English

A Superlinear Convergence Framework for Kurdyka-{\L}ojasiewicz Optimization

Optimization and Control 2023-11-14 v3

Abstract

This work extends the iterative framework proposed by Attouch et al. (in Math. Program. 137: 91-129, 2013) for minimizing a nonconvex and nonsmooth function Φ\Phi so that the generated sequence possesses a Q-superlinear convergence rate. This framework consists of a monotone decrease condition, a relative error condition and a continuity condition, and the first two conditions both involve a parameter p ⁣>0p\!>0. We justify that any sequence conforming to this framework is globally convergent when Φ\Phi is a Kurdyka-{\L}ojasiewicz (KL) function, and the convergence has a Q-superlinear rate of order pθ(1+p)\frac{p}{\theta(1+p)} when Φ\Phi is a KL function of exponent θ(0,pp+1)\theta\in(0,\frac{p}{p+1}). Then, we illustrate that the iterate sequence generated by an inexact q[2,3]q\in[2,3]-order regularization method for composite optimization problems with a nonconvex and nonsmooth term belongs to this framework, and consequently, first achieve the Q-superlinear convergence rate of order 4/34/3 for an inexact cubic regularization method to solve this class of composite problems with KL property of exponent 1/21/2.

Keywords

Cite

@article{arxiv.2210.12449,
  title  = {A Superlinear Convergence Framework for Kurdyka-{\L}ojasiewicz Optimization},
  author = {Yitian Qian and Shaohua Pan},
  journal= {arXiv preprint arXiv:2210.12449},
  year   = {2023}
}
R2 v1 2026-06-28T04:15:08.831Z