Stochastic linear programming with simple recourse arises naturally in economic problems and other applications. One way to solve it is to discretize the distribution functions of the random demands. This will considerably increase the number of variables and will involve discretization errors. Instead of doing this, we describe a method which alternates between solving some n-dimensional linear subprograms and some m-dimensional convex subprograms, where n is the dimension of the decision vector and m is the dimension of the random demand vector. In many cases, m is relatively small. This method converges in finitely many steps.
|Number of pages||8|
|Journal||Mathematical Programming Study|
|Publication status||Published - 1 Apr 1986|
ASJC Scopus subject areas