Coverage processes on spheres and condition numbers for linear programming

TL;DR

This paper presents new exact and upper bound results for coverage probabilities on spheres and analyzes the distribution and expected value of the condition number in linear programming feasibility problems, connecting geometric and optimization insights.

Contribution

It provides exact formulas and bounds for coverage probabilities on spheres and characterizes the condition number distribution for random linear programming problems, improving existing bounds.

Findings

Exact formula for coverage probability when alpha in [pi/2, pi]

Upper bounds for coverage probability when alpha in [0, pi/2]

Upper bound on the expected logarithm of the condition number: 2*log(m+1)+3.31

Abstract

This paper has two agendas. Firstly, we exhibit new results for coverage processes. Let $p(n,m,\alpha)$ be the probability that $n$ spherical caps of angular radius $\alpha$ in $S^m$ do not cover the whole sphere $S^m$. We give an exact formula for $p(n,m,\alpha)$ in the case $\alpha\in[\pi/2,\pi]$ and an upper bound for $p(n,m,\alpha)$ in the case $\alpha\in [0,\pi/2]$ which tends to $p(n,m,\pi/2)$ when $\alpha\to\pi/2$. In the case $\alpha\in[0,\pi/2]$ this yields upper bounds for the expected number of spherical caps of radius $\alpha$ that are needed to cover $S^m$. Secondly, we study the condition number ${\mathscr{C}}(A)$ of the linear programming feasibility problem $\exists x\in\mathbb{R}^{m+1}Ax\le0,x\ne0$ where $A\in\mathbb{R}^{n\times(m+1)}$ is randomly chosen according to the standard normal distribution. We exactly determine the distribution of ${\mathscr{C}}(A)$ conditioned to $A$ being feasible and provide an upper bound on the distribution function in the infeasible case. Using these results, we show that $\mathbf{E}(\ln{\mathscr{C}}(A))\le2\ln(m+1)+3.31$ for all $n>m$, the sharpest bound for this expectancy as of today. Both agendas are related through a result which translates between coverage and condition.