{"id":"arxiv/2611.19882","url":"https://arxiv.org/abs/2611.19882","tags":["math.ST","math.OC","stat.CO","stat.ML"],"score":0,"title":"Certified Algorithms for Gaussian Location Mixture NPMLEs via Discretized Gradient Flows","domain":"math.ST","source":"arxiv","authors":["Pranav Gupta","Satyaki Mukherjee"],"reviews":0,"subject":"math.ST","summary":"We study the computation of approximate nonparametric maximum likelihood estimators (NPMLEs) for Gaussian location mixtures in arbitrary dimension $d\\ge 1$ via the Fisher-Rao gradient flow. We provide a certified algorithm based on an entropic proximal discretization of the flow which, given a sufficiently small $\\varepsilon > 0$, returns an absolutely continuous mixing measure $\\tildeμ$ admitting an exact finite-dimensional representation and satisfying $\\hat L_n-L_n(f_{\\tildeμ})\\le\\varepsilon$. For almost every dataset $X_1,\\cdots, X_n \\in \\mathbb{R}^d$ satisfying $n>2(d+1)(m+1)^2$, where…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-11-15","reactions":{}}