Description Usage Arguments Value Examples
this version of HESS FBI includes the derivative w.r.t the exponential prior this function takes non-transformed theta parameters and takes the derivative with respect to log-transformed theta parameters.
1 | hessFBI(x, y, f, cor.par, lambda = 0.1)
|
x |
covariate matrix/ vector x |
y |
response vector |
f |
regression model, must be a matrix. if constant should be a vector (as.matrix) of 1s of length n, where n is the number of data points in x |
cor.par |
matrix of theta and alpha parameters for power-exponential model, includes two columns, the first for theta parameters and the second for alpha parameters. For Guassian correlation structure, alpha parameters can be initialized as 0. |
lambda |
parameter for exponential prior, defaults to 0.1 |
returns a matrix of partial second derivatives w.r.t tau = log(theta) correlation parameters. Taking the negative inverse of this hessian matrix provides the fisher approximation of the covariance
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | n <- 5
x1 <- seq(-5,10,length.out = n)
x2 <- seq(0,15,length.out = n)
x <- expand.grid(x1,x2)
d2 <- c(0.01,0.2,0,0)
cor.par <- data.frame(matrix(data = d2,nrow = dim(x)[2],ncol = 2))
names(cor.par) <- c("Theta.y","Alpha.y")
R <- cor.matrix(x,cor.par) # obtain covariance matrix
L <- chol(R)
z <- as.vector(rnorm(n^2))
y <- t(L)%*%z
f <- as.matrix(rep(1,n^2))
hess <- hessFBI(x,y,f,cor.par)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.