{"id":"arxiv/2607.09461","url":"https://arxiv.org/abs/2607.09461","tags":["q-fin.CP","econ.EM","math.OC","stat.ML"],"score":0,"title":"Deep Learning for Dynamic Programming with Recursive Utility Using First-order Conditions","domain":"q-fin.CP","source":"arxiv","authors":["Xianhua Peng","Wu Guo","Songyan Wang","Jianfei Zhu"],"reviews":0,"subject":"q-fin.CP","summary":"This paper proposes the certainty-equivalent first-order learning (CEFOL) algorithm, a deep learning algorithm for solving discrete-time dynamic programming problems with recursive utility. Dynamic programming with recursive utility is challenging because nonlinear certainty equivalent appears in the Bellman equation and the first-order optimality conditions but is difficult to evaluate. By introducing a separate neural network to represent the certainty equivalent, CEFOL enables the exploitation of the Bellman and model-specific first-order optimality conditions. In addition to certainty…","category":"research","comments":0,"disputed":false,"division":"econ","mentions":0,"discussed":0,"published":"2026-07-10","reactions":{}}