No abstract

No abstract

Abstract. The emerging large-scale computational grid infrastructure is providing an interesting platform for massive distributed computations. In this paper the problem of exploiting such computational grids for solving challenging propositional satisfiability problem (SAT) instances is studied. When designing a distributed algorithm for a large loosely coupled computational grid, a number of grid specific problems need to be tackled including the heterogeneity of the resources, inherent communication delays, and high failure probabilities of grid jobs. In this work a novel distribution method for solving SAT problem instances, called scattering, is introduced. The key advantages of scattering are that it can be used in conjunction with any sequential SAT solver (including industrial black box solvers), the distribution heuristic is strictly separated from the heuristic used in sequential solving, and it requires no communication between processes solving subproblems but still allows coordination of such processes. An implementation of the method has been developed for NorduGrid, a large widely distributed production-level grid running in Scandinavia. The implementation has been benchmarked with test cases including random 3SAT and challenging industrial benchmarks used in previous SAT competitions.

Abstract. Function summarization can be used as a means of incremental verification based on the structure of the program. HiFrog is a fully featured function-summarization-based model checker that uses SMT as the modeling and summarization language. The tool supports three encoding precisions through SMT: uninterpreted functions, linear real arithmetics, and propositional logic. In addition the tool allows optimized traversal of reachability properties, counter-example-guided summary refinement, summary compression, and user-provided summaries. We describe the use of the tool through the description of its architecture and a rich set of features. The description is complemented by an experimental evaluation on the practical impact the different SMT precisions have on model-checking.

In this paper we present Verification-Aided Regression Testing (VART), a novel extension of regression testing that uses model checking to increase the fault revealing capability of existing test suites. The key idea in VART is to extend the use of test case executions from the conventional direct fault discovery to the generation of behavioral properties specific to the upgrade, by (i) automatically producing properties that are proved to hold for the base version of a program, (ii) automatically identifying and checking on the upgraded program only the properties that, according to the developers' intention, must be preserved by the upgrade, and (iii) reporting the faults and the corresponding counter-examples that are not revealed by the regression tests. Our empirical study on both open source and industrial software systems shows that VART automatically produces properties that increase the effectiveness of testing by automatically detecting faults unnoticed by the existing regression test suites.