A Symmetry-Breaking Node Equivalence for Pruning the Search Space in Backtracking Algorithms
Autor: | Luka Fürst, Jurij Mihelič, Uros Cibej |
---|---|
Jazyk: | angličtina |
Rok vydání: | 2019 |
Předmět: |
Optimization problem
iskanje monomorfizmov Physics and Astronomy (miscellaneous) Computer science General Mathematics Backtracking 0102 computer and information sciences 02 engineering and technology symmetry breaking 01 natural sciences Monomorphism search search tree pruning razbijanje simetrij backtracking udc:004:519.17 graph equivalence graph algorithm 0202 electrical engineering electronic engineering information engineering Computer Science (miscellaneous) Graph algorithms Symmetry breaking Equivalence (formal languages) grafna ekvivalenca business.industry lcsh:Mathematics Graph algorithm Usability lcsh:QA1-939 Search tree sestopanje rezanje iskalnega drevesa TheoryofComputation_MATHEMATICALLOGICANDFORMALLANGUAGES 010201 computation theory & mathematics Chemistry (miscellaneous) Computer Science::Programming Languages Graph equivalence monomorphism search 020201 artificial intelligence & image processing Search tree pruning business Algorithm algoritem na grafih |
Zdroj: | Symmetry Volume 11 Issue 10 Symmetry, Vol 11, Iss 10, p 1300 (2019) Symmetry, vol. 11, no. 10, 1300, 2019. |
ISSN: | 2073-8994 |
DOI: | 10.3390/sym11101300 |
Popis: | We introduce a new equivalence on graphs, defined by its symmetry-breaking capability. We first present a framework for various backtracking search algorithms, in which the equivalence is used to prune the search tree. Subsequently, we define the equivalence and an optimization problem with the goal of finding an equivalence partition with the highest pruning potential. We also position the optimization problem into the computational-complexity hierarchy. In particular, we show that the verifier lies between P and NP -complete problems. Striving for a practical usability of the approach, we devise a heuristic method for general graphs and optimal algorithms for trees and cycles. |
Databáze: | OpenAIRE |
Externí odkaz: | |
Nepřihlášeným uživatelům se plný text nezobrazuje | K zobrazení výsledku je třeba se přihlásit. |