Robust diag [draft], EGP [draft], geometric aniso

Author: lbelzile

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: mev
 Type: Package
 Title: Modelling of Extreme Values
-Version: 1.17.1
+Version: 1.17.10002
 Authors@R: c(person(given="Leo", family="Belzile", role = c("aut", "cre"), email = "belzilel@gmail.com", comment = c(ORCID = "0000-0002-9135-014X")), person(given="Jennifer L.", family="Wadsworth", role=c("aut")), person(given="Paul J.", family="Northrop", role=c("aut"), comment = c(ORCID = "0000-0002-1992-4882")), person(given="Scott D.", family="Grimshaw", role=c("aut"), comment = c(ORCID = "0000-0002-6326-9360")), person(given="Jin", family="Zhang", role=c("ctb")), person(given="Michael A.", family="Stephens", role=c("ctb")), person(given="Art B.", family="Owen", role=c("ctb")), person(given="Raphael", family="Huser", role=c("aut"), comment = c(ORCID = "0000-0002-1228-2071")))
 Description: Various tools for the analysis of univariate, multivariate and functional extremes. Exact simulation from max-stable processes [Dombry, Engelke and Oesting (2016) <doi:10.1093/biomet/asw008>, R-Pareto processes for various parametric models, including Brown-Resnick (Wadsworth and Tawn, 2014, <doi:10.1093/biomet/ast042>) and Extremal Student (Thibaud and Opitz, 2015, <doi:10.1093/biomet/asv045>). Threshold selection methods, including Wadsworth (2016) <doi:10.1080/00401706.2014.998345>, and Northrop and Coleman (2014) <doi:10.1007/s10687-014-0183-z>. Multivariate extreme diagnostics. Estimation and likelihoods for univariate extremes, e.g., Coles (2001) <doi:10.1007/978-1-4471-3675-0>.
 License: GPL-3
@@ -23,6 +23,7 @@ Suggests:
     MASS,
     mvPot (>= 0.1.4),
     mvtnorm,
+    numDeriv,
     gmm,
     revdbayes,
     rmarkdown,

NAMESPACE

@@ -66,6 +66,7 @@ export(clikmgp)
 export(cvselect)
 export(dextgp)
 export(dextgp.G)
+export(dgeoaniso)
 export(dgev)
 export(dgp)
 export(distg)
@@ -161,6 +162,7 @@ export(infomat.test)
 export(intensBR)
 export(intensXstud)
 export(jac)
+export(kjtail)
 export(lambdadep)
 export(likmgp)
 export(maxstabtest)
@@ -223,6 +225,7 @@ importFrom(Rsolnp,solnp)
 importFrom(alabama,auglag)
 importFrom(alabama,constrOptim.nl)
 importFrom(grDevices,"colorRampPalette")
+importFrom(grDevices,hcl.colors)
 importFrom(grDevices,rainbow)
 importFrom(grDevices,rgb)
 importFrom(graphics,"abline")

NEWS.md

@@ -4,10 +4,12 @@
 
 - Stein's weighted generalized Pareto estimator with `fit.wgpd`
 - Multiple shape estimators (Hill, Pickands, moment estimator, generalized quantile) via `fit.shape` and specific routines `shape.moment`, `shape.pickands`, `shape.pickandsxu`, `shape.genquant` and `shape.hill`.
+- Function `dgeoaniso` for geometric anisotropy, using the parametrization of Rai and Brown (2025)
 
 ## Changes:
 
-- Function `PickandsXU` is deprecated and replaced with `shape.pickandsxu`, 
+- Function `PickandsXU` is deprecated and replaced with `shape.pickandsxu`
+- Function `fit.gpd` with method `obre` now returns ordered exceedances and weights.
 
 # mev 1.17  (Release date 2024-07-09)
 

R/RcppExports.R

@@ -6,17 +6,34 @@
 #' The function computes the distance between locations, with geometric anisotropy.
 #' The parametrization assumes there is a scale parameter, say \eqn{\sigma}, so that \code{scale}
 #' is the distortion for the second component only. The angle \code{rho} must lie in
-#' \eqn{[-\pi/2, \pi/2]}. The dilation and rotation matrix is 
+#' \eqn{[-\pi/2, \pi/2]}. The dilation and rotation matrix is
 #' \deqn{\left(\begin{matrix} \cos(\rho) & \sin(\rho) \\ - \sigma\sin(\rho) & \sigma\cos(\rho) \end{matrix} \right)}
 #' @param loc a \code{d} by 2 matrix of locations giving the coordinates of a site per row.
 #' @param scale numeric vector of length 1, greater than 1.
 #' @param rho angle for the anisotropy, must be larger than \eqn{\pi/2} in modulus.
 #' @return a \code{d} by \code{d} square matrix of pairwise distance
 #' @export
+#' @keywords internal
 distg <- function(loc, scale, rho) {
     .Call(`_mev_distg`, loc, scale, rho)
 }
 
+#' Distance matrix with geometric anisotropy
+#'
+#' The function computes the distance between locations, with geometric anisotropy.
+#' Consider real parameters \eqn{\theta_1} and \eqn{theta_2}, and the transformation \eqn{\psi=\arctan(\theta_1/\theta_2)/2} and \eqn{r=1 +\theta_1^2 + \theta_2^2}.
+#' The dilation and rotation matrix is
+#' \deqn{\left(\begin{matrix} \sqrt{r}\cos(\rho) & -\sqrt{r}\sin(\rho) \\ \sin(\rho)/\sqrt{r} & \cos(\rho)/\sqrt{r} \end{matrix} \right).}
+#'
+#' @param loc a \code{d} by 2 matrix of locations giving the coordinates of a site per row.
+#' @param theta numeric vector of length 2, real parameters
+#' @references Rai, K. and Brown, P.E. (2025), A parameter transformation of the anisotropic Matérn covariance function. Canadian Journal of Statistics e11839. \doi{10.1002/cjs.11839}
+#' @return a \code{d} by \code{d} square matrix of pairwise distance
+#' @export
+dgeoaniso <- function(loc, theta) {
+    .Call(`_mev_dgeoaniso`, loc, theta)
+}
+
 .EuclideanWeights <- function(x, mu) {
     .Call(`_mev_EuclideanWeights`, x, mu)
 }