## Evaluating Search Heuristics and Optimization Techniques in Propositional Satisfiability (2001)

### Cached

### Download Links

- [www.cs.rice.edu]
- [www.mrg.dist.unige.it]
- [www.mrg.dist.unige.it]
- DBLP

### Other Repositories/Bibliography

Venue: | In Proc. of IJCAR 2001, volume 2083 of LNCS |

Citations: | 24 - 11 self |

### BibTeX

@INPROCEEDINGS{Giunchiglia01evaluatingsearch,

author = {E. Giunchiglia and M. Maratea and A. Tacchella and D. Zambonin},

title = {Evaluating Search Heuristics and Optimization Techniques in Propositional Satisfiability},

booktitle = {In Proc. of IJCAR 2001, volume 2083 of LNCS},

year = {2001},

pages = {347--363},

publisher = {Springer}

}

### Years of Citing Articles

### OpenURL

### Abstract

This paper is devoted to the experimental evaluation of several state-ofthe -art search heuristics and optimization techniques in propositional satisfiability (SAT). The test set consists of random 3CNF formulas as well as real world instances from planning, scheduling, circuit analysis, bounded model checking, and security protocols. All the heuristics and techniques have been implemented in a new library for SAT, called SIM. The comparison is fair because in SIM the selected heuristics and techniques are realized on a common platform. The comparison is significative because SIM as a solver performs very well when compared to other state-of-the-art solvers. 1

### Citations

8530 |
Introduction to algorithms
- Cormen, Leiserson, et al.
- 2001
(Show Context)
Citation Context ... search control, and the heuristics of SIM. These parts together, define a basic implementation of the DLL algorithm [14]. The conventions we use to present the data types and algorithms are those of =-=[15]-=-, described at pages. 4,5. In particular, array elements are accessed by specifying the array name followed by the index in square brackets: for example, A[i] indexes the i-th element of the array A. ... |

944 | Temporal constraint networks
- Dechter, Meiri, et al.
- 1991
(Show Context)
Citation Context ...urs also when the formula is satisfiable, but some choice performed way up in the search tree is responsible for the constraints to be violated. The solution, borrowed from constraint network solving =-=[17]-=-, is to jump back over the choices that were not at the root of the conflict, whenever one is found. The corresponding technique is widely known as (conflict-directed) backjumping (CBJ). 4.1 CBJ-LOOK-... |

740 |
A machine program for theorem proving
- Davis, Logemann, et al.
- 1962
(Show Context)
Citation Context ...IM In this section we give a rather detailed description of the data structure, the search control, and the heuristics of SIM. These parts together, define a basic implementation of the DLL algorithm =-=[14]-=-. The conventions we use to present the data types and algorithms are those of [15], described at pages. 4,5. In particular, array elements are accessed by specifying the array name followed by the in... |

705 | Symbolic model checking without BDDs
- Biere, Cimatti, et al.
- 1999
(Show Context)
Citation Context ...lems, 3 planning problems; -- the 32 Data Encryption Standard (DES) problems, see [5, 6]; -- 34 formal equivalence verification problems, see [7]; -- 31 instances of bounded model checking (BMC), see =-=[8, 6]-=-; -- 7 instances of formal verification properties of pipelined circuits from Velev's web page http://www.ece.cmu.edu/mvelev; -- 11 quasigroup problems, see [9]; -- 39 planning problems, including 2 h... |

283 |
and easy distributions of sat problems
- Hard
- 1992
(Show Context)
Citation Context ...onding to the verification problem of a circuit C, typically the SAT-instance consists of different parts, each part corresponding to a sub-circuit of C. Randomly generated tests became popular after =-=[2]-=-, in which it is showed that, using the Fixed Clause Length model (FCL) [3], it is possible to generate very hard instances. Since then, they have been widely used to test SAT solvers' performances. H... |

193 | SATO: An efficient propositional prover
- Zhang
- 1997
(Show Context)
Citation Context ...ly available procedures. Among these, we considered some of the most effective. In particular: SATZ213, a new version of SATZ [11], POSIT ver 1.0 [12], EQSATZ [6], RELSAT ver. 2.00 [10], SATO ver 3.2 =-=[13]-=-. We do not describe these systems, for which we refer to the corresponding citation. Here, we only point out that these systems rely on very different data-structures and/or search heuristics and/or ... |

160 | Improvements to Propositional Satisfiability Search Algorithms - Freeman - 1995 |

156 |
Using CSP look-back techniques to solve real-world SAT instances
- Schrag
- 1997
(Show Context)
Citation Context ...g problems, including 2 hanoi* instances from SATLIB (www.satlib. org), and blocks world, logistics and rockets problems: most of these problems have been already used to test SAT solvers (see, e.g., =-=[10]-=-) and are available from SATLIB as well. Notice that some of these instances are known to be very hard to solve for currently available SAT solvers. This is, e.g., the case for the 8 DES instances cnf... |

125 | Algorithms for the satisfiability (SAT) problem: a survey
- Gu, Purdom, et al.
- 1997
(Show Context)
Citation Context ...(SAT) is fundamental in many areas of computer science such as formal verification, planning and theorem proving. Despite the exponential worst-case complexity of all known algorithms for solving SAT =-=[1]-=-, recent implementations of the Davis-Logemann-Loveland (DLL) algorithm are able to solve problems having thousands of propositions in a few seconds. Much of the success of these solvers is due to cle... |

84 |
Benefits of Bounded Model Checking in an Industrial Setting
- Copti, Fix, et al.
- 2001
(Show Context)
Citation Context ... we introduce backjumping and/or learning. These are clear indications of the obvious fact that the benefits of a heuristics depend on its interactions with the other implemented techniques (see also =-=[18]-=-). 6 Final remarks We conclude the paper with Table 3, comparing SIM with the various settings, and the solvers that we considered in Section 2. From this table, we see that SIMSATZ solves more real w... |

79 | Integrating equivalency reasoning into Davis–Putnam procedure
- Li
- 2000
(Show Context)
Citation Context ... see, e.g., [4]; -- the 16 instances of the Bejing Competition held in 1996: 7 circuit equivalences, 6 scheduling problems, 3 planning problems; -- the 32 Data Encryption Standard (DES) problems, see =-=[5, 6]-=-; -- 34 formal equivalence verification problems, see [7]; -- 31 instances of bounded model checking (BMC), see [8, 6]; -- 7 instances of formal verification properties of pipelined circuits from Vele... |

65 |
Probabilistic analysis of the Davis–Putnam procedure for solving the satisfiability problem
- Franco, Paull
(Show Context)
Citation Context ...ce consists of different parts, each part corresponding to a sub-circuit of C. Randomly generated tests became popular after [2], in which it is showed that, using the Fixed Clause Length model (FCL) =-=[3]-=-, it is possible to generate very hard instances. Since then, they have been widely used to test SAT solvers' performances. However, they lack the structure of real world instances. Given their differ... |

53 | Implementing the Davis–Putnam method
- Zhang, Stickel
(Show Context)
Citation Context ...of bounded model checking (BMC), see [8, 6]; -- 7 instances of formal verification properties of pipelined circuits from Velev's web page http://www.ece.cmu.edu/mvelev; -- 11 quasigroup problems, see =-=[9]-=-; -- 39 planning problems, including 2 hanoi* instances from SATLIB (www.satlib. org), and blocks world, logistics and rockets problems: most of these problems have been already used to test SAT solve... |

44 | Ordered binary decision diagrams and the Davis-Putnam procedure
- Uribe, Stickel
- 1994
(Show Context)
Citation Context ...on held in 1996: 7 circuit equivalences, 6 scheduling problems, 3 planning problems; -- the 32 Data Encryption Standard (DES) problems, see [5, 6]; -- 34 formal equivalence verification problems, see =-=[7]-=-; -- 31 instances of bounded model checking (BMC), see [8, 6]; -- 7 instances of formal verification properties of pipelined circuits from Velev's web page http://www.ece.cmu.edu/mvelev; -- 11 quasigr... |

24 |
Li and Anbulagan, Heuristics Based on Unit Propagation for Satisfiability Problems
- Min
(Show Context)
Citation Context ...s, we use the FCL model. Formulas are thus generated according to the number of propositions N , the number of literals per clause K, and the number of clauses L. Usually, K is fixed to 3 (see, e.g., =-=[11]), N is fi-=-xed to a certain value, while L is varied in order to cover the "100% satisfiable -- 100% unsatisfiable" transition. In our experiments, we considered N = 300. For N = 300, none of the solve... |

18 |
Logical cryptanalysis as a SAT problem
- Massacci, Marraro
- 2000
(Show Context)
Citation Context ... see, e.g., [4]; -- the 16 instances of the Bejing Competition held in 1996: 7 circuit equivalences, 6 scheduling problems, 3 planning problems; -- the 32 Data Encryption Standard (DES) problems, see =-=[5, 6]-=-; -- 34 formal equivalence verification problems, see [7]; -- 31 instances of bounded model checking (BMC), see [8, 6]; -- 7 instances of formal verification properties of pipelined circuits from Vele... |

10 | On the complexity of choosing the branching literal in dpll
- Liberatore
(Show Context)
Citation Context ...sition and a branching order that will cause the exploration of less nodes. Unfortunately, deciding whether a proposition p is optimal is harder than deciding the satisfiability of the formula itself =-=[16]-=-, so we resort to approximations: a good approximate heuristic does not require a lot of computation time and gives higher scores to the propositions that, once assigned, lead to simpler problems. We ... |