pod.build_kriging_gpr

build_kriging_gpr(kernel, n_restarts=KRIGING_RESTARTS_FULL)

Builds an unfitted Kriging model with the production settings.

Standardised inputs [UQLab], standardised response (normalize_y=True) [digiqual: scikit-learn has no ordinary-Kriging trend, so the prior mean is the sample mean rather than a GLS-estimated constant as in [M26] Eq. 6], hyperparameters by maximum likelihood [M26] Sec. 2.1.2, and multi-start L-BFGS-B [digiqual].