{"id":"arxiv/2610.04350","url":"https://arxiv.org/abs/2610.04350","tags":["math.NA","math.DS","math.SG"],"score":0,"title":"Inference of Hamiltonian and Lie-Poisson structures from vector field samples","domain":"math.NA","source":"arxiv","authors":["Jason E. Frank","Georg A. Gottwald"],"reviews":0,"subject":"math.NA","summary":"We present a method for the data-driven inference of Hamiltonian and Lie-Poisson dynamical systems from vector-field samples, in which the Hamiltonian and the underlying Poisson tensor are estimated jointly rather than assuming the structure is known. The Hamiltonian is represented as a linear combination of random feature maps with fixed random internal weights, so its estimation is a linear regression problem for any given Poisson structure. We treat three cases of increasing generality: a known Poisson tensor, an unknown constant invertible (symplectic) structure matrix, and Lie-Poisson…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-03","reactions":{}}