English

Towards Optimal Running Times for Optimal Transport

Data Structures and Algorithms 2020-01-29 v3

Abstract

In this work, we provide faster algorithms for approximating the optimal transport distance, e.g. earth mover's distance, between two discrete probability distributions μ,νΔn\mu, \nu \in \Delta^n. Given a cost function C:[n]×[n]R0C : [n] \times [n] \to \mathbb{R}_{\geq 0} where C(i,j)1C(i,j) \leq 1 quantifies the penalty of transporting a unit of mass from ii to jj, we show how to compute a coupling XX between rr and cc in time O~(n2/ϵ)\widetilde{O}\left(n^2 /\epsilon \right) whose expected transportation cost is within an additive ϵ\epsilon of optimal. This improves upon the previously best known running time for this problem of O~(min{n9/4/ϵ,n2/ϵ2})\widetilde{O}\left(\text{min}\left\{ n^{9/4}/\epsilon, n^2/\epsilon^2 \right\}\right). We achieve our results by providing reductions from optimal transport to canonical optimization problems for which recent algorithmic efforts have provided nearly-linear time algorithms. Leveraging nearly linear time algorithms for solving packing linear programs and for solving the matrix balancing problem, we obtain two separate proofs of our stated running time. Further, one of our algorithms is easily parallelized and can be implemented with depth O~(1/ϵ)\widetilde{O}(1/\epsilon). Moreover, we show that further algorithmic improvements to our result would be surprising in the sense that any improvement would yield an o(n2.5)o(n^{2.5}) algorithm for \textit{maximum cardinality bipartite matching}, for which currently the only known algorithms for achieving such a result are based on fast-matrix multiplication.

Keywords

Cite

@article{arxiv.1810.07717,
  title  = {Towards Optimal Running Times for Optimal Transport},
  author = {Jose Blanchet and Arun Jambulapati and Carson Kent and Aaron Sidford},
  journal= {arXiv preprint arXiv:1810.07717},
  year   = {2020}
}
R2 v1 2026-06-23T04:43:39.826Z