On the isoperimetric profile for a mixed Euclidean-log-convex measure (Q1687599)

The classical isoperimetric problem is to find the sets with minimal boundary measure under a volume constraint or, equivalently, the sets with maximal volume and given boundary measure, that are called isoperimetric sets. The typical approach to this problem is to show the existence of minimizers in the class of sets with finite perimeter and to describe the solutions via suitable symmetrization techniques in the classical case of Lebesgue measure on \(\mathbb R^n\). As a particular case, mixed densities can be considered by decomposing \(\mathbb R^n\) as a Cartesian product \(\mathbb R^n=\mathbb R^h\times\mathbb R^k\), that is, measures \(\lambda\) on \(\mathbb R^n\) arising as product measures on the factors, \(\lambda=\mu\otimes\nu\), with \(\mu\), \(\nu\) measures on \(\mathbb R^h\), \(\mathbb R^k\), respectively. In [J. Funct. Anal. 260, No. 12, 3678--3717 (2011; Zbl 1222.49058)], the first author and \textit{F. Maggi} and \textit{A. Pratelli} considered the isoperimetric problem in such a mixed framework with \(\nu={\mathcal L}^k\) the Lebesgue measure on \(\mathbb R^k\) and \(\mu\) the standard Gaussian measure on \(\mathbb R^h\). In the present paper, the authors consider a mixed density \(\lambda=\mu\times{\mathcal L}^k\), where \(\mu\) is a Log-convex measure on \(\mathbb R^h\). They consider the Euclidean space \(\mathbb R^n=\mathbb R^h\times\mathbb R^k\) whose points are denoted by \(z=(x,y)\), endowed with the measure \(\lambda=e^{\psi(|x|)}dx dy\), where \(\psi:\mathbb R\to\mathbb R\) is a convex, even function, and the isoperimetric problem is \[ \inf\left\{\int_{\partial E}e^{\psi(|x|)}d{\mathcal H}^{n-1}(x,y);\;\int_E\lambda=m>0\right\}. \] As the density of \(\lambda\) with respect to the Lebesgue measure is regular, the class \({\mathcal P}_\lambda(\mathbb R^n)\) of sets with locally finite perimeter with respect to \(\lambda\) is the same as the classical nonweighted one, so the isoperimetric problem in this class becomes \(\inf\left\{P_\lambda(E);\;\lambda(E)=m\right\}\), where \[ P_\lambda(E)=\sup\left\{\int_E\text{div}\,F\,d\lambda;\;F\in C^1_{c}(\mathbb R^n,\mathbb R^n) \text{ and } \;\|F\|_\infty\leq 1\right\}. \] The main results of this paper concern existence, geometric properties, and uniqueness of the isoperimetric sets. The authors show that a symmetric isoperimetric set exists, up to translations along the \(y\) directions, and they prove that the isoperimetric set is unique provided \(\min\{h,k\}=1\) and that if \(k=1\) and the mass \(m\) is small enough, or if \(h=1\) the isoperimetric set is strictly convex.