{"id":"arxiv/2603.29348","url":"https://arxiv.org/abs/2603.29348","tags":["math.OC"],"score":0,"title":"Stochastic Block Bregman Projection with Polyak-like Stepsize for Possibly Inconsistent Convex Feasibility Problems","domain":"math.OC","source":"arxiv","authors":["Lu Zhang","Hongzhen Chen","Hongxia Wang","Hui Zhang"],"reviews":0,"subject":"math.OC","summary":"Stochastic projection algorithms for solving convex feasibility problems (CFPs) have attracted considerable attention due to their broad applicability. In this paper, we propose a unified stochastic bilevel reformulation for possibly inconsistent CFPs that combines proximity function minimization and structural regularization, leading to a feasible bilevel model with a unique and stable regularized solution. From the algorithmic perspective, we develop the stochastic block Bregman projection method with Polyak-like and projective stepsizes, which not only subsumes several recent stochastic…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-03-31","reactions":{}}