{"id":"arxiv/2609.13677","url":"https://arxiv.org/abs/2609.13677","tags":["math.OC","cs.LG"],"score":0,"title":"Nonsmooth Optimization via Orthogonalized Momentum","domain":"math.OC","source":"arxiv","authors":["Lexiao Lai","Tianyi Lin","Jiayu Zhang"],"reviews":0,"subject":"math.OC","summary":"Modern real application problems involve matrix-valued parameters, yet conventional optimizers treat them as vectors, thereby motivating matrix-aware methods that exploit input-output geometry, such as Muon which orthogonalizes the momentum matrices before parameter updates. Its empirical success raises a conceptual question: can orthogonalized momentum remain effective beyond smooth optimization? This paper studies this question for locally Lipschitz functions using a generalized derivative framework compatible with backpropagation. Our first contribution is to identify a key limitation:…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-12","reactions":{}}