Scatter search to generate diverse MIP solutions

An objective function often is only a rough approximation of the actual goals of the organization and its stakeholders. Consequently, an optimal solution may be no more interesting than other solutions that provide good values. Beyond this, a set of “good” solutions whose members have diverse characteristics can be significantly more valuable for practical analysis and planning than any single, “best” solution. Yet the challenge of systematically uncovering such a diverse set of solutions, or even postulating what its defining features may be, has been conspicuously neglected. We address this challenge for 0-1 mixed integer programming problems and provide demonstrations of the computational efficiency of our approach. Published in: OR Computing Tools for Modeling, Optimization and Simulation: Interfaces in Computer Science and Operations Research, Eds. M. Laguna and J.L. González-Velarde, Kluwer Academic Publishers, 2000, pp. 299 – 317. 2 Glover, Løkketangen and Woodruff