It would be nice to calculate fixpoint on some unit test of program, then interpret the program and compare fixpoint with the actual data during the interpretation.
use LLVM interpreter to interpret program and check variable values against calculated fixpoint
instrument LLVM bytecode (for example via LLVM Pass) to call our library, which would check given value against value of the same variable in calculated fixpoint
implement "real value" domains - domain representing real integer, real float etc. and "calculate fixpoint" - execute the program and when executing, check the value of real value against calculated fixpoint
It would be nice to calculate fixpoint on some unit test of program, then interpret the program and compare fixpoint with the actual data during the interpretation.