Interpolation of weighted extremal functions (Q823830)

The author applies the notion of geodesics between relative extremal plurisubharmonic functions to the interpolation of compact subsets of \(\mathbb C^n\). Let \(K_j\), \(j\in\{0,1\}\), be polynomially convex, non-pluripolar compact subsets of a bounded hyperconvex domain \(\Omega \subset \mathbb C^n\), and let \(u_j=c_j\omega_{K_j}\), with \(c_j>0\), where \(\omega_{K_j}\) are the relative extremal functions of the sets \(K_j\). Denote by \(u_t\), for \(0<t<1\), the geodesic between \(u_0\) and \(u_1\). Let \(c_t=-\min\{u_t(z): z\in \Omega\}\) and \[ K_t=\{z\in \Omega: u_t(z)=-c_t\}. \] The main result of the paper is a generalization of [\textit{D. Cordero-Erausquin} and the author, Arch. Math. 113, No. 1, 63--72 (2019; Zbl 1429.32021)]. It is proved that \[ c_t\leq(1-t)c_0+tc_1 \] with equality if \(K_0\cap K_1\neq \emptyset\). Moreover, the function \[ t\to c_t^{n+1}\mathrm{Cap}(K_t) \] is convex, where \(\mathrm{Cap}\) is the Monge-Ampère capacity. In addition if the constants \(c_j\) satisfy \(c_0^{n+1}\mathrm{Cap}(K_0)=c_1^{n+1}\mathrm{Cap}(K_1)\), then the function \[ t\to \left(\mathrm{Cap}(K_t))\right)^{-\frac {1}{n+1}} \] is concave. A similar result in the case of Reinhardt sets \(K_j\) is also obtained. Finally, the above results are applied to convex geometry to develop the Brunn-Minkowski theory for volumes of (co)convex sets in \(\mathbb R^n\).