Software used for the paper "Counterexample-Guided Abstraction Refinement for Pattern Selection in Optimal Classical Planning"

This software bundle contains a modified version of Fast Downward (based on commit 48ca3e7fb6ab) with the implementations described in the paper and experiment scripts (under experiments/cegar-pdbs) that can be run using Downward-Lab (see https://doi.org/10.5281/zenodo.399255) to reproduce all experiments of the paper (the resulting data is available here: https://doi.org/10.5281/zenodo.2628693). The script paper-tables.py collects all data and creates the reports displayed in the paper. Note that this bundle is a mercurial repository containing several branches needed to reproduce all experiments. While the main implementation resides in the branch cegar-pdbs, there are additional branches used for integrating other code (explicit and symbolic CPC, SCP)