Solving two-stage stochastic programming problems with level decomposition |
| |
Authors: | Csaba I Fábián Zoltán Sz?ke |
| |
Institution: | 1. Department of Operations Research, Loránd E?tv?s University of Budapest, P.O. Box 120, Budapest, 1518, Hungary 2. Doctoral School in Applied Mathematics, Loránd E?tv?s University of Budapest, Budapest, Hungary
|
| |
Abstract: | We propose a new variant of the two-stage recourse model. It can be used e.g., in managing resources in whose supply random
interruptions may occur. Oil and natural gas are examples for such resources. Constraints in the resulting stochastic programming
problems can be regarded as generalizations of integrated chance constraints. For the solution of such problems, we propose
a new decomposition method that integrates a bundle-type convex programming method with the classic distribution approximation
schemes. Feasibility and optimality issues are taken into consideration simultaneously, since we use a convex programming
method suited for constrained optimization. This approach can also be applied to traditional two-stage problems whose recourse
functions can be extended to the whole space in a computationally efficient way. Network recourse problems are an example
for such problems. We report encouraging test results with the new method.
|
| |
Keywords: | Stochastic recourse models Decomposition Successive approximation |
本文献已被 SpringerLink 等数据库收录! |
|