Skip to content

Fast Downward 21.12#

Fast Downward 21.12 was released on February 16, 2022.

Highlights:

  • Fast Downward now has a logo!

  • We added new methods for generating patterns and pattern collections based on counterexample-guided abstraction refinement and a new highly random method for generating individual patterns based on the causal graph. These methods are due to Rovner et al. (ICAPS 2019).

  • The operator-counting heuristic now has an option to use integer operator counts rather than real-valued operator counts. This makes the heuristic more accurate at a vastly increased computational cost (not generally recommended, but very useful for targeted experiments). We added a new constraint generator for Imai and Fukunaga\'s delete relaxation constraints (JAIR 2015). With the right option settings, the operator-counting heuristic with this new constraint generator results in the optimal delete relaxation heuristic h^+.

  • Pruning methods now have a different interface. The mechanism to disable pruning automatically after a number of expansions that resulted in little pruning is now implemented as its own pruning method that wraps another pruning method. Be careful that the old syntax is still accepted by the planner, but the options that limit pruning are ignored. (This is due to an option parser bug; a fix is in the works.)

  • In our ongoing efforts to improve the landmark code, the landmark factories and landmark-count heuristic received a major overhaul. We removed irrelevant options for landmark factories, decoupled the computation of reasonable orders from landmark generation, made many internal code and data structure changes to make the code nicer to work with and fixed several long-standing bugs.

  • All pattern generators and pattern collection generators now have controllable verbosity. Similar changes to other components of the planner are planned. This is part of a general effort to make logging more configurable.

  • For developers: the internal representation of states has been overhauled, resolving the confusion between the previous classes GlobalState and State.

The Apptainer and Docker distributions of the planner include LP support using the SoPlex solver. Thank you to ZIB for their solver and for giving permission to include it in the release. SoPlex is licensed under the ZIB Academic License.

Downloads#

For Apptainer, Docker and Vagrant, see getting started for instructions on how to run the planner. That page explains how to run the latest release of Fast Downward, so you have to amend the instructions if you want to run an older release.

For the tarball, if you have satisfied all dependencies, the following should build and run the planner with configuration alias lama-first on PDDL files /path/to/domain.pddl and /path/to/problem.pddl on a Linux or macOS system:

tar -xvzf fast-downward-21.12.tar.gz
cd fast-downward-21.12
./build.py
./fast-downward.py --alias lama-first /path/to/domain.pddl /path/to/problem.pddl

For dependencies and more advanced information (including Windows builds), see the build instructions.

Limitations#

For licensing reasons, the binary distributions of Fast Downward have limited support for commercial LP solvers (CPLEX and Gurobi). The Docker and Apptainer distributions only support SoPlex. The Vagrant version supports SoPlex and/or CPLEX if they are available at provisioning time. See getting started for more information.

Referencing Fast Downward 21.12#

If you use this version of Fast Downward in a scientific publication, we encourage you to mention the version number, as in "We ran experiments with Fast Downward 21.12." The canonical reference for citing Fast Downward is the following journal paper:

@Article{helmert-jair2006,
  author =  "Malte Helmert",
  title =   "The {Fast} {Downward} Planning System",
  journal = "Journal of Artificial Intelligence Research",
  year =    "2006",
  volume =  "26",
  pages =   "191--246"
}

Changes in Fast Downward 21.12#

  • new Fast Downward logo

    https://issues.fast-downward.org/issue1024

    You can see the logo in the README. Check out our alternative suggestions for the logo designs, including the ever so popular "truck falling down the hill" logo.

  • fast-downward.py main script: The script now automatically finds domain files <taskfile>-domain.<ext> for task files called <taskfile>.<ext>.

    https://issues.fast-downward.org/issue1033

  • pdbs: Integrate the Rovner et al. pattern generation methods based on CEGAR.

    https://issues.fast-downward.org/issue1007

  • pdbs: Integrate the Rovner et al. random pattern generation methods.

    https://issues.fast-downward.org/issue1008

  • pdbs: All pattern (collection) generators now have an option verbosity to set the desired level of output.

    https://issues.fast-downward.org/issue1043

    Internally, generators now use their own logger rather than g_log.

  • pdbs, for developers: Replace size_t by int for abstract state hashes in PDB-related code.

    https://issues.fast-downward.org/issue1018

  • LP/IP: Support integer variables in linear programs.

    https://issues.fast-downward.org/issue891

    https://issues.fast-downward.org/issue1048 

    You can now use the LP solver interface to solve mixed integer programs. In particular, the operator-counting heuristics now have an option use_integer_operator_counts that improves the heuristic value by forcing operator counts to take integer values. Solving a MIP is NP-hard and usually takes much longer than solving the corresponding LP.

  • LP/IP: Delete-relation constraints can now be used in the operator-counting framework. The constraints defined by Imai and Fukunaga (JAIR 2015) encode different relaxations of the delete-relaxation heuristic. For details, see Delete relaxation constraints.

    https://issues.fast-downward.org/issue983

  • LP/IP: Fix a bug which induced inadmissible heuristic values when solving MIPs. This only occurred for operator-counting heuristics with integer variables and very large heuristic values (at least 10,000).

    https://issues.fast-downward.org/issue983

  • LP/IP, for developers: Add debugging methods to LP solver interface.

    https://issues.fast-downward.org/issue960

    You can now assign names to LP variables and constraints for easier debugging. Since this incurs a slight runtime penalty, we recommend against using this feature when running experiments.

  • LP/IP, for developers: Debug builds with LP solvers vs. the _GLIBCXX_DEBUG flag.

    https://issues.fast-downward.org/issue982

    Previously, we used the flag _GLIBCXX_DEBUG in debug builds for additional checks. This makes the binary incompatible with external libraries such as LP solvers. The flag is now disabled by default. If no LP solvers are present or LP solvers are disabled, it can be enabled by setting the CMake option USE_GLIBCXX_DEBUG. The build configurations debugnolp and releasenolp have been removed, and the build configuration glibcxx_debug has been added.

  • pruning: New LimitedPruning class replaces previous limitation options of individual pruning methods.

    https://issues.fast-downward.org/issue1042

    For example, the old command line 

    --search "astar(lmcut(),pruning=atom_centric_stubborn_sets(min_required_pruning_ratio=0.2,expansions_before_checking_pruning_ratio=1000))
    is now expressed as 
    --search "astar(lmcut(),pruning=limited_pruning(pruning=atom_centric_stubborn_sets(),min_required_pruning_ratio=0.2,expansions_before_checking_pruning_ratio=1000))

  • landmarks: Replace landmark factory option reasonable_orders by the new landmark factory lm_reasonable_orders_hps.

    https://issues.fast-downward.org/issue995

    For example, the old command line 

    --evaluator hlm=lmcount(lm_factory=lm_rhw(reasonable_orders=true))
    is now expressed as 
    --evaluator hlm=lmcount(lm_factory=lm_reasonable_orders_hps(lm_rhw()))

  • landmarks: Replace landmark factory option no_orders by the new option use_orders with the opposite meaning.

    https://issues.fast-downward.org/issue995

  • landmarks: Removed landmark factory options that have no effect.

    https://issues.fast-downward.org/issue995

    Removed options: * lm_exhaust: disjunctive_landmarks, conjunctive_landmarks * lm_hm: disjunctive_landmarks, only_causal_landmarks * lm_merged: disjunctive_landmarks, conjunctive_landmarks, only_causal_landmarks * lm_rhw: conjunctive_landmarks, reasonable_orders * lm_zg: disjunctive_landmarks, conjunctive_landmarks, only_causal_landmarks

  • landmarks: Fix a bug where lm_rhw would compute wrong greedy-necessary orderings in certain cases.

    https://issues.fast-downward.org/issue1004

  • landmarks: Fix a bug where the lm_rhw, lm_zg and lm_exhaust landmark factories used an overly optimistic approximation of relaxed reachability for planning tasks with conditional effects. This change can lead to more generated landmarks and landmark orderings in such tasks.

    https://issues.fast-downward.org/issue1041

  • landmarks: Fix a bug where the Zhu/Givan landmark factory caused a crash on relaxed unsolvable tasks due to an empty landmark graph.

    https://issues.fast-downward.org/issue998

  • landmarks: Fix a bug where cycles of natural orderings resulted in crashes in the landmark factories. This could only happen in unsolvable planning tasks.

    https://issues.fast-downward.org/issue937

  • landmarks: Fix a failing assertion in the lm_rhw landmark factory triggered by certain unsolvable tasks.

    https://issues.fast-downward.org/issue467

  • landmarks, for developers: Clean up the code of LandmarkGraph. Some of the public methods were renamed. This class will undergo further changes in the future.

    https://issues.fast-downward.org/issue989

  • landmarks, for developers: Separate the functionality of landmarks from the functionality of landmark nodes by introducing a new Landmark class.

    https://issues.fast-downward.org/issue999

  • landmarks, for developers: Move functionality used during search away from LandmarkGraph, making the landmark graph constant after creation.

    https://issues.fast-downward.org/issue988

    https://issues.fast-downward.org/issue1000

  • landmarks, for developers: Introduce new class LandmarkFactoryRelaxation for landmark factories based on delete relaxation. Move usage of exploration object to subclasses of the landmark factory base class.

    https://issues.fast-downward.org/issue990

  • build: The build system now prefers compilers cc/c++ found on the path over gcc/g++. As before, environment variables CC/CXX can be used to override this choice.

    https://issues.fast-downward.org/issue1031

  • build: Only build configurations defined in build_configs.py are loaded in the build.py script.

    https://issues.fast-downward.org/issue1016

  • for developers: Add option to use a local (configurable) logger instead of the global one.

    https://issues.fast-downward.org/issue964

    Classes which want to configure the logger (currently only the verbosity level can be configured) should now use the functions add_log_options_to_parser and get_log_from_options to obtain their local log object.

  • for developers: Unify the State and GlobalState classes.

    https://issues.fast-downward.org/issue348

    We unified the classes GlobalState and State into a new class also called State. This removed a lot of code duplication and hacks from the code. A description of the new class can be found in the blog.

  • for developers: Change public interface of generation of random ints and doubles in the RandomNumberGenerator class.

    https://issues.fast-downward.org/issue1026

  • for developers: Use RandomNumberGenerator class in VariableOrderFinder.

    https://issues.fast-downward.org/issue1032

  • infrastructure: Add support for GitHub actions.

    https://issues.fast-downward.org/issue940

  • infrastructure: Add CPLEX support to our GitHub actions for Windows.

    https://issues.fast-downward.org/issue1005

  • infrastructure: Decide on rules regarding software support and improve GitHub actions accordingly.

    https://issues.fast-downward.org/issue1003