{"id":"arxiv/2606.01787","url":"https://arxiv.org/abs/2606.01787","tags":["cs.AI","math.OC"],"score":0,"title":"Stochastic convergence of parallel asynchronous adaptive first-order methods","domain":"cs.AI","source":"arxiv","authors":["Serge Gratton","Philippe L. Toint"],"reviews":0,"subject":"cs.AI","summary":"A new class of asynchronous adaptive first-order optimization methods is introduced, comprising asynchronous variants of several popular algorithms. Versions of these methods using momentum and/or inexact normalization are also considered. The convergence of methods in the class on non-convex functions is analyzed in a fully stochastic setting, and is shown to be (up to logarithmic factors) of order O(1/sqrt{t}) under reasonable assumptions. Numerical experiments suggest that such asynchronous adaptive algorithms are very relevant in heterogeneous large-scale machine learning systems.","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-06-01","reactions":{}}