Executables:

Executable	Link	Source code
WASP	64-bit linux binary	Download
Paracoherent rewriter	64-bit linux binary	Download. Instructions: g++ --std=c++11 paracoherent_rewriter.cpp -o paracoherent_rewriter

External Executables:

Executable	Link/Source code
asprin	Download from github.com
emax (private communication with the first author of the paper).	Link to the paper

Minimal Models:

asprin-d: (echo "#preference(p1, subset){gap(X)}. #optimize(p1)."; cat $1 | ./paracoherent_rewriter semistable plain asp) | ./asprin --models=0 --ground-once asprin-h: (echo "#preference(p1, subset){gap(X)}. #optimize(p1)."; cat $1 | ./paracoherent_rewriter semistable plain asp) | ./asprin --models=0 --approximation=heuristic --ground-once asprin-w: (echo "#preference(p1, subset){gap(X)}. #optimize(p1)."; cat $1 | ./paracoherent_rewriter semistable plain asp) | ./asprin --models=0 --approximation=weak --ground-once wasp_one: cat instance.gringo | ./paracoherent_rewriter semistable plain gringo | ./wasp --minimize-predicate=gap -n0 --minimization-algorithm=core-based wasp_opt: cat instance.gringo | ./paracoherent_rewriter semistable plain gringo | ./wasp --minimize-predicate=gap -n0 --minimization-algorithm=preferences wasp_min: cat instance.gringo | ./paracoherent_rewriter semistable plain gringo | ./wasp --minimize-predicate=gap -n0 --minimization-algorithm=guess-check-minimize wasp_split: cat instance.gringo | ./paracoherent_rewriter semistable plain gringo | ./wasp --minimize-predicate=gap -n0 --minimization-algorithm=guess-check-split

The experiments were run with the option --silent=2 (for wasp) and --quiet=2 (for asprin) to print only the numbers of enumerated models. To obtain a fair comparison, comment out in the asprin code all the prints about solving/models stats.

We also tested asprin without the option --ground-once (for testing, just remove it in the above code).

All the benchmarks used in the experiments are publicly-available:

Cautious reasoning:

wasp_enum_cautious: cat instance.gringo | ./paracoherent_rewriter semistable plain gringo | ./wasp --minimize-predicate=gap --minimization-cautious-enum wasp_min_cautious: cat instance.gringo | ./paracoherent_rewriter semistable plain gringo | ./wasp --minimize-predicate=gap --minimization-cautious

The experiments were run with the option --silent=2.

All the benchmarks used in the experiments are publicly-available:

MUSes:

Download [wasp.py].

wasp_emax: python3 ./wasp.py enc.asp inst.asp | ./clingo --output=smodels | ./wasp --mus=__debug__ -n0

wasp_camus: python3 ./wasp.py enc.asp inst.asp | ./clingo --output=smodels | ./wasp --mus=__debug__ --mus-algorithm=camus -n0

wasp_camus-1: python3 ./wasp.py enc.asp inst.asp | ./clingo --output=smodels | ./wasp --mus=__debug__ --mus-algorithm=camus,1 -n0

wasp_camus-10: python3 ./wasp.py enc.asp inst.asp | ./clingo --output=smodels | ./wasp --mus=__debug__ --mus-algorithm=camus,10 -n0

wasp_camus-50: python3 ./wasp.py enc.asp inst.asp | ./clingo --output=smodels | ./wasp --mus=__debug__ --mus-algorithm=camus,50 -n0

The experiments were run with the option --silent=2 (for wasp), and we commented out in the emax code all the prints about solving stats.

We used all the benchmarks proposed in the paper: link. Moreover, additional benchmarks (i.e., the ones coming from the ASP competition) will be made available on zenodo in case of acceptance.