fvGP is an API for flexible HPC single and multi-task Gaussian processes over Euclidean and non-Euclidean (strings, molecules, materials) spaces.

The GP class is the core of most the functionality in fvGP and therefore in gpCAM.

fvGP can use HGDL optimization on supercomputers for advanced GP use cases.

The fvGP class is a multi-output GP framework which inherits most of its functionality from the GP base class.

The fvGP package holds the world-record for exact scalable GPs. It was run on 5 million data points in 2023.