Heuristic Improvements in Multi-commodity Flow Algorithm

We propose heuristic improvements in the procedure to obtain a (1+!)-approximation of the multicommodity ow which satises all demands when the cost bound is given as an input parameter. Through a series of improvements we are able to signicantly reduce the number of shortest path computations. We also present a new version of the algorithm given in (1) which is amenable to our optimizations. Finally we compare our results against those given in (2) and demonstrate signicant improvement in the results. We also propose methods to reduce the running time of the algorithm.