Skip to content

lobis/geant4-python-application

Repository files navigation

Geant4 Python Application

Binder Binder

Wheels Build and Test

This is an experiment at providing a pythonic interface to Geant4.

The goal is not to provide a full python interface to Geant4, but rather to provide a high-level interface to a generic Geant4 application which is highly configurable.

If you are looking for a full set of python bindings for Geant4, I recommend looking at geant4_pybind.

Overview

  • Configurable Geant4 application written in C++
  • Python interface to the application via pybind11
  • The complete event data is available as an awkward array

Installation

The package is pip installable and should work on all major platforms. Geant4 is not required to be installed on the system as the PyPI distribution includes a statically linked version of Geant4.

However, for the time being, it's recommended to install from the GitHub repository to get the latest version, which makes Geant4 a necessary dependency.

Geant4 is available on conda-forge and can be installed with:

conda install -c conda-forge geant4=11.2.2

This project can then be installed via pip:

python -m pip install git+https://github.com/lobis/geant4-python-application

You may need additional dependencies to perform the build. I found that these are usually enuough:

conda install cmake ninja gxx_linux-64

The provided Dockerfile can be used for development purposes and as documentation on the required dependencies. To build the image run from the root directory:

docker build -t geant4-python-application .

Geant4 can be installed using conda:

conda install -c conda-forge geant4

For development purposes however, it's recommended to install Geant4 from source with a similar configuration to the one used in the CI:

git clone https://github.com/Geant4/geant4.git ./geant4-source --depth 1 --branch v11.2.2

cmake -B ./geant4-build -S ./geant4-source -DCMAKE_INSTALL_PREFIX=./geant4-install -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=17 -DGEANT4_USE_GDML=ON -DGEANT4_INSTALL_EXAMPLES=OFF -DGEANT4_INSTALL_DATA=OFF -DGEANT4_BUILD_TLS_MODEL=global-dynamic -DBUILD_STATIC_LIBS=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_CXX_FLAGS=-fPIC -DCMAKE_C_FLAGS=-fPIC -DGEANT4_USE_SYSTEM_EXPAT=OFF
cmake --build ./geant4-build --parallel $(nproc) --config Release --target install

One common installation issue is the dependency xerces-c. You may be able to install it from your package manager (e.g. `apt-get install libxerces-c-dev) but it's also possible to build it from source. In this case make sure the installation directory is accessible or just use the default system path.

git clone https://github.com/apache/xerces-c.git ./xerces-source
git -C ./xerces-source checkout tags/v3.2.5
cmake -B ./xerces-build -S ./xerces-source -DCMAKE_INSTALL_PREFIX=./xerces-install  -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=17 -DBUILD_SHARED_LIBS=OFF -DCMAKE_CXX_FLAGS=-fPIC -DCMAKE_C_FLAGS=-fPIC -Dnetwork-accessor=socket -Dtranscoder=iconv
cmake --build ./xerces-build --parallel $(nproc) --config Release --target install

After all dependencies are met, you should be able to install the package by running the following command from the root directory:

pip install .

Geant4 data files

Geant4 comes with a large set of data files which are required in order to run. The data files are the bulk of the Geant4 installation and can be quite large. These data files are not included in the Python wheels for this package due to its size.

The python package automatically manages the download of the data files. To check the location of the data files, run the following command:

python -c "import geant4_python_application; print(geant4_python_application.get_data_path())"

Uninstalling the package will not remove the data files. The user is responsible for removing them manually.

Overriding the default data path

The default data path can be overridden by calling application_directory. This should be done before the application is initialized.

from geant4_python_application import application_directory

application_directory("/some/other/path")

Overriding the default data directory is encouraged when submitting batch jobs to a cluster in order to avoid downloading the data files multiple times. In this case the application directory should point to some shared location.

Using a temporary directory

A temporary directory can be used by calling:

from geant4_python_application import application_directory

application_directory(temp=True)

The operating system should take care of cleaning up the temporary directory.

It is possible that the operating system will delete only some of the files in the temporary directory which can lead to a Geant4 runtime error. In this case it's recommended to delete the temporary directory manually and let the application recreate it again.

Usage

import geant4_python_application as g4

# Use a temporary directory for the Geant4 data files (remove this line to use the default location)
g4.application_directory(temp=True)

with g4.Application(gdml=g4.basic_gdml, seed=137) as app:
    events = app.run(n_events=100)

print(events)