ML for DFT

During my PhD, I pursued a personal project training a deep neural network and a kernel ridge regression machine learning model to predict DFT energies and thermodynamic properties. Molecules were transformed to Coulomb matrix representation, then input into the model, which predicts the electronic energy, as well as zero point energy, enthalpy, entropy, and free energy. 

For more information, you can view my GitHub repository, which includes a write-up:

https://github.com/amcisaac/ml_dft