GENGUB: A generator for linear programs with generalized upper bound constraints

The authors report on the development and implementation of a random generator for linear programming (LP) problems. The generator extends previous efforts by considering problems that exhibit a form of special structure often found in real-world applications, namely the presence of generalized upper bound (GUB) constraints. The algorithm allows the user considerable flexibility in the form of the final problems created, and the ability to specify certain problem anomalies at the optimal solution. The generator is of use to those involved in several areas of LP research, including the computational performance of general LP algorithms and the development of methods for detecting special structure such as GUBs within an arbitrary LP.