Using CSP Look-Back Techniques to Solve Exceptionally Hard SAT Instances

  title={Using CSP Look-Back Techniques to Solve Exceptionally Hard SAT Instances},
  author={Roberto J. Bayardo and Robert C. Schrag},
  booktitle={International Conference on Principles and Practice of Constraint Programming},
  • R. BayardoR. Schrag
  • Published in
    International Conference on…
    19 August 1996
  • Computer Science
While CNF propositional satisfiability (SAT) is a sub-class of the more general constraint satisfaction problem (CSP), conventional wisdom has it that some well-known CSP look-back techniques — including backjumping and learning — are of little use for SAT. We enhance the Tableau SAT algorithm of Crawford and Auton with look-back techniques and evaluate its performance on problems specifically designed to challenge it. The Random 3-SAT problem space has commonly been used to benchmark SAT… 

Look-Ahead Versus Look-Back for Satisfiability Problems

A very simple DPL procedure called Satz is proposed which only employs some look-ahead techniques: a variable ordering heuristic, a forward consistency checking (Unit Propagation) and a limited resolution before the search, where the heuristic is itself based on unit propagation.

Using CSP Look-Back Techniques to Solve Real-World SAT Instances

The results show that incorporating CSP look-back techniques renders easy many problems which otherwise are beyond DP's reach, and recommend that such techniques be included as options in implementations of DP, just as they are in systematic algorithms for the more general constraint satisfaction problem.

The Effect of Nogood Recording in DPLL-CBJ SAT Algorithms

This paper proposes to investigate the actual usefulness of these CSP techniques in SAT solvers, with the objective of evaluating the actual role played by each individual technique.

The Quest for Efficient Boolean Satisfiability Solvers

This paper examines some of the main ideas along this passage that have led to the current capabilities of the basic DPLL search framework, and focuses on techniques with consistent demonstrated efficiency in available solvers.

CSP problems as algorithmic benchmarks: measures, methods and models

This thesis contributes to both problems, the one to comprise better the problems hardness and the other one, to create tools to build, as desired, sets of problems, with the selected hardness and having some prede ned characteristics.

Hard SAT and CSP instances with Expander Graphs

The experimental investigation conducted on complete and incomplete solvers confirms that the expansion of the graph is indeed a key factor in order to obtain harder instances than other approaches.

Nogood Processing in CSPs

This thesis proposes that a fundamental restriction of nogood learning, which is intended to be the analogous to clause learning in CSPs, can be lifted and shows that more efficient methods can be integrated into specific global constraints and demonstrated on several widely used global constraints.

The Search for Satisfaction

The state of the art for research into sati ability is reviewed, and applications in which algorithms for satis ability have proved successful are discussed.

Towards Industrial-Like Random SAT Instances

This work provides different generation models of SAT instances, extending the uniform and regular 3-CNF models, based on the use of non-uniform probability distributions to select variables that will allow us to generate random instances similar to industrial instances, of interest for testing purposes.

Complete Local Search for Propositional Satisfiability

An algorithmic framework for complete local search and an instantiation for the propositional satisfiability problem (SAT), to use constraint learning in combination with a novel objective function that converges during search to a surface without local minima are presented.



Improvements to propositional satisfiability search algorithms

This dissertation examines complete search algorithms for SAT, the satisfiability problem for propositional formulas in conjunctive normal form, and argues that any efficient SAT search algorithm should perform only a few key technique at each node of the search tree.

Some ideas on random generation of k-SAT instances

This research focuses on the two following questions : is it possible to use conditions based on symmetry in random generation in order to generate even harder instances and what lies between polynomial and NP-Complete instances.

A New Method for Solving Hard Satisfiability Problems

A greedy local search procedure called GSAT is introduced for solving propositional satisfiability problems and its good performance suggests that it may be advantageous to reformulate reasoning tasks that have traditionally been viewed as theorem-proving problems as model-finding tasks.

Experimental Results on the Crossover Point in Random 3-SAT

Sparse Constraint Graphs and Exceptionally Hard Problems

It is shown that, in binary constraint satisfaction problems, ehps are much more likely to occur when the constraints are sparse than when they are dense, and this cost can be dramatically reduced by using conflict-directed backjumping rather than a chronological backtracker.

Intelligent backtracking on constraint satisfaction problems: experimental and theoretical results

It is shown that for a large class of constraint satisfaction algorithms, there is a trade-off in intelligent backtracking between space and time, and whether it is possible for a polynomial-space algorithm to be exponential in the induced width is shown.

The Satisfiability Constraint Gap

Easy Problems are Sometimes Hard

Experimental Results on the Crossover Point in Satisfiability Problems

Empirically, it is found that for random 3-SAT problems below the crossover point, the average time complexity of satisfiability problems seems empirically to grow linearly with problem size, and at and above therossover point the complexity seems to grow exponentially, but the rate of growth seems to be greatest near the crossoverpoint.

On the Random Generation of 3SAT Instances

It is shown that it is possible to generate randomly k-SAT instances that are harder (or easier) than those obtained with the usual generation model, and the same method permits also the random generation of hard satissable instances.