1: \begin{abstract}

2: 	This paper presents an Euler--Lagrange system

3: 	for a continuous-time model of the accelerated gradient methods in smooth convex optimization

4: 	and proposes an associated Lyapunov-function-based convergence analysis framework.

5: 	Recently, ordinary differential equations (ODEs) with dumping terms have been developed to

6: 	intuitively interpret the accelerated gradient methods, and

7: 	the design of unified model describing the various individual ODE models have been examined.

8: 	In existing reports, the Lagrangian, which results in the Euler-Lagrange equation, and the Lyapunov function for the convergence analysis

9: 	have been separately proposed for each ODE.

10: 	This paper proposes a unified Euler--Lagrange system and its Lyapunov function to cover the existing various models.

11: 	In the convergence analysis using the Lyapunov function, a condition that parameters in the Lagrangian and Lyapunov function must satisfy is derived,

12: 	and a parameter design for improving the convergence rate naturally results in the mysterious dumping coefficients.

13: 	Especially, a symmetric Bregman divergence can lead to a relaxed condition of the parameters and

14: 	a resulting improved convergence rate.

15: 	As an application of this study, a slight modification in the Lyapunov function establishes the similar convergence proof

16: 	for ODEs with smooth approximation in nondifferentiable objective function minimization.

17: \end{abstract}

18: