoint aims to make the near-maximal example easier to understand. Short papers of this kind matter because clean proofs often open the way to new results in high-dimensional geometry and probability.
Catalog record
| CREATOR | Shivam Nadimpalli, Caleb Pascale |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-20079
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the Maximal Gaussian Perimeter of Convex Sets, Revisited Shivam Nadimpalli, Caleb Pascale Abstract We revisit Nazarov’s construction [ Naz03 ] of a convex set with nearly maximal Gaussian surface area and give an alternate analysis based on the notion of convex influence introduced in [ DNS21 , DNS22 ] . 1 Introduction The Gaussian surface area of a (Borel) set K\subseteq\mathbb{R}^{n} is defined as \mathbf{GSA}(K)\mathrel{\mathop{\mathchar 58\relax}}=\lim_{\delta\to 0}\frac{\mathrm{Vol}(K_{t}\setminus K)}{t}\,, where K_{t} is the t -enlargement of K (that is, K_{t}=\{x\in\mathbb{R}^{n}\mathrel{\mathop{\mathchar 58\relax}}\exists y\in K~\text{s.t.}~\|x-y\|\leq t\} ) and \mathrm{Vol}(K) denotes the measure of K under the n -dimensional standard Gaussian distribution N(0,I_{n}) . It is a fundamental complexity measure in high-dimensional geometry, arising naturally in the context of noise stability and isoperimetry [ Bor75 , ST78 ] . It has also found applications to problems in probability theory [ Ben86 , Ben03 , Rai19 ] and theoretical computer science [ KOS08 , Kan10 , DMN19 , DNS24 ] . For convex sets (and more generally, sets with smooth boundary), the above definition coincides with the following (see [ Naz03 ] ): Definition 1 . If K\subseteq\mathbb{R}^{n} is convex, then \mathbf{GSA}(K)=\int_{\partial K}\varphi_{n}(x)\,d\sigma(x)\,, where d\sigma(\cdot) denotes the surface measure in \mathbb{R}^{n} and \varphi_{n}(x)=(2\pi)^{-n/2}e^{-\|x\|^{2}/2} is the density of the n -dimensional standard Gaussian measure. Motivated by problems in high-dimensional probability, Ball [ Bal93 ] considered the following reverse isoperimetric problem for the Gaussian measure: how large can \mathbf{GSA}(K) be for a convex set K\subseteq\mathbb{R}^{n} ? It will be convenient to write \Gamma(n)\mathrel{\mathop{\mathchar 58\relax}}=\sup{\left\{\mathbf{GSA}(K)\mathrel{\mathop{\mathchar 58\relax}}\text{convex}~K\subseteq\mathbb{R}^{n}\right\}}\,. Ball established the following uniform bound on the Gaussian surface area of convex sets: Theorem 2 (Theorem 4 of [ Bal93 ] ) . We have \Gamma(n)\leq 4n^{1/4} . This order of growth is sharp: Nazarov [ Naz03 ] showed the existence of a convex set K\subseteq\mathbb{R}^{n} with surface area on the order of \Omega(n^{1/4}) for n sufficiently large. Nazarov’s construction has since found applications in learning theory [ KOS08 ] , polyhedral approximation [ DNS24 ] , and property testing [ CDN+25 ] . Theorem 3 ( [ Naz03 ] ) . We have \liminf_{n\to\infty}\frac{\Gamma(n)}{n^{1/4}}\geq e^{-5/4}\,. Nazarov [ Naz03 ] lamented that his proof of Theorem 3 (and of his sharpening of Theorem 2 ; see below) was “a pretty boring and technical computation” and speculated that “there should exist some simple and elegant way leading to the result,” even encouraging the reader to “stop reading and (try to) prove the theorem themselves.” This note attempts to give such a proof. In particular, our argument avoids the heavier analytic machinery (such as the Laplace asymptotic formula) and technical estimates of Nazarov’s original proof. Our key idea is to analyze an isoperimetric quantity closely related to Gaussian surface area, namely convex influence (introduced in [ DNS21 , DNS22 ] and recalled below in Section 2.2 ), which is considerably easier to calculate. We note that a gap remains between the best known upper and lower bounds for \Gamma(n) . Nazarov [ Naz03 ] himself sharpened Ball’s bound ( Theorem 2 ), and a subsequent refinement by Raič [ Rai19 ] obtained the following estimate, which, to the best of our knowledge, is the current state of the art: \Gamma(n)\leq\sqrt{\frac{2}{\pi}}+0.59(n^{1/4}-1)\,. Theorem 3 continues to be the best known lower bound on \Gamma(n) ; note that e^{-5/4}\approx 0.2865 . Organization. We recall useful preliminaries in Section 2 and present our proof of Theorem 3 in Section 3 . 2 Preliminaries We write [n]\mathrel{\mathop{\mathchar 58\relax}}=\{1,\dots,n\} . We use boldfaced letters (such as \boldsymbol{x} , \boldsymbol{v} , and \boldsymbol{K} ) to denote random variables (which may be real-valued, vector-valued, or set-valued; the intended type will be clear from the context. We write \boldsymbol{x}\sim\mathcal{D} to indicate that the random variable \boldsymbol{x} is distributed according to probability distribution \mathcal{D} . Finally, \mathbb{S}^{n-1}=\{x\in\mathbb{R}^{n}\mathrel{\mathop{\mathchar 58\relax}}\|x\|=1\} will denote the unit sphere in \mathbb{R}^{n} . 2.1 Distributions and Tail Bounds We write N(0,I_{n}) for the n -dimensional standard Gaussian distribution over \mathbb{R}^{n} with density \varphi_{n}(\cdot) . When the dimension n is clear from context, we will sometimes write \varphi instead of \varphi_{n} . The Gaussian measure of a (measurable) set K\subseteq\mathbb{R}^{n} will be denoted by \mathrm{Vol}(K) , i.e., \mathrm{Vol}(K)\mathrel{\mathop{\mathchar 58\relax}}=\mathop{{\bf Pr}\/}_{\boldsymbol{x}\sim N(0,I_{n})}{\left[\boldsymbol{x}\in K\right]}\,. We recall a standard concentration bound for univariate Gaussian random variables: Proposition 4 (Proposition 2.1.2 of [ Ver18 ] ) . Let \boldsymbol{x}\sim N(0,1) . Then for all t>0 , we have {\left(\frac{1}{t}-\frac{1}{t^{3}}\right)}\cdot\frac{1}{\sqrt{2\pi}}e^{-t^{2}/2}\leq\operatorname{{\bf Pr}}{\left[\boldsymbol{g}\geq t\right]}\leq\frac{1}{t}\cdot\frac{1}{\sqrt{2\pi}}e^{-t^{2}/2}\,. We will also require the following concentration bound on the norm of a Gaussian random vector: Proposition 5 (Equation 3.3 of [ Ver18 ] ) . Let t>0 . We have \mathop{{\bf Pr}\/}_{\boldsymbol{x}\sim N(0,I_{n})}{\left[{\left|\|\boldsymbol{x}\|-\sqrt{n}\right|}\geq t\right]}\leq 2e^{-Ct^{2}}\,, where C is an absolute constant. Let \mathbb{S}^{n-1}\mathrel{\mathop{\mathchar 58\relax}}=\{x\mathrel{\mathop{\mathchar 58\relax}}\|x\|=1\} denote the unit sphere in \mathbb{R}^{n} endowed with the uniform measure. It is known (see, for example, Appendix D of [ KOS08 ] ) that for any x\in\mathbb{R}^{n} and r\geq 0 , we have \mathop{{\bf Pr}\/}_{\boldsymbol{v}\sim\mathbb{S}^{n-1}}{\left[x\cdot\boldsymbol{v}\leq r\right]}=\tau_{n}\int_{-1}^{\min{\left\{\frac{r}{\|x\|},1\right\}}}(1-z^{2})^{\frac{(n-3)}{2}}\,dz\qquad\text{where}~\tau_{n}\mathrel{\mathop{\mathchar 58\relax}}=\frac{\Gamma{\left(\frac{n}{2}\right)}}{\sqrt{\pi}\cdot\Gamma{\left(\frac{n-1}{2}\right)}}\,. (1) Stirling’s approximation (or alternatively, Gautschi’s inequality [ Gau59 ] ) yields the following: \sqrt{\frac{n}{2\pi}}(1+O(n^{-1}))\leq\tau_{n}\leq\sqrt{\frac{n}{2\pi}}\,. (2) We will also require the following estimate on the measure of spherical caps: Lemma 6 . Let x\in\mathbb{R}^{n} and r\geq 0 such that \|x\|\geq r . Then \mathop{{\bf Pr}\/}_{\boldsymbol{v}\sim\mathbb{S}^{n-1}}{\left[x\cdot\boldsymbol{v}\leq r\right]}\geq 1-\tau_{n}\cdot\frac{\|x\|}{r(n-3)}\cdot e^{-\frac{r^{2}(n-3)}{2\|x\|^{2}}}\,. Standard estimates on the measure of spherical caps—see, for example, [ Bal97 , Tko12 ] —obtain the coarser bound \mathop{{\bf Pr}\/}_{\boldsymbol{v}\sim\mathbb{S}^{n-1}}{\left[x\cdot\boldsymbol{v}\leq r\right]}\geq 1-e^{-\frac{nr^{2}}{2\|x\|^{2}}}\,. Lemma 6 is sharper when \|x\|\ll r\sqrt{n} ; this is exactly the regime that will arise later in our proof of Theorem 3 . Proof. Thanks to Equation 1 , we have \displaystyle\mathop{{\bf Pr}\/}_{\boldsymbol{v}\sim\mathbb{S}^{n-1}}{\left[x\cdot\boldsymbol{v}>r\right]} \displaystyle=\tau_{n}\int_{\frac{r}{\|x\|}}^{1}(1-z^{2})^{\frac{(n-3)}{2}}\,dz \displaystyle\leq\tau_{n}\int_{\frac{r}{\|x\|}}^{\infty}e^{\frac{-z^{2}(n-3)}{2}}\,dz ( 1+y\leq e^{y} ) \displaystyle=\tau_{n}\cdot\sqrt{\frac{2\pi}{n-3}}\cdot\mathop{{\bf Pr}\/}_{\boldsymbol{y}\sim N(0,1)}{\left[\boldsymbol{y}>\frac{r\sqrt{n-3}}{\|x\|}\right]} \displaystyle\leq\tau_{n}\cdot\frac{\|x\|}{r(n-3)}\cdot e^{-\frac{r^{2}(n-3)}{2\|x\|^{2}}}\,. ( Proposition 4 ) The result follows immediately. ∎ 2.2 Convex Influence The following quantity was introduced in [ DNS21 , DNS22 ] as a convex set analogue of the well-studied notion of total influence from the analysis of Boolean functions [ O’D14 ] : Definition 7 (Total convex influence) . Given a convex set K\subseteq\mathbb{R}^{n} , we define its total convex influence \mathbf{I}[K] as \mathbf{I}[K]\mathrel{\mathop{\mathchar 58\relax}}=\frac{d}{dt}\mathrm{Vol}{\left((1+t)K\right)}\bigg|_{t=0}\,. We will also make use of the following alternative characterization of total convex influence from [ DNS24 ] , and give a self-contained proof for completeness: Lemma 8 (Lemma 67 of [ DNS24 ] ) . Given a convex set K\subseteq\mathbb{R}^{n} , we have \mathbf{I}[K]=\int_{\partial K}(x\cdot\nu_{x})\,\varphi(x)\,d\sigma(x) where \nu_{x} denotes the unit normal to \partial K at x and d\sigma(\cdot) denotes the surface measure. Proof. We have \displaystyle\mathbf{I}[K]=\frac{d}{dt}{\left(\int_{(1+t)K}\varphi(x)\,dx\right)}\bigg|_{t=0} \displaystyle=\frac{d}{dt}{\left(\int_{K}(1+t)^{n}\,\varphi((1+t)y)\,dy\right)}\bigg|_{t=0} \displaystyle=\int_{K}{\left(\frac{d}{dt}(1+t)^{n}\,\varphi((1+t)y)\right)}\bigg|_{t=0}\,dy (Fubini) \displaystyle=\int_{K}(n-\|y\|^{2})\,\varphi(y)\,dy (3) \displaystyle=\int_{K}\mathcal{L}\frac{\|y\|^{2}}{2}\,\varphi(y)\,dy\,, where, for a function f\mathrel{\mathop{\mathchar 58\relax}}\mathbb{R}^{n}\to\mathbb{R} , we define \mathcal{L}f(x)\mathrel{\mathop{\mathchar 58\relax}}=\Delta f(x)-\nabla f(x)\cdot x . Note that \mathcal{L} is the Ornstein–Uhlenbeck operator (cf. Definition 11.24 and Proposition 11.26 of [ O’D14 ] ). Integrating by parts then gives \mathbf{I}[K]=\int_{\partial K}(x\cdot\nu_{x})\,\varphi(x)\,d\sigma(x)\,. as desired. ∎ Note that the proof of Lemma 8 does not require convexity of K and in fact applies more generally to any set with finite Gaussian surface area. We additionally note that the expression in Equation 3 is the original definition of convex influence from [ DNS21 , DNS22 ] ; we record a few easy consequences of it that rely on Hermite analysis (see Appendix A or Chapter 11 of [ O’D14 ] ). Equation 3 can be rewritten as \mathbf{I}[K]=\mathop{{\bf E}\/}_{\boldsymbol{x}\sim N(0,I_{n})}{\left[K(\boldsymbol{x})(n-\|\boldsymbol{x}\|)^{2}\right]}\,, where we identify K\subseteq\mathbb{R}^{n} with its 0/1