TROT: Tsallis-Regularized Optimal Transport Unifies Wasserstein and KL Divergences

FrnkNlsn · x · 2026-09-16

A 2016 paper by Muzellec, Nock, Patrini, and Nielsen proposes Tsallis-regularized optimal transport (TROT): regularizing mass transport with a Tsallis entropy term instead of Shannon entropy, it unifies the Monge-Kantorovitch and Sinkhorn-Cuturi approaches and interpolates a family of divergences from Wasserstein to KL, Pearson, Neyman, and Hellinger, with efficient Sinkhorn-like algorithms and convergence proofs.

It also applies optimal transport to ecological inference — reconstructing joint distributions from marginals — demonstrated on 2012 US presidential election data.

Original post →

More from Research

Research channel →