Flata

From Numerical Transition Systems
Revision as of 12:20, 19 December 2014 by Radu iosif (Talk | contribs)

Jump to: navigation, search

FLATA <ref name="fm12">"A Verification Toolkit for Numerical Transition Systems." H. Hojjat, R. Iosif, F. Garnier, F. Konecny, V. Kuncak, and P. Rummer. In Proc. of FM'12, volume 7436 of LNCS, pages 247-251, 2012. Springer-Verlag. </ref> is a toolset for the manipulation and the analysis of non-deterministic integer programs (also known as counter automata). The main functionalities of FLATA are:

  • reachability analysis of non-recursive programs - checking if an error control state is reachable
  • termination analysis of non-recursive programs - computation of termination preconditions
  • computation of summaries of recursive programs

Download

FLATA is a free software under LGPL license. The current distribution of FLATA is available as an [File:Flata.tar.gz archive] or on this git repository

Prerequisites:

  • JAVA version 1.6.0 or later
  • YICES has to be installed in your executable path
  • GLPK Java has to be installed in LD_LIBRARY_PATH (required for termination analysis only)


Run

The input to the tool is a textual description of a counter automaton, essentially a control flow graph with edges labeled with arithmetic relations. A good way to get started using FLATA is to go through some of the examples (a subset of NTS benchmarks that FLATA can verify) contained in the distribution and run FLATA as e.g.:

  • reachability analysis ./flata-reachability.sh benchmarks-reach/VHDL/synlifo.correct.nts
  • termination analysis ./flata-termination.sh benchmarks-term/anubhav.correct.nts

Reachability Analysis

The reachability analysis semi-algorithm implemented in FLATA is based on computatation of procedure summaries. The core of the method is an algorithm for computing transitive closures of octagonal relations <ref name="cav10">"Fast Acceleration of Ultimately Periodic Relations." M. Bozga, R. Iosif, and F. Konecny. In Proc. of CAV'10, volume 6174 of LNCS, pages 227-242, 2010. Springer-Verlag. </ref><ref name="cav10-TR">"Relational Analysis of Integer Programs" M. Bozga, R. Iosif, and F. Konecny. VERIMAG technical report, TR-2012-10, 2012. </ref>.

Examples

  • ./flata-reachability.sh benchmarks-reach/VHDL/synlifo.correct.nts (a correct program)
  • ./flata-reachability.sh benchmarks-reach/L2CA/listcounter.error.nts (program with a counterexample trace)

Termination Analysis

The semi-algorithm implemented in FLATA first attempts to compute a transition invariant as a disjunction of octagonal relations (by adapting the procedure summary algorithm) and then computes a termination precondition by applying an algorithm that computes the weakest termination precondition of octagonal relations <ref name="tacas12"> Deciding Conditional Termination. M. Bozga, R. Iosif, and F. Konecny. In Proc. of TACAS'12, volume 7214 of LNCS, pages 252-266, 2012. Springer-Verlag. (Extended journal submission.) </ref>.

Example

  • ./flata-termination.sh benchmarks-term/anubhav.correct.nts

Verification of Recursive Programs

Given a recursive program, FLATA attempts to compute its summary by computing increasingly precise underapproximations of the program <ref name="tacas13">"Underapproximation of Procedure Summaries for Integer Programs." P. Ganty, R. Iosif and F. Konecny. In Proc. of TACAS'13. To appear. </ref>. Note that error control states are ignored and that a reachability relation between initial and final control states is computed.

Example

  • ./flata-reachability.sh benchmarks-recur/mccarthy.nts


Decision Procedure for the SAT problem of the Logic SIL

FLATA decides the satisfiability problem of the SIL logic<ref name="lpar-08">"A Logic of Singly Indexed Arrays" P. Habermehl, R. Iosif, T. Vojnar. In Proc. of LPAR'08, volume 5330 of LNCS, pages 558-573, 2008. Springer-Verlag. </ref> by reduction to the reachability problem of flat counter automata.

Examples

  • ./flata-sil.sh benchmarks-sil/rotation_vc.nts (valid formula)
  • ./flata-sil.sh benchmarks-sil/rotation_vc-f.nts (falsifiable formula)

Contributors


Acknowledgements

This work was supported by the French national project ANR-09-SEGI-016 VERIDYC, by the Czech Science Foundation (projects P103/10/0306 and 102/09/H042), the Czech Ministry of Education (projects COST OC10009 and MSM 0021630528), and the internal FIT BUT grant FIT-S-10-1.


References

<references />