Linear complementarity problems solvable by integer programming

Abstract

In this paper it is shown how to solve a linear complementarity problem (or verify that a solution does not exist) using different equivalent mixed zero-one integer programming formulations. In the general case, the number of zero-one integer variables introduced is equal to the dimension of the problem. When it is known that a solution exists, the number of integer variables can be reduced, on the average, using the solution of a multiple-cost-row program.