Study: Symmetry breaking for ASP
Artificial Intelligence
2012-12-13 v1
Abstract
In their nature configuration problems are combinatorial (optimization) problems. In order to find a configuration a solver has to instantiate a number of components of a some type and each of these components can be used in a relation defined for a type. Therefore, many solutions of a configuration problem have symmetric ones which can be obtained by replacing some component of a solution by another one of the same type. These symmetric solutions decrease performance of optimization algorithms because of two reasons: a) they satisfy all requirements and cannot be pruned out from the search space; and b) existence of symmetric optimal solutions does not allow to prove the optimum in a feasible time.
Cite
@article{arxiv.1212.2657,
title = {Study: Symmetry breaking for ASP},
author = {Anna Ryabokon},
journal= {arXiv preprint arXiv:1212.2657},
year = {2012}
}