Proximal point type algorithms with relaxed and inertial effects beyond convexity

被引：1
|
作者：
Grad, S. -M. [1 ]
Lara, F. [2 ]
Marcavillaca, R. T. [2 ]
机构：
[1] Inst Polytech Paris, Unite Math Appl, ENSTA Paris, Palaiseau, France
[2] Univ Tarapaca, Inst Alta Invest IAI, Arica, Chile
来源：
OPTIMIZATION | 2024年 / 73卷 / 11期
关键词：
Proximal point algorithms; relaxed iterative methods; inertial iterative methods; generalized convexity; prox-convexity; MAXIMAL MONOTONE-OPERATORS; WEAK-CONVERGENCE;
D O I：
10.1080/02331934.2024.2329779
中图分类号：
C93 [管理学]; O22 [运筹学];
学科分类号：
070105 ; 12 ; 1201 ; 1202 ; 120202 ;
摘要：
We show that the recent relaxed-inertial proximal point algorithm due to Attouch and Cabot remains convergent when the function to be minimized is not convex, being only endowed with certain generalized convexity properties. Numerical experiments showcase the improvements brought by the relaxation and inertia features to the standard proximal point method in this setting, too.
引用
收藏
页码：3393 / 3410
页数：18
相关论文
共 50 条
12345