Each directory represents one of the experiments from our paper. All numbered experiments match the experiments in the Evaluation of our paper. The dead_code_elimination experiment allows to reproduce behavior of Table 1 (in the Introduction).
Each experiment has a README.md describing its individual steps. Furthemore, most experiments have a script reproducing all steps of the experiment, called experiment_N.py. We recommend using this script.
To reduce time spent on reproducing our experiments, these scripts define (as global value in form of NUM_INSTANCES or NUM_FORMULAS) a lower number of targets to evaluate on than our actual evaluation (oftentimes 10 instead of 1,000). Feel free to change this value to your needs.
Make sure you have installed Loki and all dependencies before running the experiments. To do so, you can run ../setup.sh.