Sample average approximation of stochastic dominance constrained programs (Q431031)

The paper develops an algorithm for the solution of optimization problems with multi-variate second-order stochastic dominance constraints. The sample Average Approximation (SAA) method is applied to a relaxation of this problem. For the resulting semi-infinite program, asymptotic convergence of optimal values and optimal solutions (as the sample size goes to infinity) is studied, under suitable structural and distributional assumptions. Additional convexity assumptions are exploited to develop a finitely convergent method that provides an \(\varrho\)-optimal solution of the solved SAA problem. Finally, statistical lower and upper bounds for the optimal value of the relaxed true problem are suggested.