Saved in:
Bibliographic Details
Main Authors: Liu, Chengchang, Chen, Cheng, Luo, Luo
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2303.16188
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909265790238720
author Liu, Chengchang
Chen, Cheng
Luo, Luo
author_facet Liu, Chengchang
Chen, Cheng
Luo, Luo
contents This paper proposes a novel class of block quasi-Newton methods for convex optimization which we call symmetric rank-$k$ (SR-$k$) methods. Each iteration of SR-$k$ incorporates the curvature information with~$k$ Hessian-vector products achieved from the greedy or random strategy. We prove that SR-$k$ methods have the local superlinear convergence rate of $\mathcal{O}\big((1-k/d)^{t(t-1)/2}\big)$ for minimizing smooth and strongly convex function, where $d$ is the problem dimension and $t$ is the iteration counter. This is the first explicit superlinear convergence rate for block quasi-Newton methods, and it successfully explains why block quasi-Newton methods converge faster than ordinary quasi-Newton methods in practice. We also leverage the idea of SR-$k$ methods to study the block BFGS and block DFP methods, showing their superior convergence rates.
format Preprint
id arxiv_https___arxiv_org_abs_2303_16188
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Symmetric Rank-$k$ Methods
Liu, Chengchang
Chen, Cheng
Luo, Luo
Optimization and Control
This paper proposes a novel class of block quasi-Newton methods for convex optimization which we call symmetric rank-$k$ (SR-$k$) methods. Each iteration of SR-$k$ incorporates the curvature information with~$k$ Hessian-vector products achieved from the greedy or random strategy. We prove that SR-$k$ methods have the local superlinear convergence rate of $\mathcal{O}\big((1-k/d)^{t(t-1)/2}\big)$ for minimizing smooth and strongly convex function, where $d$ is the problem dimension and $t$ is the iteration counter. This is the first explicit superlinear convergence rate for block quasi-Newton methods, and it successfully explains why block quasi-Newton methods converge faster than ordinary quasi-Newton methods in practice. We also leverage the idea of SR-$k$ methods to study the block BFGS and block DFP methods, showing their superior convergence rates.
title Symmetric Rank-$k$ Methods
topic Optimization and Control
url https://arxiv.org/abs/2303.16188