As I understand it, this function should pick one of ds or ztnb, whichever fits best. The problem is that if it selects ztnb, it doesn't return the same as ztnb. There is an obvious bug in the code that gives this effect, see below. Perhaps this is meant to be that way?
The functions differ because of a suspected bad variable assignment:
preseqR.rSAC <- function(n, r=1, mt=20, size=SIZE.INIT, mu=MU.INIT)
{
para <- preseqR.ztnb.em(n) ##this call is also different, all params are not sent in
shape <- para$size ##### HERE IT STARTS, THIS IS NOT ASSIGNED TO size
mu <- para$mu
the population is heterogeneous
because the coefficient of variation is large $1 / sqrt(shape)$
if (shape <= 1) {
f.rSAC <- ds.rSAC(n=n, r=r, mt=mt)
} else {
## the population is close to be homogeneous
## the ZTNB approach is applied
## the probability of a species observed in the initial sample
p <- 1 - dnbinom(0, size = size, mu = mu) ########HERE size is used
## L is the estimated number of species in total
L <- sum(as.numeric(n[, 2])) / p
## ZTNB estimator
f.rSAC <- function(t) {
L * pnbinom(r - 1, size=size, mu=mu*t, lower.tail=FALSE) ########HERE size is used
}
}
return(f.rSAC)
}
As I understand it, this function should pick one of ds or ztnb, whichever fits best. The problem is that if it selects ztnb, it doesn't return the same as ztnb. There is an obvious bug in the code that gives this effect, see below. Perhaps this is meant to be that way?
The functions differ because of a suspected bad variable assignment:
preseqR.rSAC <- function(n, r=1, mt=20, size=SIZE.INIT, mu=MU.INIT)
{
para <- preseqR.ztnb.em(n) ##this call is also different, all params are not sent in
shape <- para$size ##### HERE IT STARTS, THIS IS NOT ASSIGNED TO size
mu <- para$mu
the population is heterogeneous
because the coefficient of variation is large$1 / sqrt(shape)$
if (shape <= 1) {
f.rSAC <- ds.rSAC(n=n, r=r, mt=mt)
} else {
## the population is close to be homogeneous
## the ZTNB approach is applied
}
return(f.rSAC)
}