Sampling from convex sets with a cold start using multiscale decompositions
Hariharan Narayanan, Amit Rajaraman, Piyush Srivastava
TL;DR
The paper tackles efficient sampling from a convex body $K$ by proving rapid mixing of multiscale Whitney-based Markov chains $\\mathcal{M}_p$ from cold starts, with a central technical development being a $g_p$-metric isoperimetric inequality that magnifies near-boundary geometry. By relating isoperimetric bounds to conductance, the authors obtain polynomial mixing times in the dimension $n$ and aspect ratio $R/r$, namely $T = O(n^{4+2/p}(R/r)^2\log(M/\\epsilon))$ for $\\mathcal{M}_p$ and $T = O(n^9(R/r)^2\log(M/\\epsilon))$ for coordinate hit-and-run (CHR) from cold starts, where $M$ reflects warm-start softness. A key byproduct is a new $\\ell_0$-isoperimetric inequality for axis-disjoint sets that removes the need for warm starts in CHR, coupling multiscale conductance with axis-disjoint geometry. The work also provides algorithmic implementations for $p=1$ and distance-oracle requirements for $p>1$, and discusses extensions to point-starts and close-to-boundary scenarios, offering a substantial step toward practical, provably efficient sampling from high-dimensional convex bodies. Overall, the results unify multiscale decompositions with non-Euclidean isoperimetry to achieve cold-start rapid mixing for fundamental sampling walks in convex geometry.
Abstract
Running a random walk in a convex body $K\subseteq\mathbb{R}^n$ 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 $π_K$ on $K$ after a number of steps polynomial in $n$ and the aspect ratio $R/r$ (i.e., when $rB_2 \subseteq K \subseteq RB_{2}$). Proofs of rapid mixing of such walks often require the probability density $η_0$ of the initial distribution with respect to $π_K$ to be at most $\mathrm{poly}(n)$: 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 $η_0$ can be as high as $\exp(\mathrm{poly}(n))$. 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, Lovász 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 $\mathbb{R}^n$ 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 $n$ and the aspect ratio. Our main technical ingredient is an isoperimetric inequality for $K$ for a metric that magnifies distances between points close to the boundary of $K$. 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 $K$.