Peeling potatoes near-optimally in near-linear time

TL;DR

This paper presents a randomized near-linear-time algorithm that approximates the maximum-area convex polygon inside a simple polygon with high probability, advancing geometric optimization techniques.

Contribution

It introduces a near-linear-time approximation algorithm for maximum convex polygon inside a simple polygon, utilizing new geometric probability bounds.

Findings

Achieves a (1-ε)-approximation in near-linear time

Provides probabilistic bounds relating visibility and convex body area

Develops bounds on expected perimeter differences in random sampling

Abstract

We consider the following geometric optimization problem: find a convex polygon of maximum area contained in a given simple polygon $P$ with $n$ vertices. We give a randomized near-linear-time $(1-\varepsilon)$-approximation algorithm for this problem: in $O(n( \log^2 n + (1/\varepsilon^3) \log n + 1/\varepsilon^4))$ time we find a convex polygon contained in $P$ that, with probability at least $2/3$, has area at least $(1-\varepsilon)$ times the area of an optimal solution. We also obtain similar results for the variant of computing a convex polygon inside $P$ with maximum perimeter. To achieve these results we provide new results in geometric probability. The first result is a bound relating the probability that two points chosen uniformly at random inside $P$ are mutually visible and the area of the largest convex body inside $P$. The second result is a bound on the expected value of the difference between the perimeter of any planar convex body $K$ and the perimeter of the convex hull of a uniform random sample inside $K$.