0

I thought I was fairly confident in the formulation for “two-stage stochastic program with recourse” until I read the Wikipedia page.

I am used to seeing the problem formulated with one decision variable for the first stage, and another, different decision variable for the second stage (and the decision variable for the second stage is called “recourse decision variable”).

In fact, Wikipedia seems to imply this under the section “Two-stage problems”:

enter image description here

But, then, when I want to find out information about how to actually solve this, I look under the section later on in the document called Stochastic linear program (specifically, “Deterministic equivalent of a stochastic problem”). Now in that case it seems to be treating three decision variables x, y, and z.

enter image description here Why are there three decision variables in the deterministic equivalent? I thought there would be only two, x, and y?

makansij
  • 1,643

1 Answers1

1

The wikipedia page is not uniform in the naming of the variables. In the first part $x$ are first stage and $y$ are second stage variables. In the Deterministic Equalivalent (DE) formulation both $x$ and $y$ are first stage and $z$ are second stage. They make a distinction between $x$ and $y$ in how they appear in the second stage. The $x$ variable do not appear in the second stage whereas the $y$ variables appear both in the first stage and second stage. (The second stage variables $z$ only appear in the second stage).

You can recognize a stage 2 variable as being indexed by the scenario. Stage 1 variables do not have a scenario index.

Erwin Kalvelagen
  • 4,367
  • 2
  • 13
  • 16