{"id":"arxiv/2605.23871","url":"https://arxiv.org/abs/2605.23871","tags":["stat.ML","cs.LG","math.ST"],"score":0,"title":"Move on Muon : A Hamiltonian probability gradient flow perspective of Muon optimizer","domain":"stat.ML","source":"arxiv","authors":["Aratrika Mustafi","Soumya Mukherjee","Bharath K. Sriperumbudur"],"reviews":0,"subject":"stat.ML","summary":"We develop a gradient flow on the space of probability measures defined on matrix-valued parameters induced by regularized Muon, an analytically smoothed version of the idealized Muon optimizer. The key observation is that the regularized orthogonalization map is the gradient of a smooth Fenchel-dual smoothing of the nuclear norm. This identifies the (regularized) Muon update as a mirror/prox step in the update variable, with momentum acting as the dual coordinate. We use this structure to lift Muon from a single matrix parameter to finite-particle probability objectives of the form…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-05-22","reactions":{}}