Local Approximate Gaussian Process Regression (laGP R package)
laGP is an R package providing approximate GP regression routins for large computer experiments and spatial datasets. The approximation is based on finding small local designs for prediction (independently) at particular inputs. The current version supports
- ALC, MSPE and NN-based local approximation, as well as EFI-based global heuristics
- local MLE/MAP inference for (isotropic) lengthscale and nugget parameters
- OpenMP for approximation over a vast out-of-sample testing set; requires special compilation
- GPU acceleration for local ALC subroutine evaluations; requires special compilation
- SNOW/parallel-package cluster parallelization
- functions to support computer model calibration via optimization, e.g., using NOMAD
- functions automating the optimization of blackbox functions under (blackbox) constraints
- an interface to lower-level (full) GP inference and prediction
- Obtain R from cran.r-project.org by selecting the version for your operating system.
- Install the laGP package, from within R.
- Optionally, install the mvtnorm, and snow packages, which
are helpful for some of the comparisons in the examples and demos.
> install.packages(c("mvtnorm", "snow"))
- Load the library as you would for any R library.
- The laGP tutorial is
implemented as a package vignette, authored in Sweave. The pdf can be obtained from within R with the following code.
> vignette("laGP")To obtain the source code contained in the vignette, use the Stangle command.
> Stangle(vignette("laGP")$file)The code from Section 4 of the vignette, on Calibration, is available as a standalone demo.
> demo("calib", package="laGP")
- See the package documentation. A pdf version of the
reference manual, or help pages, as also available.
The help pages can be accessed from within
R. The best way to acquaint yourself with the functionality
of this package is to run the demos which illustrate the examples
contained in the papers referenced below. Try starting with...
> ?laGP # follow the examples
> ?aGP # follow the examples - this is the main workhorse
- Local Gaussian process approximation for large computer experiments (2014) with Dan Apley; to appear in Journal of Computational and Graphical Statistics; preprint on arXiv:1303.0383
- Massively parallel approximate Gaussian process regression (2014) with Jarad Niemi and Robin Weiss; to appear in Journal of Uncertainty Quantification; preprint on arXiv:1310.5182
- Speeding up neighborhood search in local Gaussian process prediction (2014) with Ben Haaland; preprint on arXiv:1409.0074
- Calibrating a large computer experiment simulating radiative shock hydrodynamics (2014) with Derek Bingham, James Paul Holloway, Michael J. Grosskopf, Carolyn C. Kuranz, Erica Rutter, Matt Trantham, Paul R. Drake; preprint on arXiv:1410.3293
Please send questions and comments to rbgramacy_AT (_chicagobooth_DOT_edu). Enjoy!
Robert B. Gramacy -- 2013