Python package for highly flexible function-valued Gaussian processes (fvGP)
It is recommended to use this package via gpCAM.
Specialties: Extreme-Scale GPs, GPs Tailored for HPC training, Advanced Kernel Designs, Domain-Aware Stochastic Function Approximation Coming soon: All those advancements for stochastic manifold learning
fvGP holds the world record for exact large-scale Gaussian Processes!
This package uses the HGDL package of David Perryman and Marcus Noack, which is based on the HGDN algorithm by Noack and Funke.