{"id":"arxiv/2602.23854","url":"https://arxiv.org/abs/2602.23854","tags":["math.OC","cs.LG","math.NA","stat.ML"],"score":0,"title":"A distributed semismooth Newton based augmented Lagrangian method for distributed optimization","domain":"math.OC","source":"arxiv","authors":["Qihao Ma","Chengjing Wang","Peipei Tang","Dunbiao Niu","Aimin Xu"],"reviews":0,"subject":"math.OC","summary":"This paper proposes a novel distributed semismooth Newton based augmented Lagrangian method for solving a class of optimization problems over networks, where the global objective is defined as the sum of locally held cost functions, and communication is restricted to neighboring agents. Specifically, we employ the augmented Lagrangian method to solve an equivalently reformulated constrained version of the original problem. Each resulting subproblem is solved inexactly via a distributed semismooth Newton method. By fully leveraging the structure of the generalized Hessian, a distributed…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-02-27","reactions":{}}