Entropic optimization: a novel framework for optimization/decision making under uncertainty

A novel framework for optimization/decision making under uncertainty is proposed. Within this framework, a solution is associated with a map from the set of possible system states (scenarios) and the set of possible decisions. This allows for a consideration of solutions with nonzero entropy on equal footing with traditional zero-entropy solutions that correspond to all-into-one maps. Thus the not previously realized trade-off between solution suboptimality and its uncertainty can be explored. One of practical benefits of the proposed approach is a method for identifying cases where a much higher quality decision can be obtained by acquiring a small amount of additional information.