{"id":"arxiv/2610.19027","url":"https://arxiv.org/abs/2610.19027","tags":["math.OC","cs.LG"],"score":0,"title":"Optimization for One, Feasibility for Most: Decoupling Optimization from Feasibility in Chance-Constrained Programs","domain":"math.OC","source":"arxiv","authors":["Justin Dumouchelle","Shunyu Yao","Beste Basciftci","Thiago Serra"],"reviews":0,"subject":"math.OC","summary":"Chance-constrained stochastic programs (CCSPs) require solutions that satisfy uncertain constraints with high probability. Standard approaches either embed scenarios directly into optimization by considering sample average approximation (SAA)-based reformulations, growing prohibitively expensive and often failing to find feasible solutions; or employ conservative reformulations that guarantee feasibility but sacrifice objective quality by design. We propose ACCESS (Adaptive Chance-Constrained Exploration via Structured Sampling), a structured sampling framework that decouples optimization…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-13","reactions":{}}