.png){kind=icon}
.png){kind=icon}
Return
ACCELERATING CONTINUOUS-TIME OPTIMIZATION WITH FIXED-TIME TIME-VARYING SECOND-ORDER DYNAMICS
.png){kind=icon}
PRE
AI
DOI：10.23952/jnva.10.2026.3.02.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper proposes a new fixed-time second-order time-varying dynamics to accelerate continuous-time optimization. We derive a sufficient condition for the existence and uniqueness of solutions to a general ordinary differential equation (ODE) and present a rigorous proof establishing the well-posedness of our proposed algorithm, addressing a significant gap in the literature. We show that the time-varying coefficients in the proposed dynamics can be used to accelerate convergence speed to optimal solutions. The efficiency of the proposed method is demonstrated by numerical simulations on the ridge regression and training neural networks.
Keywords:
. Fixed-time convergence
Gradient flow
Lyapunov stability
Second-order dynamics
Time-varying systems
Journal
IF:
1.9
Papers:
72
Citations:
356