SL-COMP

Competition of Solvers for Separation Logic

The Separation Logic Competition (SL-COMP) is an initiative to bring together researchers and practitioners interested on improving the state of the art of automated deduction methods for Separation Logic. The competition is run on more than 1K decision problems for different theories of Separation Logic. The solvers are compared on each theory on the number of problems solved and the running time. This initiative started at FLOC 2014 and it has been afiliated with different events: SMT-COMP 2014, FLOC 2018 workshop ADSL 2018 and TOOLympics 2019.

NEWS

Specification and rules

The competition compares solvers for decision problems in Separation Logic with respect to effectiveness and running time. The competition consists of two phases: a training phase, in which solver developers try their tool on the competition benchmark and may provide feedback to organizers, and an evaluation phase, in which all participating solvers are executed on benchmark problems, and the number of correctly solved instances as well as the runtime is measured.

A decision problem is either a satisfiability or an entailment problem in a fixed fragment of Separation Logic. The possible answers of a solver are: sat, unsat and unknown. The answer is compared with the known status of the problem specified in the problem’s file. The result of the comparison determines the evaluation of the solver on this problem, which is correct, incorrect or unsolved.

The problems are specified using the format described here and commented in this post. The input format of problems extends the format SMT-LIB with SL constructs, and exploit the new features of SMT-LIB like datatypes definition and mutually recursive functions.

Teams may contribute with problems in the input format above. The number of problems submitted by a team in some division is limited. The problems submitted are revised by the organizing committee which may or may not accept it. An accepted problem may be changed when this improves the overall quality of the final set of problems.

The solvers shall run on the StarExec platform. Solvers may use a pre-processor to transform the input file format to the solver’s format.The time spent on this pre-processor is not included in the evaluation time. The input problems are not scrambled before their submission to he solver.

Benchmark and divisions

The set of decision problems collected until now are available for browsing and download on the following GIT repository.

Each problem is contributed by a team, as specified in the file header. Thanks to all participants that contributed programs, sent patches, and commented on the sets.

The file header also specifies the status of the problems (sat, unsat) and, most importantly, the logic fragment used in this problem. A division is defined by its logic fragment. The following division are now present:

Tools

The GIT repository provides tools for parsing the input format (in C, C++ and Ocaml). Based on these tools, the organizing committee provides pre-processors of this input to solvers’ input format.

The same repository includes a translator of the input format used for SL-COMP 2014 to the new input format.

Participants

The following solvers have participated to at least one edition of SL-COMP:

Previous editions

Papers

Mailing list

Any question related to this competition shall be sent to the organisation committee and to the mailing list.