New convergence results for the inexact variable metric forward–backward method
作者：
Highlights：
• The minimization of nonconvex Kurdyka–Lojasiewicz functions is addressed.
• Convergence results on forward-backward methods are provided.
• Inexact computation of the proximal gradient point is allowed.
• Implementable rules to fulfill the theoretical requirements are discussed.
• Numerical tests of image deconvolution problems are shown.
摘要
•The minimization of nonconvex Kurdyka–Lojasiewicz functions is addressed.•Convergence results on forward-backward methods are provided.•Inexact computation of the proximal gradient point is allowed.•Implementable rules to fulfill the theoretical requirements are discussed.•Numerical tests of image deconvolution problems are shown.
论文关键词：Numerical optimization,Inexact forward–backward methods,Nonconvex problems,Kurdyka–Łojasiewicz property
论文评审过程：Received 26 February 2020, Revised 13 August 2020, Accepted 28 September 2020, Available online 21 October 2020, Version of Record 21 October 2020.
论文官网地址：https://doi.org/10.1016/j.amc.2020.125719
