Implementations of the Simplex method differ only in very specific aspects such as the pivot rule. Similarly, most relaxation methods for mixed-integer programming differ only in the type of cuts and the exploration of the search tree. Implementing instances of those frameworks would therefore be more efficient if linear and mixed-integer programming solvers let users customize such aspects easily. We provide a scripting mechanism to easily implement and experiment with pivot rules for the Simplex method by building upon COIN-OR’s open-source linear programming package CLP. Our mechanism enables users to implement pivot rules in the Python scripting language without explicitly interacting with the underlying C++ layers of CLP. In the same manner, it allows users to customize the solution process while solving mixed-integer linear programs using the CBC and CGL COIN-OR packages. The Cython programming language ensures communication between Python and COIN-OR libraries and activates user-defined customizations as callbacks. For illustration, we provide an implementation of a well-known pivot rule as well as the positive edge rule—a new rule that is particularly efficient on degenerate problems, and demonstrate how to customize branch-and-cut node selection in the solution of a mixed-integer program.
Cahier du GERAD G-2012-07, GERAD, Montreal, Quebec, Canada, March 2012
View Customizing the Solution Process of COIN-OR’s Linear Solvers with Python