The generic functions dens, pp, qq and
rl create the diagnostic plots generated by
plot.uvevd.
Similarly, bvdens, bvcpp, bvdp and
bvqc create the diagnostic plots generated by
plot.bvevd and plot.bvpot.
There are internal fitting, simulation, distribution and density
functions for each bivariate and multivariate parametric model,
which are called from functions such as rbvevd and
rmvevd. There also exists internal functions for the
calculation and plotting of dependence and spectral density
functions, which are called from abvevd, hbvevd
and amvevd.
Dependence functions are ultimately plotted by the low-level
functions bvdepfn and tvdepfn.
The function pcint calculates profile confidence intervals,
and is called from the function plot.profile.evd. The fitting
function fgev calls the internal functions fgev.quantile
and fgev.norm for fits under different parameterizations.
The fitting function fpot calls the internal functions
fpot.norm and fpot.quantile.
Gev parameters used in marginal transforms are calculated using
frobgev, which avoids numerical data scaling issues.
The function ccop calculates condition copulas (i.e.
conditional distributions under uniform margins) for each
bivariate parametric model, and ccop.case does the
same for when a case indicator is implemented, conditioning
also on the case. They are needed to create the conditional
P-P plots generated by bvcpp.
The functions nsloc.transform, na.vals,
bvpost.optim, bvstart.vals and sep.bvdata
are used in the fitting of bivariate models. The function
mvalog.check checks and transforms the asy argument
for the multivariate asymmetric model. The function
subsets lists all subsets of 1:n; it is
called by mvalog.check and multivariate distribution
functions.
For fitting bivariate threshold models, internal functions exist
for the censored and (undocumented) point process likelihoods,
and each of these calls a further internal function
corresponding to the specified model. The internal function
bvtpost.optim is then used for post optimization processing.