A cyclic block coordinate descent method with generalized gradient projections

Abstract

The aim of this paper is to present the convergence analysis of a very general class of gradient projection methods for smooth, constrained, possibly nonconvex, optimization. The key features of these methods are the Armijo linesearch along a suitable descent direction and the non Euclidean metric employed to compute the gradient projection. We develop a very general framework from the point of view of block--coordinate descent methods, which are useful when the constraints are separable.

arXiv admin note: substantial text overlap with arXiv:1406.6601

References in corpus (2)

• Accelerated gradient methods for the X-ray imaging of solar flares
• A blind deconvolution method for ground based telescopes and Fizeau interferometers

Cited by in corpus (2)

• Block Coordinate Descent for smooth nonconvex constrained minimization
• Randomized Bregman Coordinate Descent Methods for Non-Lipschitz Optimization