{"id":"arxiv/2605.28894","url":"https://arxiv.org/abs/2605.28894","tags":["math.OC","stat.ML"],"score":0,"title":"Saddle Networks: Structure-Preserving Architectures for Convex-Concave Functions","domain":"math.OC","source":"arxiv","authors":["Xavier Warin"],"reviews":0,"subject":"math.OC","summary":"Saddle-point models arise throughout optimization, optimal transport, robust learning, and control. In many applications, the relevant function f(x,y) is convex in x and concave in y, and preserving this geometry is essential for obtaining tractable min--max formulations and reliable certificates. We introduce a structured separable decomposition that preserves the convex-concave geometry and prove a complete one-dimensional approximation theorem under a mixed Monge-type convexity condition. We then describe practical saddle network architectures that preserve convexity in x and concavity in…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-05-27","reactions":{}}