PolyhedralDangerInvariants

Polyhedral Danger Invariants

High level comparison of danger invariants and safety invariants implementation

Commonalities

• Input is ICFG (interprocedural control flow graph)

• We map templates to locations of program

• Calls and Returns are not yet supported. We use our Boogie procedure inlining to handle programs with several procedures (hence, we cannot handle recursive programs).

• Polyhedra-based (both use Boolean combinations of linear inequalities) Need translation of transition relation into linear inequalities.

• Apply Motzkin transformation to constraints in order to reduce non-linear arithmetic and to get rid of quantifier alternation

• Use SMT solver to get satisfying assignments for contraints. Apply simplification to obtain simple solution.

• We start with small templates and iteratiable increase the size of the template.

• Use large block encoding to improve performance (LargeBlockEncodingIcfgTransformer, mApplyLargeBlockEncoding)

• Use live variables to optimize size of templates (generateLiveVariables, USE_LIVE_VARIABLES)

Differences

• Synthesis of safety invariants also works with overapproximations of transition relation.
• We can check correctness of a safety invariant (by checking Hoare triples)
• safety invariants: error locations annotated with false, danger invariants: error locations annotated with true

Tasks

• Translation of transition relation into linear inequalities: check if we can detect overapproximations.
• How can we check correctness of a danger invariant (warning: do not implement this, something similar might have been already implemented in Ultimate)?
• Which statistical data is interesting for our synthesis of danger invariants?