Application of dual projected pseudo quasi Newton algorithm in multicommodity network flow problems

Abstract

In the paper, we apply a method for solving multicommodity network flow problems, which has important applications in communication networks. This method combines a projected quasi-Newton method and a dual projected pseudo quasi-Newton (DPPQN) method. We use a sparse-block-matrix technique and the finite iteration scaled projection technique to enhance the computational efficiency of DPPQN method. Compared with the Frank-Wolfe with PARTAN algorithm, we show the efficiency of our method by several numerical examples.