nllmvsdir: Negative log-likelihood of scaled Dirichlet extreme value...

In lbelzile/ExtLiouv: Functions for 'Extremal attractor of Liouville copulas'

Description

This is the workhorse with the likelihood, which transforms observations and returns the objective function value.

Usage

nllmvsdir(y, thid, N, lambda, u, alpha, rho, scale, shape)

Arguments

y	n by d matrix of observations on original scale
thid	n by d matrix of logical vector indicating whether or not a threshold exceedance was observed
N	total number of observations (not just exceedances)
lambda	vector of percentage of threshold exceedances
u	threshold (on y-scale), to be substracted from y
alpha	vector of parameters for the Dirichlet model
rho	regular variation parameter
scale	vector of scale parameters
shape	vector of shape parameters

Value

the negative log-composite likelihood value

Note

The N argument is there mostly for cases in which function returns observations stripped from their contribution; in such cases, the count would otherwise be ‘incorrect’.

lbelzile/ExtLiouv documentation built on May 20, 2019, 8:28 p.m.