DocumentCode :
3572482
Title :
Narrowing Extended Resolution
Author :
Prcovic, N.
Author_Institution :
LSIS, Aix-Marseille Univ., Marseille, France
Volume :
1
fYear :
2012
Firstpage :
556
Lastpage :
563
Abstract :
Extended Resolution (i.e., Resolution incorporating the extension rule) is a more powerful proof system than Resolution because it can find polynomially bounded refutations of some SAT instances where Resolution alone cannot (and at the same time, every proof with resolution is still a valid proof with extended resolution). However it is very difficult to put it into practice because the extension rule is an additionnal source of combinatorial explosion, which tends to lengthen the time to discover a refutation. We call a restriction of Resolution forbiding the production of resolvents of size greater than 3 Narrow Resolution. We show that Narrow Extended Resolution p-simulates (unrestricted) Extended Resolution. We thus obtain a proof system whose combinatorics is highly reduced but which is still as powerful as before. However, the algorithms based on Resolution cannot be easily modified to accommodate this restriction on the resolution rule. This is why we define Splitting Resolution, a variant of Narrow Extended Resolution suitable for integrating into any resolution-based solver.
Keywords :
computability; theorem proving; SAT; combinatorial explosion; narrow extended resolution; polynomially bounded refutation; proof system; resolution-based solver; splitting resolution; Artificial intelligence; Conferences; Erbium; Explosions; Polynomials; Production; Space exploration;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Tools with Artificial Intelligence (ICTAI), 2012 IEEE 24th International Conference on
ISSN :
1082-3409
Print_ISBN :
978-1-4799-0227-9
Type :
conf
DOI :
10.1109/ICTAI.2012.81
Filename :
6495093
Link To Document :
بازگشت