The previous issues reported were mostly pertaining DSharp since there seems to be some problem with projected compilation. While that is still an issue, I also wanted to report problems with the actual sample generation when D4 is invoked on cnf files with no sample set.
I have attached a couple of example files for which WAPS generates faulty samples and also a script (requiring pysat) to test whether the samples actually satisfy the formula. Almost all samples generated from register..cnf are wrong, while for lime_constraints.cnf, 3-4 samples out of ~30k are wrong.
The previous issues reported were mostly pertaining DSharp since there seems to be some problem with projected compilation. While that is still an issue, I also wanted to report problems with the actual sample generation when D4 is invoked on cnf files with no sample set.
I have attached a couple of example files for which WAPS generates faulty samples and also a script (requiring pysat) to test whether the samples actually satisfy the formula. Almost all samples generated from register..cnf are wrong, while for lime_constraints.cnf, 3-4 samples out of ~30k are wrong.
Thanks!
waps_faulty_sampling.zip