1: \begin{abstract}

2: This paper optimizes the step coefficients of first-order methods

3: for smooth convex minimization

4: in terms of the worst-case convergence bound (\ie, efficiency)

5: of the decrease of the gradient norm.

6: This work is based on the performance estimation problem approach. %~\cite{drori:14:pof}.

7: The worst-case gradient bound 

8: of the resulting method

9: is optimal up to a constant

10: for large-dimensional smooth convex minimization problems, %~\cite{nemirovsky:92:ibc};

11: under the initial bounded condition on the cost function value.

12: This paper then illustrates that

13: the proposed method %, named OGM-G, 

14: has a computationally efficient form

15: that is similar to the optimized gradient method. %(OGM). %~\cite{kim:16:ofo}.

16: \keywords{First-order methods

17: \and Gradient methods

18: \and Smooth convex minimization 

19: \and Worst-case performance analysis}

20: \subclass{90C25 \and 90C30 \and 90C60 \and 68Q25 \and 49M25 \and 90C22}

21: \end{abstract}

22: