CSP search algorithms are exponential in the worst-case. A trivial upper bound on the time complexity of CSP search algorithms is O*(dn), where n and d are the number of variables and the maximal domain size of the underlying CSP, respectively.
In this paper we show that a combination of heuristic methods of constraint solving can reduce the time complexity. In particular, we prove that the FC-CBJ algorithm combined with the fail-first variable ordering heuristic (FF) achieves time complexity of O*((d – 1)n), where n and d are the number of variables and the maximal domain size of the given CSP, respectively. Furthermore, we show that the combination is essential because neither FC-CBJ alone nor FC with FF achieve the above complexity. The proposed results are interesting because they establish connection between theoretical and practical approaches to CSP research.
[1]
Fahiem Bacchus,et al.
Extending Forward Checking
,
2000,
CP.
[2]
J. Christopher Beck,et al.
Trying Again to Fail-First
,
2004,
CSCLP.
[3]
Gerhard J. Woeginger,et al.
Exact Algorithms for NP-Hard Problems: A Survey
,
2001,
Combinatorial Optimization.
[4]
Robert M. Haralick,et al.
Increasing Tree Search Efficiency for Constraint Satisfaction Problems
,
1979,
Artif. Intell..
[5]
Patrick Prosser,et al.
HYBRID ALGORITHMS FOR THE CONSTRAINT SATISFACTION PROBLEM
,
1993,
Comput. Intell..
[6]
Patrick Prosser,et al.
MAC-CBJ: maintaining arc consistency with conflict-directed backjumping
,
1995
.