Aggregation and Disaggregation in Integer Programming Problems

Column aggregation of 0/1 programming problems is considered. By relaxing the integer requirements in the aggregated problem, an easy problem results, the solution of which can be used to compute bounds on the loss of accuracy in the objective function value. It is shown how these bounds can be improved by identification of valid inequalities and by exploiting primal degeneracy. A small example is given to illustrate the procedure.