English

Online metric allocation

Data Structures and Algorithms 2021-12-01 v1

Abstract

We introduce a natural online allocation problem that connects several of the most fundamental problems in online optimization. Let MM be an nn-point metric space. Consider a resource that can be allocated in arbitrary fractions to the points of MM. At each time tt, a convex monotone cost function ct:[0,1]R+c_t: [0,1]\to\mathbb{R}_+ appears at some point rtMr_t\in M. In response, an algorithm may change the allocation of the resource, paying movement cost as determined by the metric and service cost ct(xrt)c_t(x_{r_t}), where xrtx_{r_t} is the fraction of the resource at rtr_t at the end of time tt. For example, when the cost functions are ct(x)=αxc_t(x)=\alpha x, this is equivalent to randomized MTS, and when the cost functions are ct(x)=1x<1/kc_t(x)=\infty\cdot 1_{x<1/k}, this is equivalent to fractional kk-server. We give an O(logn)O(\log n)-competitive algorithm for weighted star metrics. Due to the generality of allowed cost functions, classical multiplicative update algorithms do not work for the metric allocation problem. A key idea of our algorithm is to decouple the rate at which a variable is updated from its value, resulting in interesting new dynamics. This can be viewed as running mirror descent with a time-varying regularizer, and we use this perspective to further refine the guarantees of our algorithm. The standard analysis techniques run into multiple complications when the regularizer is time-varying, and we show how to overcome these issues by making various modifications to the default potential function. We also consider the problem when cost functions are allowed to be non-convex. In this case, we give tight bounds of Θ(n)\Theta(n) on tree metrics, which imply deterministic and randomized competitive ratios of O(n2)O(n^2) and O(nlogn)O(n\log n) respectively on arbitrary metrics. Our algorithm is based on an 22\ell_2^2-regularizer.

Keywords

Cite

@article{arxiv.2111.15169,
  title  = {Online metric allocation},
  author = {Nikhil Bansal and Christian Coester},
  journal= {arXiv preprint arXiv:2111.15169},
  year   = {2021}
}
R2 v1 2026-06-24T07:57:11.889Z