Results on Preprocessing, Postprocessing and Node Processing in Industrial LP And MIP Models
OMP, a general solver for industrial LP and MIP models, is based on the simplex and branch and bound algorithms. Performance has been substantially improved through the implementation of various techniques, which reduce the model size and prune the branch and bound tree. Furthermore, the basic algorithms have been made faster, and the choice of branching variables has been sophisticated. The model reduction techniques include variable and constraint elimination and substitution. Pruning of the branch and bound tree is obtained by sharpening variable bounds, applying coefficient reduction to the constraints, and generating cuts.