Faster Logconcave Sampling from a Cold Start in High Dimension

Yunbum Kook, Santosh S. Vempala · 2025

Read Faster Logconcave Sampling from a Cold Start in High Dimension by Yunbum Kook, Santosh S. Vempala (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-01937.txt.

Introduction

Sampling from a complex probability shape in many dimensions is a core task in statistics and machine learning. Yunbum Kook and Santosh Vempala speed up the cold start of that task in a 2025 theory paper.

Most known samplers need a warm start, a rough sample that is already close to the target, and earning one used to cost at least a cubic penalty in dimension. The new method relaxes how close the warm start must be, using a weaker distance measure. It also sharpens a classic inequality that controls mixing speed, tying it to the shape's spread. Together, these steps give the first samplers that beat the cubic barrier for near-round inputs.

Catalog record

CREATORYunbum Kook, Santosh S. Vempala
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE167 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-01937

Browse the full library · About the Tower of Babel