Experimental data and benchmarks used in the paper "Theoretical Foundations for Structural Symmetries of Lifted PDDL Tasks"

This dataset contains both benchmarks and data used in the paper.

PDDL benchmark files can be found in the files benchmarks.tar.gz and
bagged-benchmarks.tar.gz. The former contains all domains from all IPCs from the
repository https://bitbucket.org/aibasel/downward-benchmarks, without duplicate
domains that have been used in multiple IPCs. The latter contains the subset of
these tasks for which the reformulation in the "bagged representation" from the
following paper succeeded:

Riddle, P.; Douglas, J.; Barley, M.; and Franco, S. 2016. Improving
performance by reformulating PDDL into a bagged representation.
In ICAPS 2016 Workshop on Heuristics and Search for Domain-
independent Planning, 28–36.

We obtained the implementation of the baggy reformluation from the authors.

All other files in this dataset contain raw and processed data of all
experiments, which were generated using Downward-Lab (see
https://doi.org/10.5281/zenodo.399255). The scripts used to run the experiments
can be found in the software bundle for this paper (see
https://doi.org/10.5281/zenodo.2621897).

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,
there are symbolic links to the input PDDL files domain.pddl and problem.pddl
(can be resolved by putting the benchmarks directory to the right place), 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, and a "properties"
file that contains data parsed from the log file(s). Some directories (not for
the ground experiments, where these files got too large) also contain a file
generators.py that contain all symmetry group generators in permutation
notation.

Directories with the "-eval" ending contain a "properties" file, which contains
a JSON directory with 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.

本数据集包含本论文中使用的基准测试集与实验数据。

PDDL（Planning Domain Definition Language，规划领域定义语言）基准测试文件可于benchmarks.tar.gz与bagged-benchmarks.tar.gz压缩包中获取。前者包含仓库https://bitbucket.org/aibasel/downward-benchmarks中历届国际规划竞赛（IPC）的所有领域，剔除了多次出现在不同IPC中的重复领域；后者则包含经下述论文提出的「打包表示」重写后成功的任务子集：

Riddle, P.; Douglas, J.; Barley, M.; 及 Franco, S. 2016. 通过将PDDL重写为打包表示提升性能. 见：2016年ICAPS国际自动规划与调度会议启发式搜索与域独立规划研讨会论文集, 28–36.

我们从该论文作者处获取了打包式重写方法的实现代码。

本数据集其余文件包含所有实验的原始与处理后数据，此类实验均通过Downward-Lab生成（详见https://doi.org/10.5281/zenodo.399255）。本实验所用的运行脚本可在本论文的配套软件包中获取（详见https://doi.org/10.5281/zenodo.2621897）。

不带“-eval”后缀的目录包含原始数据，按实验分为多个子目录。每个实验子目录下均存在“runs-*”形式的子目录树，每个规划器运行结果均对应一个独立子目录。每个运行子目录中包含：指向输入PDDL文件domain.pddl与problem.pddl的符号链接（需将基准测试目录放置在正确路径后方可解析）、运行日志文件run.log（标准输出stdout）、可选的运行错误文件run.err（标准错误stderr）、用于启动实验的运行脚本run，以及包含从日志文件解析所得数据的properties文件。部分目录（非基础实验目录，基础实验的此类文件体积过大）还包含generators.py文件，其中存储了所有以置换形式表示的对称群生成元。

带“-eval”后缀的目录包含一个properties文件，该文件为包含对应实验所有运行结果合并数据的JSON目录。本质上，该properties文件是所有单个规划器运行结果对应的properties文件的并集。