Article ID: | iaor20001122 |
Country: | China |
Volume: | 17 |
Issue: | 2 |
Start Page Number: | 160 |
End Page Number: | 172 |
Publication Date: | May 1995 |
Journal: | Mathematica Numerica Sinica |
Authors: | Tang H.W., Zhang L.W. |
Keywords: | entropy |
In this paper, an ε-optimal solution to a linear programming problem in Karmarkar standard form is given by reducing its dual problem to a differentiable and strictly convex programming via the maximum entropy principle. A perturbation solution of the original linear programming can be achieved by solving this convex programming. Moreover, an algorithm for finding an ε-optimal solution is proposed and the algorithm is proven to be convergent as ε tends to zero. Numerical examples are given in this paper.