Code and data for the ICAPS 2024 paper "Efficiently Computing Transitions in Cartesian Abstractions"

Code

The file seipp-icaps2024-code.zip contains an extended version of the Fast Downward planning system (http://fast-downward.org). The code for Cartesian abstraction refinement can be found in "src/search/cartesian_abstractions/". Please see README.md and BUILD.md for instructions on how to compile the planner. Here is the short version for building the planner and running A* with a Cartesian abstraction that caches optimal abstract transitions and computes the remaining transitions on demand:

./build.py
./fast-downward.py PDDL_TASK --search "astar(cegar(subtasks=[original()], max_states=infinity, max_transitions=infinity, \
    max_time=1200, sort_transitions=true, transition_representation=sg_rh, pick_flawed_abstract_state=batch_min_h, \
    pick_split=max_cover, tiebreak_split=max_refined, memory_padding=512, random_seed=0, \
    max_concrete_states_per_abstract_state=1K, max_state_expansions=1M, store_shortest_path_tree_children=True, \
    store_shortest_path_tree_parents=True, use_max=true))"

The latest version of the code is available at https://github.com/jendrikseipp/scorpion.

Scripts

The file seipp-icaps2024-scripts.zip contains the Python code that was used for running the experiment.

Benchmarks

The file ipc-benchmarks-optimal-strips-1998-2018.zip contains the STRIPS PDDL benchmarks from sequential optimization tracks of IPC 1998-2018.

Experiment data

The remaining zipfiles contain the raw experiment data, parsed values and basic reports for the experiments in the paper. For each experiment there are two files. The first contains the raw data of all experiment runs. The code directories and benchmark files have been removed to avoid duplication and to save space. The second file (*-eval.zip) contains a "properties" file with all parsed values, HTML reports and scatter plots.