Sampling from convex sets with a cold start using multiscale decompositions
Abstract
Running a random walk in a convex body is a standard approach to sample approximately uniformly from the body. The requirement is that from a suitable initial distribution, the distribution of the walk comes close to the uniform distribution on after a number of steps polynomial in and the aspect ratio (i.e., when ). Proofs of rapid mixing of such walks often require the probability density of the initial distribution with respect to to be at most : this is called a "warm start". Achieving a warm start often requires non-trivial pre-processing before starting the random walk. This motivates proving rapid mixing from a "cold start", wherein can be as high as . Unlike warm starts, a cold start is usually trivial to achieve. However, a random walk need not mix rapidly from a cold start: an example being the well-known "ball walk". On the other hand, Lov\'asz and Vempala proved that the "hit-and-run" random walk mixes rapidly from a cold start. For the related coordinate hit-and-run (CHR) walk, which has been found to be promising in computational experiments, rapid mixing from a warm start was proved only recently but the question of rapid mixing from a cold start remained open. We construct a family of random walks inspired by classical decompositions of subsets of into countably many axis-aligned dyadic cubes. We show that even with a cold start, the mixing times of these walks are bounded by a polynomial in and the aspect ratio. Our main technical ingredient is an isoperimetric inequality for for a metric that magnifies distances between points close to the boundary of . As a corollary, we show that the CHR walk also mixes rapidly both from a cold start and from a point not too close to the boundary of .
Cite
@article{arxiv.2211.04439,
title = {Sampling from convex sets with a cold start using multiscale decompositions},
author = {Hariharan Narayanan and Amit Rajaraman and Piyush Srivastava},
journal= {arXiv preprint arXiv:2211.04439},
year = {2024}
}
Comments
Changes from v3: Added further discussion/details, and fixed some typos. This version should be close to the final version