{"id":"arxiv/2604.07404","url":"https://arxiv.org/abs/2604.07404","tags":["cond-mat.stat-mech","cs.LG","math.AP","stat.ML"],"score":0,"title":"Score Shocks: The Burgers Equation Structure of Diffusion Generative Models","domain":"cond-mat.stat-mech","source":"arxiv","authors":["Krisanu Sarkar"],"reviews":0,"subject":"cond-mat.stat-mech","summary":"We analyze the score field of a diffusion generative model through a Burgers-type evolution law. For VE diffusion, the heat-evolved data density implies that the score obeys viscous Burgers in one dimension and the corresponding irrotational vector Burgers system in $\\R^d$, giving a PDE view of \\emph{speciation transitions} as the sharpening of inter-mode interfaces. For any binary decomposition of the noised density into two positive heat solutions, the score separates into a smooth background and a universal $\\tanh$ interfacial term determined by the component log-ratio; near a regular…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-04-08","reactions":{}}