Compute Approximated Variance Covariance Matrix

Find the approximated variance covariance matrix of the parameters.

compute_vcov(obj)

Arguments

obj

a fitted object, either with fitCauchy or cauphylm.

Value

The same object, with added vcov entry.

Details

This function computes the numerical Hessian of the likelihood at the optimal value using function hessian, and then uses its inverse to approximate the variance covariance matrix. It can be used to compute confidence intervals with functions confint.cauphylm or confint.cauphyfit.

confint.cauphylm and confint.cauphyfit internally call compute_vcov, but do not save the result. This function can be used to save the vcov matrix.

See also

fitCauchy, cauphylm, confint.cauphylm, confint.cauphyfit, vcov.cauphylm, vcov.cauphyfit

Examples

# Simulate tree and data
set.seed(1289)
phy <- ape::rphylo(20, 0.1, 0)
dat <- rTraitCauchy(n = 1, phy = phy, model = "cauchy",
                    parameters = list(root.value = 10, disp = 0.1))
# Fit the data, without computing the Hessian at the estimated parameters.
fit <- fitCauchy(phy, dat, model = "cauchy", method = "reml", hessian = FALSE)
# Precompute the vcov matrix
fit <- compute_vcov(fit)
# Approximate confidence intervals
confint(fit)
#> Approximated asymptotic confidence interval using the Hessian.
#>          2.5 %    97.5 %
#> disp 0.0240104 0.1010921