{"id":"arxiv/2604.22581","url":"https://arxiv.org/abs/2604.22581","tags":["math.OC"],"score":0,"title":"Stochastic Krasnoselskii-Mann Iterations: Convergence without Uniformly Bounded Variance","domain":"math.OC","source":"arxiv","authors":["Daniel Cortild","Coralia Cartis","Juan Peypouquet"],"reviews":0,"subject":"math.OC","summary":"We investigate the Stochastic Krasnoselskii-Mann iterations for expected nonexpansive fixed-point problems in a real separable Hilbert space. We establish convergence guarantees under significantly weaker assumptions on the variance than those typically used in the literature. In particular, instead of a uniform bound on the variance of the stochastic oracle, we only assume finite variance at a single fixed point. Under this assumption, we prove almost sure weak convergence of the iterates, derive convergence rates for the expected residual and the last-iterate residual, and obtain almost…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-04-24","reactions":{}}