{"id":"arxiv/2601.12398","url":"https://arxiv.org/abs/2601.12398","tags":["math.OC"],"score":0,"title":"Anderson Acceleration for Distributed Constrained Optimization over Time-varying Networks","domain":"math.OC","source":"arxiv","authors":["Haijuan Liu","Xuyang Wu"],"reviews":0,"subject":"math.OC","summary":"This paper applies the Anderson Acceleration (AA) technique to accelerate the Fenchel dual gradient method (FDGM) to solve constrained optimization problems over time-varying networks. AA is originally designed for accelerating fixed-point iterations, and its direct application to FDGM faces two challenges: 1) FDGM in time-varying networks cannot be formulated as a standard fixed-point update; 2) even if the network is fixed so that FDGM can be expressed as a fixed-point iteration, the direct application of AA is not distributively implementable. To overcome these challenges, we first…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-01-18","reactions":{}}