Code from the IJCAI 2019 paper "Pattern Selection for Optimal Classical Planning with Saturated Cost Partitioning"

The zipfile contains an extended version of the Fast Downward planning system (http://fast-downward.org). The code for Sys-SCP and saturated cost partitioning can be found in the "src/search/pdbs" and "src/search/cost_saturation" directories.

Please see http://www.fast-downward.org for instructions on how to compile the planner.

The following configuration computes diverse SCP heuristics over Sys-SCP PDBs:

--search "astar(saturated_cost_partitioning([projections(systematic_scp(max_pattern_size=infinity, max_pdb_size=2M, max_collection_size=20M, max_time=100, max_time_per_restart=10, store_orders=True, store_dead_ends=True, order=cg_down, random_seed=0))], max_orders=infinity, max_optimization_time=0, max_time=200, diversify=True, random_seed=0))"

The Scorpion variant with the highest total coverage uses Cartesian abstraction heuristics and Sys-SCP PDBs, prunes irrelevant operators in a preprocessing step and uses partial order reduction during the A* search:

--search "astar(saturated_cost_partitioning([projections(systematic_scp(max_pattern_size=infinity, max_pdb_size=2M, max_collection_size=20M, max_time=100, max_time_per_restart=10, store_orders=true, store_dead_ends=true, order=cg_down, max_patterns=infinity, random_seed=0)),cartesian()], max_orders=infinity, max_optimization_time=0, max_time=200, diversify=true, random_seed=0), pruning=stubborn_sets_simple(min_required_pruning_ratio=0.2))"

Note that for operator pruning you need to make the h² preprocessor (https://bitbucket.org/atorralba/h2-fd-preprocessor) available on the PATH (e.g., using the name "h2-mutexes") and then pass "--transform-task h2-mutexes" to the fast-downward.py script (in Lab you can use the driver_options argument of add_algorithm for this).

We actively maintain the code at https://github.com/jendrikseipp/scorpion.