kMatern: Matérn Kernels

In kergp: Gaussian Process Laboratory

Matérn Kernels

Description

Matérn kernels, obtained by plugging the Euclidian norm into a 1-dimensional Matérn function.

Usage

   kMatern(d, nu = "5/2")

Arguments

d	Dimension.
nu	Character corresponding to the smoothness parameter \nu of Matérn kernels. At this stage, the possible values are "1/2" (exponential kernel), "3/2" or "5/2".

Value

An object of class "covMan" with default parameters: 1 for ranges and variance values.

Note

Notice that these kernels are NOT obtained by tensor product.

References

C.E. Rasmussen and C.K.I. Williams (2006), Gaussian Processes for Machine Learning, the MIT Press, \Sexpr[results=rd]{tools:::Rd_expr_doi("10.7551/mitpress/3206.001.0001")}

Examples

kMatern()  # default: d = 1, nu = 5/2
kMatern(d = 2)
myMatern <- kMatern(d = 5, nu = "3/2")
coef(myMatern) <- c(range = 1:5, sigma2 = 0.1)
myMatern
try(kMatern(nu = 2))  # error

kergp documentation built on May 29, 2024, 10:25 a.m.