NLTGCR: A class of Nonlinear Acceleration Procedures based on Conjugate Residuals

Abstract

This paper develops a new class of nonlinear acceleration algorithms based on extending conjugate residual-type procedures from linear to nonlinear equations. The main algorithm has strong similarities with Anderson acceleration as well as with inexact Newton methods - depending on which variant is implemented. We prove theoretically and verify experimentally, on a variety of problems from simulation experiments to deep learning applications, that our method is a powerful accelerated iterative algorithm.

References in corpus (2)

• Acceleration Methods
• Convergence analysis of adaptive DIIS algorithms with application to electronic ground state calculations