Zero-one programming with many variables and few constraints

Abstract One of the more successful techniques for solving zero-one integer programs has been the implicit enumeration strategy first introduced by E. Balas. However, experience has shown that the efficiency of these enumerative techniques depends critically upon the bumber of variables. In this paper an algorithm is developed and computational experience provided for solving zero-one integer programs with many variables and few constraints. Sub-problems solved via implicit enumeration are generated from the linear programming relaxation and the variables in these sub-problems correspond to the fractional variables obtained in the linear program. Since the number of fractional variables in the linear program is bounded by the number of constraints in the linear program, the sub-problems will in general contain many fewer variables than the original zero-one integer program.