Data and analysis setup for "Machine Learning of Atomic Dynamics and Statistical Surface Identities in Gold Nanoparticles"

All of the required package will be installed with:

python3 -m venv NPenv
source ./NPenv/bin/activate
pip install -U pip
pip install -r requirements.txt
pip install --upgrade joblib==1.1.0

the last line, with the downgrade of joblib, is due to a bug in the requirements of hdbscan 0.8.28

NB

The file in this repository are tailored to reproduce precisely the results obtained in our simulations (starting from the data that are present in the zenodo link and/or in the release on github), so if you want to apply the analysis to your data, you will need to modify the scripts.

Simulations

If you have a recent version of lammps (with the smatb/single pair active) simply launch the script in simulations And then preprocess the trajectories with bash ./run00_preprocessTrajectories.sh

OR

You may download the precompressed trajectory from the release page on github or run the following commands:

wget https://github.com/GMPavanLab/dynNP/releases/download/V1.0-trajectories/dh348_3_2_3.hdf5
wget https://github.com/GMPavanLab/dynNP/releases/download/V1.0-trajectories/dh348_3_2_3_fitted.hdf5
wget https://github.com/GMPavanLab/dynNP/releases/download/V1.0-trajectories/ico309.hdf5
wget https://github.com/GMPavanLab/dynNP/releases/download/V1.0-trajectories/ico309_fitted.hdf5
wget https://github.com/GMPavanLab/dynNP/releases/download/V1.0-trajectories/to309_9_4.hdf5
wget https://github.com/GMPavanLab/dynNP/releases/download/V1.0-trajectories/to309_9_4_fitted.hdf5

SOAP calculations

• Prepare the SOAP trajectories with bash ./run01_SOAP.sh

Bottom-up analysis

• Prepare and run the BU analysis with bash ./run02_BU.sh

Top-Down analysis

• Prepare and run the TD analysis with bash ./run02_TD.sh

Visualization

• Prepare the visualizations and create the figures with with bash ./run03_prepareFigs.sh

This scripts contains a small workaround that we encountered in working in an virtual environment with both ovito and matplotlib.

Remarks

The Bottom-up and the Top-Down can be executed in any order or in parallel