1: \begin{abstract}

2: This paper generalizes 

3: the optimized gradient method (OGM)~\cite{drori:14:pof,kim:16:ofo,kim:17:otc}

4: that achieves

5: the optimal worst-case cost function bound of first-order methods

6: for smooth convex minimization~\cite{drori:17:tei}.

7: Specifically,

8: this paper studies a generalized formulation of OGM

9: and analyzes its worst-case rates

10: in terms of both the function value and the norm of the function gradient.

11: This paper also develops a new algorithm called OGM-OG

12: that is in the generalized family of OGM and

13: that has the best known analytical worst-case bound

14: with rate $O(1/N^{1.5})$

15: on the decrease of the gradient norm

16: among fixed-step first-order methods.

17: This paper also proves

18: that Nesterov's fast gradient method~\cite{nesterov:83:amf,nesterov:05:smo}

19: has an $O(1/N^{1.5})$ worst-case gradient norm rate

20: but with constant larger than OGM-OG.

21: The proof is based on the worst-case analysis 

22: called Performance Estimation Problem in~\cite{drori:14:pof}.

23: %

24: %Include keywords, PACS and mathematical

25: %subject classification numbers as needed.

26: %\keywords{First-order algorithms \and Gradient methods

27: %\and Convergence rate 

28: %\and Smooth convex minimization \and Worst-case performance analysis}

29: % \PACS{PACS code1 \and PACS code2 \and more}

30: % \subclass{MSC code1 \and MSC code2 \and more}

31: \end{abstract}