Scikit-Matter - ORE Paper Data and Figures

Jupyter notebooks used to produces results of the paper

Goscinski, A, VP Principe, G Fraux, S Kliavinek, BA Helfrecht, P Loche, M Ceriotti, and RK Cersonsky. 2023. “Scikit-Matter : A Suite of Generalisable Machine Learning Methods Born Out of Chemistry and Materials Science” Open Research Europe 3 (81). https://doi.org/10.12688/openreseurope.15789.2.

Installation

To run the notebooks use Python 3.8.10 and install the required packages using

pip install -r requirements.txt

or directly use the conda environment

conda env create -f environment.yml

In the conda evironment also the dependencies are fixed.

Figures

You can find the notebooks in paper/

• Figure 1: manually created
• Figure 2: atomrings.ipynb and NeighborFig.ipynb
• Figure 3: manually created
• Figure 4: PCovR Charge Model.ipynb
• Figure 5: PCovR Charge Model.ipynb
• Figure 6: WhoDataset-PCovR.ipynb
• Figure 7: Ice Selection.ipynb
• Figure 8: WhoDataset-Selection.ipynb
• Figure 9: convex_hull_toy_figure.ipynb
• Figure 10: ice_convex_hull.ipynb

License

The code is licensed undere BSD-3. For the ice dataset please refer to https://archive.materialscloud.org/record/2018.0010/v1