For the past thirty five years artificial intelligence researchers have been studying heuristic search techniques. People writing AI programs have had strong intuitions about what it means for a program to search. Intuitively they classified programs accor
Lecture Notes for 6.824, Arti cial Intelligence c 1992 David McAllester, all rights reserved
Rev. October, 1992
Constraint Satisfaction Search For the past thirty ve years arti cial intelligence researchers have been studying heuristic search techniques. People writing AI programs have had strong intuitions about what it means for a program to search. Intuitively they classi ed programs according to the kinds of search performed. Certain programs, such as matrix multiplication routines, did no search whatsoever. Other programs searched for solutions to a xed nite set of constraints. Others searched for paths in graphs, or for strategies in game trees, or for proofs in formal inference systems. The intuitive classi cation of search programs seems to roughly correspond to complexity classes. The no-search procedures correspond to the complexity class P. Constraint satisfaction search procedures correspond to the complexity class NP, graph search and game search procedures to the class PSPACE, and search in theorem proving to the class of recursively enumerable functions. This chapter discusses heuristic techniques for solving search problems of the simplest type| constraint satisfaction problems. Heuristic techniques for other forms of search are discussed in later chapters.
1 Constraint Satisfaction Problems Intuitively, a search problem involves nding an object that satis es a given speci cation. A classic example is the eight queens problem. This is the problem of placing eight queens on a chess board so that no two queens attack each other, i.e., so that no single row, column, or diagonal contains more than one queen. This problem can be solved (ine ciently) by a\generate and test" program that rst generates a list of all the possible ways eight queens can be placed a chess board and then tests each possibility to see if it is a solution to the problem. 1
Another example is the SAT problem. This is the problem of determining
Abstract. In this paper we examine the use of Tabu Search for Constraint Satisfaction Problems (CSPs). We show that with Tabu Search for most problems...
Solving partial constraint satisfaction problems using local search and abstraction Partial constraint satisfaction problems (PCSPs) were proposed by Freuder and...
An Empirical Study of Multi-Point Constructive Search for Constraint Satisfaction Ivan Heckman & J. Christopher Beck Department of Mechanical & Industrial ...
Stochastic local search strategies have been recently reintroduced into the satis ability and constraint satisfaction literature under the umbrella name GSAT ...
Backtrack-free search for real-time constraint satisfaction Abstract. A constraint satisfaction problem (CSP) model can be preprocessed to ensure that any ...
The impact of ANDOR search spaces on constraint satisfaction and counting Abstract. The contribution of this paper is in demonstrating the impact of AND/...
Genetic local search algorithm for random binary constraint satisfaction problemsGenetic local search algorithm for random binary constraint satisfaction problems...
risto@cs.utexas.edu Technical Report AI94-218 May 1994 A new method for developing good value-ordering strategies in constraint satisfaction search is ...
Optimal refutations for constraint satisfaction problems Variable ordering heuristics have long been an important component of constraint satisfaction search ...
Abstract. Yet, two classes of algorithms have been used in partial constraint satisfaction: local search methods and branch&bound search extended by the ...
热门文档