Splitting the Atom: A New Approach to Neighbourhood Interchangeability in Constraint Satisfaction Problems
Abstract
We investigate interchangeability of values in CSPs, based on an approach where a single value in the domain of a variable can be treated as a combination of "sub-values". An algorithm for removing overlapping sub-values is presented. The resulting CSPs take less time to find all solutions and yield a more compactly-representable, but equivalent, solution set. Experimental results show that, especially in loose problems with large numbers of solutions, dramatic savings in search cost are achieved. 1
Cite
Text
Bowen and Likitvivatanavong. "Splitting the Atom: A New Approach to Neighbourhood Interchangeability in Constraint Satisfaction Problems." International Joint Conference on Artificial Intelligence, 2003.Markdown
[Bowen and Likitvivatanavong. "Splitting the Atom: A New Approach to Neighbourhood Interchangeability in Constraint Satisfaction Problems." International Joint Conference on Artificial Intelligence, 2003.](https://mlanthology.org/ijcai/2003/bowen2003ijcai-splitting/)BibTeX
@inproceedings{bowen2003ijcai-splitting,
title = {{Splitting the Atom: A New Approach to Neighbourhood Interchangeability in Constraint Satisfaction Problems}},
author = {Bowen, James and Likitvivatanavong, Chavalit},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {2003},
pages = {1366-1367},
url = {https://mlanthology.org/ijcai/2003/bowen2003ijcai-splitting/}
}