The Peaceman-Rachford splitting method is very efficient for minimizing sum of two functions each depends on its variable, and the constraint is a linear equality. However, its convergence was not guaranteed without extra requirements. Very recently, He et al. (SIAM J. Optim. 24: 1011 - 1040, 2014) proved the convergence of a strictly contractive Peaceman-Rachford splitting method by employing a suitable underdetermined relaxation factor. In this paper, we further extend the so-called strictly contractive Peaceman-Rachford splitting method by using two different relaxation factors, and to make the method more flexible, we introduce semi-proximal terms to the subproblems. We characterize the relation of these two factors, and show that one factor is always underdetermined while the other one is allowed to be larger than 1. Such a flexible conditions makes it possible to cover the Glowinski's ADMM whith larger stepsize. We show that the proposed modified strictly contractive Peaceman-Rachford splitting method is convergent and also prove $O(1/t)$ convergence rate in ergodic and nonergodic sense, respectively. The numerical tests on an extensive collection of problems demonstrate the efficiency of the proposed method.
Citation
@techreport{gu2015semi-proximal-based title={A semi-proximal-based strictly contractive Peaceman-Rachford splitting method}, author={Gu, Yan and Jiang, Bo and Han, Deren}, year={2015}, institution={Optimization online} }
Article
View A semi-proximal-based strictly contractive Peaceman-Rachford splitting method