powered by
Transforms GPD observations to unit Frechet ones and vice versa
gpd2frech(x, loc = 0, scale = 1, shape = 0, pat = 1) frech2gpd(z, loc = 0, scale = 1, shape = 0, pat = 1)
A numeric vector.
The vector of observations.
The location, scale and shape parameters respectively.
The proportion above the threshold, i.e. Pr[X > log] = pat.
Mathieu Ribatet
Let \(x_i\), \(i=1,\ldots,n\) be the realisation of a GPD random variable. Thus, the transformation to unit Frechet is defined as:
$$z_i = - \frac{1}{\log\left[1 - pat \left(1 + shape \frac{x_i - loc}{scale}\right)_+^{-1/shape}\right]}$$
x <- rgpd(10, 0, 1, 0.25) z <- gpd2frech(x, 0, 1, 0.25) z all(frech2gpd(z, 0, 1, 0.25) == x)
Run the code above in your browser using DataLab