/VecGP

Gaussian Process Regression designed for machine learning predictions of force vectors in chemical environments

Primary LanguagePython

VecGP

Theano implementation of Gaussian Process Regression designed for machine learning predictions of force vectors in chemical environments.

gpr.py implements a general GP regression scheme in theano

kernels.py implements many kernel functions

For more information, see the documentation in each of these files and the references below:

[1] C.E. Rasmussen and C.K.I Williams. MIT Press (2006)

[2] A.P. Bartok, R. Kondor and G. Csanyi. Phys. Rev. B. 87 (2013)

[3] A. Glielmo, P. Sollich, and A. De Vita. Phys. Rev. B. 95 (2017)

[4] G. Ferre, J.B. Maillet, and G. Stoltz. J. Chem. Phys. 143 (2015)