# Deterministic equivalent of a stochastic problem

 1 1 Dear All, I am new to the field of stochastic programming. I am going to formulate and solve a multi-stage stochastic problem by using recourse method. I wondering if I formulate and solve the deterministic equivalent of the problem, will I get the same answer as the stochastic problem? asked 04 Apr '11, 23:14 Meysam 13●1●3 accept rate: 0%

 4 Part of the answer depends on whether you are talking about theory or whether you are looking for computational results. Theoretically speaking, it's always possible to formulate a deterministic equivalent. The question is whether that can be solved directly or whether you have to solve it through other means such as sampling, decomposition, etc. Practically speaking, whether you can solve the deterministic equivalent directly depends primarily on the nature of the random variates: is there a relatively small number of (finite) random variates? This is only true if the random variables have a small number of discrete outcomes. If so, then the deterministic equivalent can be solved as a traditional LP or MIP. If not, then you're better off with sampling methods. answered 04 Apr '11, 23:40 Greg Glockner 437●1●6 accept rate: 20% Many thanks Greg! Very helpful and clear answer, indeed! All I wanted to know was, if I can formulate the deterministic equivalent of a stochastic problem and solve it by using LP/NLP/MIP solvers, OR I have to use a stochastic programming solver. According your answer, I believe I can directly solve the deterministic equivalent bevause I roughly have 10 scenarios (after using scenario reduction algorithm). (06 Apr '11, 08:36) Meysam
 1 Yes, this should work though it could be hard if there are lots of scenarios and stages. I have been working on a pretty simple implementation of Bender's Decomposition for Multistage Stochastic Programs that I can send you if the Deterministic Equivalent gets intractable. The code runs in MATLAB but uses the CPLEX API to solve LPs. I've purposely tried to make user-input as simple as possible. In general, try to stay away from the SP solvers out there, the SMPS file format will drive you nuts. A better option may be to use AIMMS... This has a very nice stochastic programming implementation but can be a little tricky to learn. Feel free to e-mail/message me in case you need anything! answered 30 Apr '11, 02:35 Berk Ustun 307●1●3●11 accept rate: 0% Hi Berk, Could you kindly send me the MATLAB code? I am working on a multi stage SP and the scenarios are getting messier and I am finding it hard to implement Benders. Thanks a ton! (29 Mar '12, 01:30) rahulsharm9
 toggle preview community wiki

By Email:

Markdown Basics

• *italic* or _italic_
• **bold** or __bold__
• image?![alt text](/path/img.jpg "Title")
• numbered list: 1. Foo 2. Bar
• to add a line break simply add two spaces to where you would like the new line to be.
• basic HTML tags are also supported

Tags:

Seen: 4,039 times

Last updated: 29 Mar '12, 12:10