Back to [[Releases]].
= 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 Singularity and Docker distributions of the planner include LP support using the [[https://soplex.zib.de/|SoPlex]] solver. Thank you to ZIB for their solver and for giving permission to include it in the release. !SoPlex is licensed under the [[https://scipopt.org/academic.txt|ZIB Academic License]].
== Downloads ==
* source tarball: [[attachment:fast-downward-21.12.tar.gz||&do=get]]
* Singularity container: [[attachment:fast-downward.sif||&do=get]]
* Docker container: [[https://hub.docker.com/r/aibasel/downward|aibasel/downward:21.12 on Docker Hub]]
* Vagrantfile: [[attachment:Vagrantfile||&do=get]] for Fast Downward 21.12
For Singularity, Docker and Vagrant, see QuickStart 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:
{{{#!bash
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 [[https://github.com/aibasel/downward/blob/main/BUILD.md|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 Singularity distributions only support !SoPlex. The Vagrant version supports !SoPlex and/or CPLEX if they are available at provisioning time. See QuickStart 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 file on https://github.com/aibasel/downward. Check out [[ForDevelopers/Blog/LogoDesigns]] for alternative suggestions including the ever so
popular "truck falling down the hill" logo.
* `fast-downward.py` main script: The script now automatically finds domain files `-domain.` for task files called `.`. <
>
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 [[Doc/ConstraintGenerator#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 wiki:
[[ForDevelopers/Blog/ADeeperLookAtStates]]
* 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