A computationally stable approach of fitting a Gaussian Process (GP) model to a deterministic simulator.
News
GPfit v1.0-8 (Release date: 2019-02-07)
Changes:
Removed reverse dependency from test suite.
Reverted export of plot.GP().
GPfit v1.0-6 (Release date: 2019-01-15)
Changes:
Refactored GP_fit() for improved calculation speed.
Refactored predict.GP() for improved prediction speed.
Added scale_norm() as helper to rescale dimensions into 0,1 range.
Added fitted.GP() method.
Added testthat unit tests.
GPfit v1.0-0 (Release date: 2015-04-01)
Changes:
Housekeeping changes only: added inst/CITATION file and NEWS file, including retroactive listing of changes
Version number coincides with publication of JSS paper (see references).
GPfit v0.2-1 (Release date: 2014-09-03)
Changes:
fixed bug in predict() for the Matern correlation function
GPfit v0.2-0 (Release date: 2014-06-24)
Changes:
Matern correlation function added to corr_matrix()
In corr_matrix(), Gaussian correlation generalized to power exponential.
Power exponential correlation with power=1.95 is now default for corr_matrix() and thus for GP_fit.
Iterative approach to selecting the nugget implemented in predict.GP().
In GP_fit(), the argument optim_start was added, to allow user-specified initial parameter values for optim(), in addition to those generated automatically.