Abstract

Unpredictability in the running time of complete search procedures can often be explained by the phenomenon of “heavy-tailed cost distributions”, meaning that at any time during the experiment there is a non-negligible probability of hitting a problem that requires exponentially more time to solve than any that has been encountered before (Gomes et al. 1998a). We present a general method for introducing controlled randomization into complete search algorithms. The “boosted ” search methods provably eliminate heavy-tails to the right of the median. Furthermore, they can take advantage of heavy-tails to the left of the median (that is, a nonnegligible chance of very short runs) to dramatically shorten the solution time. We demonstrate speedups of several orders of magnitude for state-of-the-art complete search procedures running on hard, real-world problems.

Citations

636	A machine program for theorem proving - Davis, Logemann, et al. - 1962
463	Pushing the envelope: Planning, propositional logic, and stochastic search - KAUTZ, SELMAN - 1996
323	Dynamic backtracking - Ginsberg - 1993
256	Stable Non-Gaussian Random Processes: Stochastic Models with Infinite Variance - SAMORODNITSKY, S - 1994
207	Domain-independent extensions to GSAT: solving large structured satis ability problems - Selman, Kautz - 1993
182	Using CSP look-back techniques to solve real-world SAT instances - Bayardo, Schrag - 1997
93	Experimental results on the application of satisfiability algorithms to scheduling problems - Crawford, Baker - 1994
75	Heuristic-biased stochastic sampling - Bresina - 1996
75	Easy Problems are Sometimes Hard - Gent, Walsh - 1994
67	B.: Problem structure in the presence of perturbations - Gomes, Selman - 1997
55	Scheduling a major college basketball conference - Nemhauser, Trick - 1998
43	Sparse constraint graphs and exceptionally hard problems - Smith, Grant - 1995
35	The Pareto-Lévy law and the distribution of income - Mandelbrot - 1960
31	Constraint networks. In: Encyclopedia of Artificial Intelligence, 2nd edn - Dechter - 1992
27	Constraint-based reasoning - Freuder, Mackworth, et al. - 1994
24	Stochastic Procedures for Generating Feasible Schedules - Oddi, Smith - 1997
12	An interior point approach to Boolean vector function synthesis - Kamath, Karmarkar, et al. - 1993
8	Min and Anbulagan - Li - 1997
7	Beyond the Black Box: Constraints as objects - Puget, Leconte - 1995
2	Heavy-Tailed Phenomena in Combinatorial Search - Gomes, Selman, et al. - 1998
2	Combinatorial Aspects of Construction of - Schreuder - 1992