I found that the rkdevine function, from the package kdevine, is returning constant zero-valued simulations when it shouldn't be the case. The function is meant to simulate from a a kdevine object, which is vine-copula based density estimator
Example of the problem:
#1. Set the data
x <- c(0.5150864, 0.1661704, 0.5138176, 0.5044893, 0.4984769, 0.8359263, 0.4838860, 0.4799104, 0.4800063, 0.5012104)
y <- c(0.02299000, 0.03423000, 0.00649000, 0.01970000, 0.02184000, 0.01053000, 0.02399000, 0.02067000, 0.10918000, 0.04334681)
z <- c(0.13505051, 0.06968379, 0.06484263, 0.08655820, 0.07044886, 0.08455797, 0.06206888, 0.07187104, 0.93282970, 0.07841944)
#2. Create the copula density estimation
k.den <- kdevine(x = cbind(x, y, z), xmin = c(0.01, 0, 0.04), xmax = c(0.99, 2, 2), method = "TLL2nn")
#3. Generate simulations
rkdevine(1, k.den)
The output rkdevine(1, k.den) is c(0, 0, 0), which is obviously mistaken.
It tried looking at the function rkdevine but cannot find the mistake. Earlier versions of the package and/or R used to work.
Edward
22.3k3 gold badges18 silver badges37 bronze badges
lang-r
kdevinethat you use. Typeversion$version.stringfor your R version andpackage_version("kdevine")for the package version.