Multi-stage stochastic optimization: the distance between stochastic scenario processes (Q2356157)

The stochastic programming problems with the stage-wise available information and information constraints are considered. A solution method proposed is based on the approximation techniques using nested distributions and their distances. The proposed type of distances between a stochastic scenario process and a tree, which includes infinite trees, and two stochastic processes given by their continuous distributions, facilitates the distribution quantization. The latter is applied to the development of approximation techniques for multi-stage stochastic programming problems. Two examples of application of the theoretical results are included: a multi-stage inventory control problem, and the catastrophic risk-management.