Many (and probably most) quantitative decision problems involve some uncertainty as to the eventual consequences of decisions that must be made now. In such an uncertain environment, it is important to make decisions that balance current implementation costs with the costs of potential future decisions contingent on the current decision and the outcomes of uncertain future events. Experience has shown repeatedly that explicit incorporation of the uncertainty in a decision problem, via techniques called stochastic optimization, leads to vastly superior current decisions. This result applies across many application domains, including finance, the energy sector, manufacturing, and distribution.
DECIS™ is a state-of-the-art system for solving general, large-scale stochastic optimization problems for which the uncertain parameters can be characterized by probability distributions. DECIS™ particularly shines in providing approximately optimal solutions to problems with far too many possible outcomes to allow for an exact solution. It does so by judiciously combining methods drawn from both optimization and simulation disciplines. The combination is rigorously proven in theory and efficiently implemented in software available in multiple forms. (More...)