mc_build_sigma: Build variance-covariance matrix
In mcglm: Multivariate Covariance Generalized Linear Models

mc_build_sigma

R Documentation

Build variance-covariance matrix

This function builds a variance-covariance matrix, based on the variance function and omega matrix.

mc_build_sigma(
  mu,
  Ntrial = 1,
  tau,
  power,
  Z,
  sparse,
  variance,
  covariance,
  power_fixed,
  compute_derivative_beta = FALSE
)

`mu`	A numeric vector. In general the output from `mc_link_function`.
`Ntrial`	A numeric vector, or NULL or a numeric specifying the number of trials in the binomial experiment. It is useful only when using variance = binomialP or binomialPQ. In the other cases it will be ignored.
`tau`	A numeric vector.
`power`	A numeric or numeric vector. It should be one number for all variance functions except binomialPQ, in that case the argument specifies both p and q.
`Z`	A list of matrices.
`sparse`	Logical.
`variance`	String specifying the variance function: constant, tweedie, poisson_tweedie, binomialP or binomialPQ.
`covariance`	String specifying the covariance function: identity, inverse or expm.
`power_fixed`	Logical if the power parameter is fixed at initial value (TRUE). In the case power_fixed = FALSE the power parameter will be estimated.
`compute_derivative_beta`	Logical. Compute or not the derivative with respect to regression parameters.

A list with the Cholesky decomposition of Σ, Σ^{-1} and the derivative of Σ with respect to the power and tau parameters.

Wagner Hugo Bonat