English

Pure entropic regularization for metrical task systems

Data Structures and Algorithms 2020-09-04 v3 Metric Geometry

Abstract

We show that on every nn-point HST metric, there is a randomized online algorithm for metrical task systems (MTS) that is 11-competitive for service costs and O(logn)O(\log n)-competitive for movement costs. In general, these refined guarantees are optimal up to the implicit constant. While an O(logn)O(\log n)-competitive algorithm for MTS on HST metrics was developed by Bubeck et al. (SODA 2019), that approach could only establish an O((logn)2)O((\log n)^2)-competitive ratio when the service costs are required to be O(1)O(1)-competitive. Our algorithm can be viewed as an instantiation of online mirror descent with the regularizer derived from a multiscale conditional entropy. In fact, our algorithm satisfies a set of even more refined guarantees; we are able to exploit this property to combine it with known random embedding theorems and obtain, for any nn-point metric space, a randomized algorithm that is 11-competitive for service costs and O((logn)2)O((\log n)^2)-competitive for movement costs.

Keywords

Cite

@article{arxiv.1906.04270,
  title  = {Pure entropic regularization for metrical task systems},
  author = {Christian Coester and James R. Lee},
  journal= {arXiv preprint arXiv:1906.04270},
  year   = {2020}
}

Comments

COLT 2019

R2 v1 2026-06-23T09:49:30.031Z