{"id":"arxiv/2610.01700","url":"https://arxiv.org/abs/2610.01700","tags":["math.NA"],"score":0,"title":"Deterministic and stochastic particle methods for the Fokker-Planck equation in S--formulation with application to point set registration","domain":"math.NA","source":"arxiv","authors":["Klaas Willems","Angelo Iollo","Giovanni Russo","Tommaso Taddei"],"reviews":0,"subject":"math.NA","summary":"We present two particle methods for point set registration in bounded domains based on the Fokker-Planck equation. The first method relies on a moving least squares discretization of the S--formulation, in which moving grid points (particles) are advected by the drift with the target distribution, diffusion is resolved on a dynamically evolving particle cloud. This setting naturally leads to strong compression and expansion of the particle cloud. Obstacles are handled by enforcing reflective boundary conditions through a novel ghost point method. The second method is a Monte Carlo solver for…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-01","reactions":{}}