API Reference

gpCAM comes with two main classes: the gpOptimizer, for single-task Gaussian processes and Bayesian optimization, and the fvgpOptimizer for the multi-task equivalent. Both are capable of dealing with non-Euclidean input spaces and world-record-holding scaling of exact GPs.

These optimizer classes are a thin Bayesian-optimization layer on top of the fvGP engine — most of the GP machinery (kernels, training, posterior evaluation) is inherited from fvgp.GP, and acquisition-function optimization can run on supercomputers via HGDL.

To get to know gpCAM, check out the examples, download the repository and look in ./tests, or visit the project website.

New in fvgp 4.8 (gpCAM 8.4)

Because the optimizers inherit from fvgp.GP, the following fvgp 4.8 additions are available directly on any GPOptimizer/fvGPOptimizer instance and are listed on the gpOptimizer / fvgpOptimizer pages:

• Model-validation metrics — mae, mape, msll, interval_score, mpiw, and coverage_curve for quantifying predictive accuracy and calibration, plus plot_observed_vs_predicted for a quick diagnostic plot.

• New kernels — bump and sle_kernel (in addition to the existing library), all re-exported through gpcam.kernels and documented on the Kernels page.

• Linear-algebra modes — the linalg_mode argument now accepts CholInv/Inv (the replacement for the removed calc_inv option) and preconditioned sparse solvers such as sparseCGpre/sparseMINRESpre for large gp2Scale problems.

See Also

• Repository

• The fvGP package

• The HGDL package

Have suggestions for the API or found a bug?

Please submit an issue on GitHub.