{"id":"arxiv/2604.12751","url":"https://arxiv.org/abs/2604.12751","tags":["math.OC","eess.SY"],"score":0,"title":"Finite-Time Optimization via Scaled Gradient-Momentum Flows","domain":"math.OC","source":"arxiv","authors":["Yu Zhou","Mengmou Li","Masaaki Nagahara"],"reviews":0,"subject":"math.OC","summary":"In this paper, we develop a scaled gradient-momentum framework for continuous-time optimization that achieves global finite-time convergence. A state-dependent scaling mechanism is introduced to enable classical dynamics, such as Heavy-Ball-type and proportional-integral (PI)-type flows, to attain finite-time convergence. We establish explicit conditions that bridge the gradient-dominance property of the objective function and finite-time stability of the proposed scaled dynamics. Numerical experiments validate the theoretical results.","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-04-14","reactions":{}}