The Method of Pairwise Variations with Tolerances for Linearly Constrained Optimization Problems

We consider a method of pairwise variations for smooth optimization problems, which involve polyhedral constraints. It consists in making steps with respect to the difference of two selected extreme points of the feasible set together with special threshold control and tolerances whose values reduce sequentially. The method is simpler and more flexible than the well-known conditional gradient method, but keeps its useful sparsity properties and is very suitable for large dimensional optimization problems. We establish its convergence under rather mild assumptions. Efficiency of the method is confirmed by its convergence rates and results of computational experiments.