{"id":"arxiv/2609.36282","url":"https://arxiv.org/abs/2609.36282","tags":["math.OC"],"score":0,"title":"A projected descent method for minimization of weakly semismooth functions over polyhedral sets","domain":"math.OC","source":"arxiv","authors":["Morteza Maleknia","Majid Soleimani-damaneh"],"reviews":0,"subject":"math.OC","summary":"This study develops a projection-based descent method for minimizing weakly semismooth functions over polyhedral sets, taking an initial step toward extending projection-based descent algorithms to nonsmooth, nonconvex optimization problems. At each iteration, the method considers an inner approximation of the Clarke $\\varepsilon$-subdifferential at the current iterate over the feasible region and uses its least-norm element to generate a search direction. An exponential limited backtracking line search is then developed to assess the quality of the generated direction, while a projected…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-28","reactions":{}}