Scalable Gaussian-Process Approximations.

GPvecchia

Fast Gaussian-process inference using general Vecchia approximations

For examples of how to use the package, please see the vignettes folder. Please note that GPvecchia is under active development and not stable at this time.

Reporting problems

If you have an issue with GPvecchia, including unsatisfactory runtime, please open a Github ticket.

References

Katzfuss, M., & Guinness, J. (2017). A general framework for Vecchia approximations of Gaussian processes. arXiv:1708.06302.

Katzfuss, M., Guinness, J., Gong, W., & Zilber, D. (2018). Vecchia approximations of Gaussian-process predictions. arXiv:1805.03309.

Zilber, D., & Katzfuss, M. (2019). Vecchia-Laplace approximations of generalized Gaussian processes for big non-Gaussian spatial data. arXiv:1906.07828.

Installation

This package can be installed directly from CRAN by running

install.packages("GPvecchia")

OMP library is required for the package to run fast. Mac OS users, where OMP has not been supported for a while could consult this page

Alternatively, the package can be installed using the github source. This can be done either by using devtools

devtools::install_github("katzfuss-group/GPvecchia")

or by downloading the repository and then building the package manually:

R CMD build GPvecchia
R CMD INSTALL GPvecchia_0.1.4.tar.gz

Note that Rtools is required for compiling C/C++ with OpenMP on Windows systems. When installing Rtools, the system PATH needs to be set so that the C++ compiler included in Rtools can be found by R. Once Rtools is installed, system('g++ -v') can be used to check if the compiler is accessible from R.