Full Text: PDF
Volume 6, Issue 2, 1 April 2022, Pages 5-14
Abstract. In this paper, we study a Bregman proximal point algorithm (BPPA) for convex optimization problems. Though the convergence and sublinear convergence rate for BPPA are well-understand, the linear convergence rate for BPPA has yet been thoroughly studied in the literature. In this paper, we analyze the linear convergence rate of BPPA. Under the assumption that the objective function is strongly convex relative to a Legendre function, we establish the linear convergence for the function values sequence. Moreover, if the Legendre function is strongly convex and smooth, the linear convergence for the iterative sequence of BPPA is obtained.
How to Cite this Article:
K. Guo, C. Zhu, On the linear convergence of a Bregman proximal point algorithm, J. Nonlinear Var. Anal. 6 (2022), 5-14.