home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.theory
- Path: sparky!uunet!decwrl!sdd.hp.com!wupost!gumby!destroyer!ubc-cs!unixg.ubc.ca!kakwa.ucs.ualberta.ca!acs.ucalgary.ca!gu
- From: gu@enel.ucalgary.ca (Jun Gu)
- Subject: Satisfiability (I)
- Sender: news@acs.ucalgary.ca (USENET News System)
- Message-ID: <92Sep03.221830.18724@acs.ucalgary.ca>
- Date: Thu, 03 Sep 92 22:18:30 GMT
- References: <1992Sep1.183853.5844@news.cs.indiana.edu>
- Nntp-Posting-Host: eneli.enel.ucalgary.ca
- Organization: ECE Department, U. of Calgary, Calgary, Alberta, Canada
- Keywords: Average time complexity, local search, satisfiability
- Lines: 82
-
- Since last January, I have been receiving messages from many researchers.
- A typical message was one from Paul Purdom on Sept. 1, 1992. Some of the
- questions are fairly interesting. I have been asked by many people and
- I would be happy to share my opinion on some questions in this group.
-
- The first set of questions I sorted out from numerous messages include: why
- we had chosen this algorithm (i.e., SAT) to study, what are other results
- from the past and present research, what are the other SAT algorithm(s),
- how to evaluate the hardness of the SAT problem instances. I will discuss
- these questions soon.
-
- Jun Gu
-
- ---------------------------------------------------------------------------
-
- The following two manuscripts were written last year and were submitted for
- review. You may ask a copy from Ruchir Puri:
-
- puri@enel.UCalgary.CA
-
-
- ---------------------------------------------------------------------------
-
- Average Time Complexities of
- Several Local Search Algorithms for the Satisfiability Problem
-
- Jun Gu and Qian Ping Gu
- Dept. of Electrical and Computer Engineering
- Univ. of Calgary, Canada T2N 1N4
-
- 1991
-
- Abstract (LaTex format)
-
- The satisfiability (SAT) problem is a basic problem in computing theory.
- It is fundamental in solving many practical application problems in logic
- programming, inference, machine learning, and VLSI engineering. Recently,
- many families of local search algorithms for the SAT problem have been
- developed. In this report, we analyze the average time complexities of
- three basic algorithms in the $SAT1$ family. For the randomly generated
- {\it conjunctive normal form} ($CNF$) formulae with $n$ clauses, $m$ variables,
- and $l$ literals in each clause, we give a qualitative analysis of the average
- time complexity of the $SAT1.1$ and $SAT1.2$ algorithms, and a quantitative
- analysis of the average time complexity of the $SAT1.3$ algorithm. For
- $l\geq \log m-\log\log m-c$ and $n/m\leq 2^{l-1}/l$, where $c\geq 0$ is any
- constant, we show that $SAT1.1$ exhibits a polynomial average time complexity
- of $O(m^{O(1)}n)$. For $l\geq 3$ and $n/m=O(2^l/l)$, we show that $SAT1.2$
- exhibits a polynomial average time complexity of $O(m^{O(1)}n^2)$. The average
- run time of the $SAT1.3$ algorithm is $O(m\log n(ln+m))$ for $l\geq 3$ and
- $n/m=O(2^l/l)$. For $l=o(m)$, while all $SAT1.1$, $SAT1.2$, and $SAT1.3$
- algorithms have the polynomial average time complexity, the well-known Davis-
- Putnam algorithm demonstrates a superpolynomial average run time
- ($O(n^{O(m/l)})$). Thus the $SAT1.1$, $SAT1.2$ and $SAT1.3$ algorithms
- have significantly improved on the Davis-Putnam algorithm for certain classes
- of $CNF$ formulae.
-
-
- Average Time Complexity of the SAT1.3 Algorithm
-
- Jun Gu and Qian Ping Gu
- Dept. of Electrical and Computer Engineering
- Univ. of Calgary, Canada T2N 1N4
-
- 1991, Rev. Feb. 1992
-
- Abstract (LaTex format)
-
-
- The satisfiability (SAT) problem is a basic problem in computing theory.
- It is fundamental in solving many practical application problems in logic
- programming, inference, machine learning, and VLSI engineering. In this paper,
- we analyze the average time complexity of the $SAT1.3$ algorithm. For the
- randomly generated $CNF$ formulae with $n$ clauses, $m$ variables, and
- average $l$ literals in each clause, the average run time of the $SAT1.3$
- algorithm is $O(m\log n(ln+m))$ for $l\geq 3$ and $n/m=O(2^l/l)$. This result
- improves significantly on the well-known Davis-Putnam algorithm which has an
- $O(n^{O(m/l)})$ average time complexity. Numerous real algorithm executions
- were performed which indicate that the $SAT1.3$ algorithm is much more
- efficient than the Davis-Putnam algorithm for certain classes of
- {\it conjunctive normal form} ($CNF$) formulae.
-
-
-