Three main libraries are required to run the code : Dscribe (SOAP descriptors), Scikit-learn (Machine Learning tools) and ASE (Atomic Simulation Environment).
More precisely :
More precisely :
- python==3.10
- ase==3.22.1
- scikit-learn==1.4.1
...
...
@@ -22,9 +22,9 @@ pip install -r requirements.txt
The code builts Energy Adsorption Maps (EAMs) using machine learning.
The file "run.py" (i) trains a model from a training set containing adsorption energies calculated on selected sites (".traj" file), and (ii) predicts adsorption energies of new sites located on a n $\times$ n regular grid. Results are written in the file "result.txt".
The file "run.py" (i) trains a model from a training set containing adsorption energies calculated on selected sites (".traj" file), and (ii) predicts adsorption energies of new sites located on a $n$ $\times$ $n$ regular grid. Results are written in the file "result.txt".
The ".traj" file contains data useful to describe the simulation box, the atomic positions, the grid (n) as well as parameters for the SOAP descriptors and the Gaussian Process Regression.
The ".traj" file contains data useful to describe the simulation box, the atomic positions, the grid ($n$) as well as parameters for the SOAP descriptors and the Gaussian Process Regression.
The code to create the Adsorption Enery Map ("Ag111_AEM.pdf", Working Example) is also given and uses the matplotlib package. Code execution must take place in the same directory as the file "result.txt" previously obtained.