MaMMoS documentation#

Framework#

The MaMMoS framework provides software components for magnetic multiscale modeling. The following table provides a short overview and contains links to example and API reference for the individual components. The binder badges allow running the examples for the individual packages interactively in the cloud.

Package repository	Examples	API
mammos	mammos	–
mammos-analysis	mammos-analysis	mammos_analysis
mammos-dft	mammos-dft	mammos_dft
mammos-entity	mammos-entity	mammos_entity
mammos-mumag	mammos-mumag	mammos_mumag
mammos-spindynamics	mammos-spindynamics	mammos_spindynamics
mammos-units	mammos-units	mammos_units

Additional tools#

An overview of other tools created through or supported by the MaMMoS project is available at https://mammos-project.github.io/#additional-tools.

Framework installation#

The MaMMoS framework consists of a collection of packages (see design for more details). The metapackage mammos can be used to install a consistent set of these packages.

The package mammos-mumag depends on jax. To get jax with GPU support you will need to manually install jax with the required optional dependencies matching your GPU hardware/software, e.g. for an NVIDIA GPU you may need to install jax[cuda12]. For details please refer to the jax installation instructions.

pixi

Requirements: pixi (https://pixi.sh/)

Pixi will install Python and mammos.

To conveniently work with the notebook tutorials we install jupyterlab. (packaging needs to be pinned due to a limitation of pixi/PyPI.):

Some examples also require esys-escript. On linux we can install it from conda-forge. On Mac or Windows refer to the esys-escript installation instructions:

Linux:

pixi init
pixi add python jupyterlab "packaging<25" "pandas<2.3" esys-escript
pixi add mammos --pypi
pixi add --pypi "jax[cuda12]"  # assuming an NVIDIA GPU with CUDA 12, see comment above

Mac/Windows:

pixi init
pixi add python jupyterlab "packaging<25" "pandas<2.3"
pixi add mammos --pypi
pixi add --pypi "jax[cuda12]"  # assuming an NVIDIA GPU with CUDA 12, see comment above

Finally start a shell where the installed packages are available:

pixi shell

conda

Requirements: conda (https://conda-forge.org/download/)

Use conda in combination with pip to get packages from conda-forge and PyPI.

To conveniently work with the notebook tutorials we install jupyterlab. (packaging needs to be pinned due to a dependency issue in mammos-entity.)

Some examples also require esys-escript. On linux we can install it from conda-forge. On Mac or Windows refer to the esys-escript installation instructions.

conda create -n mammos-environment python pip jupyterlab "packaging<25" "pandas<2.3" esys-escript
conda activate mammos-environment
pip install mammos
pip install "jax[cuda12]"  # assuming an NVIDIA GPU with CUDA 12, see comment above

pip

Requirements: python>=3.11 and pip

When using pip we recommend creating a virtual environment to isolate the MaMMoS installation.

First, create a new virtual environment. Here, we choose the name mammos-venv.

python3 -m venv mammos-venv

To activate it, run

on MacOS/Linux
```
. mammos-venv/bin/activate
```
on Windows
```
mammos-venv/bin/activate.sh
```

Finally install mammos from PyPI:

pip install mammos
pip install "jax[cuda12]"  # assuming an NVIDIA GPU with CUDA 12, see comment above

Some examples also require esys-escript, which must be installed separately. Please refer to the documentation of esys-escript for installation instructions.

Framework example notebooks#

To conveniently download all example notebooks use the mammos-fetch-examples script, which is installed as part of the mammos package (further details).

Acknowledgements#

This software has been supported by the European Union’s Horizon Europe research and innovation programme under grant agreement No 101135546 MaMMoS.