A Random Generator for Large-Scale Linearly Constrained Quadratic Programming Test Problems

In this paper we describe a random generator for large and sparse quadratic programming problems that frequently arise in different areas of applied science. This generator is an useful tool in testing algorithms on QP problems with different features, since it allows us to vary many parameters which characterize the problems. The procedure used to generate a QP problem as well as some details for its implementation are explained.Finally, we report an analysis of the numerical results, obtained by the routine E04NFK of the NAG library on the test problems produced by the generator.