Code, benchmarks, and experiment data for the ICAPS 2022 paper "Detecting Unsolvability Based on Separating Functions"

This bundle contains code, scripts and benchmarks for reproducing all experiments reported in the paper. It also contains the data generated for the paper.

christen-et-al-icaps2022-code.zip contains two code bases, f2-separating-potential-functions (which is the implementation of the approach described in the paper) and aidos (which is a version of the Aidos planner from https://zenodo.org/record/4058213 with slight modifications to make it compatible with newer Python versions). 

christen-et-al-icaps2022-benchmarks.zip contains all benchmarks used in the evaluation of the paper. The folder lights-out contains a new PDDL encoding of the Lights Out puzzle. The folder sliding-tiles-korf24 contains a PDDL encoding of unsolvable versions of the 24-puzzle instances described in [1]. Finally, the folder uipc contains the benchmarks used in the unsolvability IPC 2016 (https://github.com/AI-Planning/unsolve-ipc-2016/tree/master/domains/FINAL).

christen-et-al-icaps2022-lab.zip contains a copy of Lab 7.0 (https://github.com/aibasel/lab).

christen-et-al-icaps2022-data.zip contains the experiment scripts and data. Directories without the "-eval" ending contain raw data, distributed over a subdirectory for each experiment. Each of these contain a subdirectory tree structure "runs-*" where each planner run has its own directory. For each run, it contains: the run log file "run.log" (stdout), possibly also a run error file "run.err" (stderr), the run script "run" used to start the experiment, the PDDL files for the task, and a "properties" file that contains data parsed from the log file(s). Directories with the "-eval" ending contain a "properties" file - a JSON file with the combined data of all runs of the corresponding experiment. In essence, the properties file is the union over all properties files generated for each individual planner run. Note that some adjustments to the scripts would need to be done for rerunning the experiments because, e.g., the entire tree is not a repository anymore.

[1] Korf, R. E.; and Felner, A. 2002. Disjoint Pattern Database Heuristics. AIJ, 134(1–2): 9–22.