This is the data needed to replicate the simulations discussed in the paper "Modeling the Ga/As binary system across temperatures and compositions from first principles" by G. Imbalzano and M. Ceriotti.

The data record contains:

- README.txt : this README file
- gaas-dataset.chemiscope.json.gz : a data file with an interactive visualization of the training data, that can be loaded on http://chemiscope.org
- gaas-data.zip : an archive file, containing 4 folders:
    - chemiscope: contains the chemiscope standalone viewer necessary to visualize the dataset. It does not need any additional program and can be run anywhere
    - database: contains the database of configurations used for this work. The structures are divided into TRAINING.data (the training + validation set) and ALPHA.data (test set, used to evaluate alpha as shown in the paper)
    - MD_example: contains 2 examples of simulations that have been run to produce the data shown in the paper. To run them, one needs to install i-PI, LAMMPS, and N2P2
    - potentials: contains the trained Behler-Parrinello neural network potentials used to run the simulations. They are compatible with N2P2 and RuNNer.