Convergence properties of two-stage stochastic programming

From MaRDI portal





The aim of the authors is to investigate a convergence rate of empirical estimates in the case of stochastic programming problems with mathematical expectation in the objective function and a ``deterministic constraint set. Of course, two-stage stochastic programming problems belong to this type of the problems. First, they recall a (rather complete) survey of the papers dealing with the empirical estimates in the case of independent random sample. Papers devoted to dependent (mixing) random samples are not mentioned. Furthermore, employing the ``later results on the large deviations, a valuable generalization of the former results is obtained. Especially, valuable results are introduced for the empirical estimates of the optimal solution. The main results are presented for independent random samples. However, generalization to the Markov type of dependence is also discussed. Special attention is paid to the case of the quadratic objective function. To illustrate the problem as well as to demonstrate the results on a numerical example the vendor problem is employed.




Cited in
(37)








This page was built for publication: Convergence properties of two-stage stochastic programming

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5925743)