Mathematical Methods in Operations Research and Computer Science

Our progress on the fundamental problem of decision science, solving large-scale optimization problems whose parameters are subject to uncertainty, has been so rapid during the past year that we are now in a position to tackle the most general type of stochastic models one is likely to encounter in practice. It is particularly relevant to the Navy during this critical period of deciding how best to down-size the military and yet maintain a force robust and ready to handle any combination of contingencies that might arise in the future with high probability. A major breakthrough in this area has the potential of profoundly affecting the quality of planning, the reduction of cycle time to improve products, the reduction of time to militarily deploy in crisis, and generally improve the industrial competitiveness of U.S. until the time that other countries catch on. By way of background, our approach uses a combination of decomposition methods D-W and Benders, importance sampling, and if necessary parallel processors.