lagpax
Copyright (c) 2026 Davide Sartor

This product is a derivative work. `src/lagpax/` is a translation into JAX of the
separable (`GPsep`) family of the laGP R package, together with the criteria and
subsampling routines laGP ships only for its isotropic family (`mspeGP`, `fishGP`,
`dmus2GP`) and its block Latin hypercube subsampling (`blhs`, `blhs.loop`):

    laGP: Local Approximate Gaussian Process Regression, version 1.5-9
    Copyright (c) Robert B. Gramacy <rbg@vt.edu> and Furong Sun <furongs@vt.edu>
    Licensed under the GNU Lesser General Public License
    https://cran.r-project.org/package=laGP

Because lagpax is derived from laGP's source, it is distributed under the same
license. laGP's DESCRIPTION declares `License: LGPL` with no version; per CRAN's
"Writing R Extensions", the bare form means version 2 or later. lagpax therefore
uses LGPL-2.0-or-later and ships that text in LICENSE. The "or later" clause lets
you adopt LGPL-2.1 or LGPL-3 instead if you prefer.

The methods implemented here are described in:

    Gramacy, R. B. and Apley, D. W. (2015). "Local Gaussian Process Approximation
    for Large Computer Experiments." Journal of Computational and Graphical
    Statistics, 24(2), 561-578.

    Gramacy, R. B. (2016). "laGP: Large-Scale Spatial Modeling via Local
    Approximate Gaussian Processes in R." Journal of Statistical Software, 72(1).

    Seo, S., Wallat, M., Graepel, T. and Obermayer, K. (2000). "Gaussian Process
    Regression: Active Data Selection and Test Point Rejection." IJCNN.

    Sun, F., Gramacy, R. B., Haaland, B., Lawrence, E. and Walker, A. (2019).
    "Emulating Satellite Drag from Large Simulation Experiments." SIAM/ASA Journal
    on Uncertainty Quantification, 7(2), 720-759. (Block Latin hypercube
    subsampling; `blhs.py`.)

lagpax is not affiliated with or endorsed by the laGP authors.
