{"id":"arxiv/2610.32169","url":"https://arxiv.org/abs/2610.32169","tags":["math.NA"],"score":0,"title":"Weak adversarial tensor neural network for high-dimensional Fokker-Planck equations","domain":"math.NA","source":"arxiv","authors":["Sihong Shao","Zhongwei Xu"],"reviews":0,"subject":"math.NA","summary":"Solving high-dimensional stationary Fokker-Planck (SFP) equations remains challenging. Recently, deep learning methods have shown great potential for solving high-dimensional SFP equations due to their universal approximation and mesh-free training. However, the high-dimensional integrals involved in loss computation and normalization can be expensive to approximate accurately using Monte Carlo integration. To address this issue, we propose a weak adversarial tensor neural network (WATNN) method for solving high-dimensional SFP equations, which combines a weak adversarial (WA) formulation…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-23","reactions":{}}