{"id":"arxiv/2611.08505","url":"https://arxiv.org/abs/2611.08505","tags":["stat.ML","cs.LG","math.OC"],"score":0,"title":"Drift-admissibility via Optimal Transport","domain":"stat.ML","source":"arxiv","authors":["Abigail Langbridge","Enrico Biffis","Jakub Mareček","Robert Shorten"],"reviews":0,"subject":"stat.ML","summary":"We introduce structure-aware optimal transport diagnostics for assessing whether empirical time-series dynamics are compatible with a prescribed predictable-drift class. Our approach translates a pathwise property of stochastic processes into a finite-dimensional projection problem over empirical transition laws, encoding admissibility as row-wise moment constraints on local couplings and quantifying deviations via a Kullback--Leibler projection gap $G_{λ,k}$. This yields both a principled diagnostic and a scalable Bregman--Dykstra algorithm that retains the efficiency of Sinkhorn-type…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-09-10","reactions":{}}