1: \begin{abstract}

2: 	This paper addresses the gradient flow--the continuous-time representation of the gradient method--with the smooth approximation of a non-differentiable objective function

3: 	and presents convergence analysis framework.

4: 	Similar to the gradient method, the gradient flow is inapplicable to the non-differentiable function minimization;

5: 	therefore, this paper addresses the smoothing gradient method,

6: 	which exploits a decreasing smoothing parameter sequence in the smooth approximation.

7: 	The convergence analysis is presented using conventional Lyapunov-function-based techniques,

8: 	and a Lyapunov function applicable to both strongly convex and non-strongly convex objective functions is provided

9: 	by taking into consideration the effect of the smooth approximation.

10: 	Based on the equivalence of the stepsize in the smoothing gradient method and

11: 	the discretization step in the forward Euler scheme for

12: 	the numerical integration of the smoothing gradient flow,

13: 	the sample values of the exact solution of the smoothing gradient flow are compared with

14: 	the state variable of the smoothing gradient method, and

15: 	the equivalence of the convergence rates is shown.

16: \end{abstract}

17: