Mimir is a C++20 planning library with Python bindings for grounded and lifted classical planning.

Focus on the intersection of planning and learning - search algorithms, parallel data processing, and knowledge representation languages.

• Supports a rich fragment of PDDL in both the grounded and lifted planning setup - strips, typing, negative-preconditions, disjunctive-preconditions, equality, existential-preconditions, universal-preconditions, conditional-effects, derived-predicates, action-costs.
• Fully written in C++ - no intermediate files.
• Provides Python bindings that even allow you to write heuristics for AStar search directly in Python.
• Provides a generic graph library that supports forward and backward traversal of edges with adapters to use the powerful Boost BGL library.
• Uses the zero-copy serialization library Cista for efficient parallel data generation.

Mimir is available on pypi.

pip install pymimir

import pymimir as mm

parser = mm.PDDLParser("domain.pddl", "problem.pddl")
aag = mm.LiftedApplicableActionGenerator(parser.get_problem(), parser.get_pddl_factories())
brfs = mm.BrFSAlgorithm(aag)
status, plan = brfs.find_solution()

Mimir depends on the following set of libraries:

• Loki for parsing PDDL files,
• Boost header-only libraries (Fusion, Spirit x3, Container, BGL),
• Cista for zero-copy serialization,
• Nauty and Traces for graph isomorphism testing,
• GoogleBenchmark for automated performance benchmarking, and
• GoogleTest for unit testing.

Run the following sequence of commands to download, configure, build, and install all dependencies:

1. Configure the dependencies CMake project with the desired installation path:

cmake -S dependencies -B dependencies/build -DCMAKE_INSTALL_PREFIX=dependencies/installs

2. Download, build, and install all dependencies:

cmake --build dependencies/build -j$(nproc)

Run the following sequence of commands to configure, build, and install Mimir:

1. Configure Mimir in the build directory build/ with the CMakePrefixPath pointing to the installation directory of the dependencies:

cmake -S . -B build -DCMAKE_PREFIX_PATH=${PWD}/dependencies/installs

2. Build Mimir in the build directory:

cmake --build build -j$(nproc)

3. (Optional) Install Mimir from the build directory to the desired installation prefix directory:

cmake --install build --prefix=<path/to/installation-directory>

#include  <mimir/mimir.hpp>

const auto parser = PDDLParser("domain.pddl", "problem.pddl")
const auto aag = std::make_shared<LiftedApplicableActionGenerator>(parser.get_problem(), parser.get_pddl_factories())
const auto brfs = BrFSAlgorithm(aag)
const auto [status, plan] = brfs.find_solution()

We developed Mimir in Visual Studio Code. We recommend installing the C/C++ and CMake Tools extensions by Microsoft. To get maximum IDE support, you should set the following Cmake: Configure Args in the CMake Tools extension settings under Workspace:

• -DCMAKE_PREFIX_PATH=${workspaceFolder}/dependencies/installs

After running CMake: Configure in Visual Studio Code (ctrl + shift + p), you should see all include paths being correctly resolved.

Alternatively, you can create the file .vscode/settings.json with the content:

{
"cmake.configureArgs": [ "-DCMAKE_PREFIX_PATH=${workspaceFolder}/dependencies/installs" ]
}