A Parallel Projection for the Multicommodity Network Model

We present an application of parallel computing techniques to the solution of a quadratic programme that arises in the resource-directive decomposition method for multicommodity problems. A sequential algorithm for the quadratic programme is discussed, and its extension to a parallel implementation is given. Computational testing of the sequential and parallel algorithms was done on the Sequent Symmetry S81 parallel computer located in the Parallel Processing Laboratory at Southern Methodist University. On several large test problems the parallel version achieved a speed-up of 10 with 12 processors.