The Models Module

The ciderpress.models module is the workhorse for training Cider exchange-correlation functionals and mapping them to efficient models for evaluation in DFT codes. The train module contains the MOLGP and MOLGP2 classes, which are used to construct Gaussian process models for the exchange and correlation energies. While these Gaussian process classes are custom to CiderPress, the kernel functions used to construct the covariance matrix are built on top of those in the scikit-learn package.

Contents:

• ciderpress.models.train
• ciderpress.models.dft_kernel
• ciderpress.models.kernels
• Kernel Tools