GPmp-contrib documentation

gpmp-contrib extends gpmp with computer-experiment objects, model containers, Matérn container classes, sequential strategies, optimization criteria, excursion/set-inversion tools, and relaxed Gaussian-process utilities.

The provided Maternp classes fix the Matérn regularity at \(\nu=p+1/2\). Model_ConstantMean_Matern_REML selects \(\nu\) with the variance and lengthscales.

Use gpmp directly when you need models, kernels, numerical backends, or parameter-selection functions. Use gpmp-contrib when you want computer experiments, model containers, or sequential-design classes built from these objects.

Documentation contents

Installation

Package dependencies, backend notes, and documentation build commands.

Getting started

Hartmann4 run: define a problem, build a model, select parameters, predict, and inspect diagnostics.

User guide

Concepts and procedures: package organization, model containers, parameter objects, parameter selection, priors, diagnostics, sequential strategies, and reGP.

Examples

Script-oriented explanations for the main examples in examples/.

API reference

Public modules, classes, and functions.

References

Literature cited by the guide and examples.

Technical entry points

Hartmann4 example

Getting started builds the Hartmann4 example used in the core gpmp tutorial, selects Matérn covariance parameters, predicts at test points, and shows the expected diagnostic output.

Model construction

Models and computer experiments, Model state and parameter objects, and Parameter selection describe model containers, fixed or selected Matérn regularity, parameter-selection rules, stored covariance vectors, and readable Param objects.

Priors in REMAP selection

Priors for REMAP selection documents set_prior, get_prior, prior anchors, and REMAP hyperparameters.

Sequential design

Sequential design, optimization, and set estimation describes expected improvement, excursion sets, set inversion, SMC particles, and BSS particles.

API details

API reference gives signatures, shapes, return values, stored side effects, and failure conditions.