Tabu search—Uncharted domains

Abstract

The setting could have come from a Hollywood science fiction movie. A dozen figures, a handful in uniforms of the U.S. Strategic Air Command (SAC) and a slightly larger contingent variously in business suits and in shirtsleeves, were gathered around a large mainframe computer whose blinking lights signaled a run in progress. The computer run — on an early70s behemoth that represented the state-of-the-art of the period—was seeking a response to a scenario of a first wave nuclear strike on the U.S.. Each of us present had a security clearance at the level of Secret or higher. Mine, a lowly Secret clearance, had come via an engagement with the Defense Communications Agency at the other-worldly SAC installation buried beneath Cheyenne Mountain. My involvement in the present project was the result of a labyrinthine trail that had wended its way through my earlier associations with research groups at Carnegie Mellon University and at the University of California, Berkeley to reach me at my present post at the University of Colorado. The computer scenario currently in progress was engaged in testing a new type of search procedure applied to the objective of discovering an appropriate counter to the hypothetical nuclear strike. The conditions surrounding this objective, which hinged on satisfactorily meeting a collection of prioritized sub-goals, were embodied in a complex combinatorial optimization model. For the search routine to be successful, an effective retaliatory response had to be identified in less than 5 minutes. Experience suggested such an undertaking might not be easy. Previous efforts, utilizing a diverse array of approaches implemented by the “shirtsleeves crowd” surrounding me — a computer science and OR group from a California Think Tank organization engaged by SAC for this project — had been unable to generate an effective response in under 47 minutes, and the resulting solution had been deemed marginal. A concerted initiative to remedy this outcome had, reasonably or unreasonably, wound up on my doorstep and confronted me with the challenge of designing an approach that might do better. The California Think Tank group did not conceal their doubts about the likelihood of my pulling this off, and had openly