Gaussian random projections of convex cones: approximate kinematic formulae and applications

arXiv2212.05545MaRDI QIDQ6420195FDOQ6420195

Authors: Qiyang Han, Huachen Ren 

Publication date: 11 December 2022

Abstract: Understanding the stochastic behavior of random projections of geometric sets constitutes a fundamental problem in high dimension probability that finds wide applications in diverse fields. This paper provides a kinematic description for the behavior of Gaussian random projections of closed convex cones, in analogy to that of randomly rotated cones studied in [ALMT14]. Formally, let $K$ be a closed convex cone in $mathbb{R}^n$, and $Ginmathbb{R}^{mimesn}$ be a Gaussian matrix with i.i.d. $mathcalN(0,1)$ entries. We show that $GKequivGmu:muinK$ behaves like a randomly rotated cone in $mathbb{R}^m$ with statistical dimension $mindelta(K),m$, in the following kinematic sense: for any fixed closed convex cone $L$ in $mathbb{R}^m$, �egin{align*} &delta(L)+delta(K)ll m, Rightarrow, Lcap GK = {0} hbox{ with high probability},\ &delta(L)+delta(K)gg m, Rightarrow, Lcap GK eq {0} hbox{ with high probability}. end{align*} A similar kinematic description is obtained for $G^{-1}Lequivmuinmathbb{R}^n:GmuinL$. The practical usefulness and broad applicability of the prescribed approximate kinematic formulae are demonstrated in a number of distinct problems arising from statistical learning, mathematical programming and asymptotic geometric analysis. In particular, we prove (i) new phase transitions of the existence of cone constrained maximum likelihood estimators in logistic regression, (ii) new phase transitions of the cost optimum of deterministic conic programs with random constraints, and (iii) a local version of the Gaussian Dvoretzky-Milman theorem that describes almost deterministic, low-dimensional behaviors of subspace sections of randomly projected convex sets.

This page was built for publication: Gaussian random projections of convex cones: approximate kinematic formulae and applications

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6420195)