Parallel Decomposition of Multistage Stochastic Programming Problems

Ruszczynski, A. (1988). Parallel Decomposition of Multistage Stochastic Programming Problems. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-88-094

[thumbnail of WP-88-094.pdf]
Preview
Text
WP-88-094.pdf

Download (554kB) | Preview

Abstract

A new decomposition method for multistage stochastic linear programming problems is proposed by the author. The method combines the ideas of the regularized decomposition method for two-stage programs and dynamic programming. With each node of the decision tree of the multistage stochastic problem a certain regularized subproblem is associated which generates decisions for its successors and some backward information for its predecessor. The subproblems are solved in parallel and exchange information in an asynchronous way through special buffers. After a finite time the method either finds an optimal solution to the problem or discovers its inconsistency. This method is especially convenient for implementation on a parallel computer.

Item Type: Monograph (IIASA Working Paper)
Research Programs: System and Decision Sciences - Core (SDS)
Depositing User: IIASA Import
Date Deposited: 15 Jan 2016 01:58
Last Modified: 27 Aug 2021 17:13
URI: https://pure.iiasa.ac.at/3112

Actions (login required)

View Item View Item