The major purpose of this release is to fix an error on CRAN checks for Oracle Developer Studio. We have also parallelized the computation of the inverse cholesky factor, and included safeguards against large smoothness parameters in matern_isotropic.

Many thanks for Youssef Fahmy for finally cracking the solaris error.

Added matern_anisotropic3D_alt covariance Fixed some problems with #includes in src files

Now uses OpenMP for parallel computations of the likelihood.

Updated behavior of Fisher scoring algorithm when information matrix is ill-conditioned (simple regularization of info matrix).

Fixed bug in “fit_model” when missing values are present. Updated behavior of Fisher scoring algorithm when information matrix ill-conditioned Allow user to fix a subset of parameters in “fit_model” Allow user to specify maximum number of iterations in “fit_model” New faster computational algorithm for predictions Several new covariance functions, including

matern15_isotropic matern25_isotropic matern35_isotropic matern45_isotropic matern15_scaledim

matern25_scaledim

matern35_scaledim

matern45_scaledim

Bug fix for overloaded use of ‘pow’ function in ‘basis.h’

This update includes an implementation of the Fisher Scoring algorithm described in this paper https://arxiv.org/abs/1905.08374, computed in a single pass through the data.

Much of the C++ code has been rewritten and reorganized, making use of the Armadillo C++ linear algebra library, with the help of RcppArmadillo.

There are also several new covariance functions. The complete list of covariance functions is now:

matern_isotropic exponential_isotropic matern_spacetime exponential_spacetime matern_scaledim exponential_scaledim matern_anisotropic2D exponential_anisotropic2D exponential_anisotropic3D matern_nonstat_var exponential_nonstat_var matern_sphere exponential_sphere matern_spheretime exponential_spheretime matern_sphere_warp exponential_sphere_warp matern_spheretime_warp exponential_spheretime_warp

This is a minor release fixing numerical stability problems that arise during optimization of the likelihood.

Added a check in each of the proflik_mean* functions to avoid inverting the information matrix when it is numerically singular.

Changed the default number of Nelder-Mead iterations in fit_model to 100