Optimality functions in stochastic programming (Q715095): Difference between revisions

A stochastic programming problem is considered with nonlinear and possibly non-convex expected value objective and constraint functions. The concept of an optimality function is extended to stochastic programs, and applied to evaluate the quality of a candidate solution by means of confidence intervals. An algorithm is proposed which almost surely converges to a stationary (in the sense of Fritz-John conditions) point. Several numerical examples are included.