{"id":"arxiv/2604.06675","url":"https://arxiv.org/abs/2604.06675","tags":["math.OC"],"score":0,"title":"A Sample-Wise Adjoint Regression Framework for Mean-Field Control with Connections to Adjoint Matching","domain":"math.OC","source":"arxiv","authors":["Hui Sun"],"reviews":0,"subject":"math.OC","summary":"This work proposes a novel numerical approach for solving mean-field control (MFC) problems using an adjoint-based optimization framework motivated by the stochastic maximum principle (SMP). Rather than solving the adjoint processes $(Y_t,Z_t)$ explicitly, we construct sample-wise unbiased estimators of their discretized counterparts and use them to approximate the Hamiltonian control gradient through recursive regression. The control is then updated by a gradient-descent scheme. This approach differs from conventional deep-learning methods, which typically follow a…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-04-08","reactions":{}}