{"id":"arxiv/2604.26932","url":"https://arxiv.org/abs/2604.26932","tags":["math.OC","cs.LG"],"score":0,"title":"Learning Over-Relaxation Policies for ADMM with Convergence Guarantees","domain":"math.OC","source":"arxiv","authors":["Junan Lin","Paul J. Goulart","Luca Furieri"],"reviews":0,"subject":"math.OC","summary":"The Alternating Direction Method of Multipliers (ADMM) is a widely used method for structured convex optimization, and its practical performance depends strongly on the choice of penalty and relaxation parameters. Motivated by settings such as Model Predictive Control (MPC), where one repeatedly solves related optimization problems with fixed structure and changing parameter values, we propose learning online updates of the relaxation parameter to improve average performance on problem classes of interest, while guaranteeing that asymptotic convergence is not compromised for the worst-case…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-04-29","reactions":{}}