A translator from SMTLIB2 to QDIMACS. Restricted to the bitvector theory and still incomplete.

SMTLIB2 is the standard format for Satisfiability Modulo Theory (SMT) solvers and used in many industrial and academic tools. Recent progress in algorithms for Quantified Boolean Formulas (QBF) makes it tempting to run QBF solvers on problems formulated in SMTLIB2 that contain quantifiers. I therefore implemented a translation to QDIMACS, the standard format for QBF solvers.

It turns out it is surprisingly hard to bitblast quantified formulas and maintain a variable mapping. This script is a partial implementation using the python Z3 API. The script only understands linear quantifier prefixes (no quantifier trees) and will not necessarily detect quantifiers inside 'Not' or 'Or' operators. Please contribute!

Beware that this script is pretty slow. It may take a couple of seconds to produce formulas. (The bottleneck is the extraction of clauses from Z3. Maybe a reimplementation in C# may help.)

You will need python2.7 and pyZ3 to execute the script.

python2 smtlib2qdimacs.py -o=output.qdimacs input.smt

If no output file is specified, the tool will write to the file input.smt.qdimacs. Use the command line option '-2' to negate formulas. Depending on your problem, this may reduce the number of quantifier alternations by one.