stilt is an open-source R library for Separable Gaussian Process Interpolation, a type of statistical emulation.

The package, which is available on CRAN, comprises a set of R functions to build and use an interpolator (emulator) for time series or 1D regularly spaced data in multidimensional space. The standard usage is for interpolating time-resolved computer model output between model parameter settings. It can also be used for interpolating multivariate data (e.g., oceanographic time-series data, etc.) in space. There are functions to test the emulator using cross-validation, and to produce contour plots over 2D slices in model input parameter (or physical) space.

For more information or to download the source material, please visit the stilt page on CRAN: or visit the Stilt page on GitHub.


R. Olson and W. Chang (2013): Mathematical framework for a separable Gaussian Process Emulator, Technical report.


stilt is licensed under the GNU General Public License, version 3 or later.

This work was supported by the National Science Foundation through the Network for Sustainable Climate Risk Management (SCRiM) under NSF cooperative agreement GEO-1240507.