Probabilistic estimation of stability of solutions of optimization problems

Abstract

The issues of stability of solution of optimization problems are considered on the example of transportation problem of linear programming, in which the transportation costs are random variables with a known distribution density. The complexity of solving such optimization problems by classical methods is substantiated. It is proved that the problem of estimating stability admits an analytic solution if the optimal solution of the problem is sought using the matrix minimum method. This solution is based on the search for the probability that characterizes the level of stability of solving the optimization problem. The corresponding computational procedure is described. An example is considered that illustrates it graphically.