problems of phylogenetic signal

[dongyiyi]

CASE B # : no phylogenetic signal
##########
trait0 <- c(2,3,1,3,3,3,3,2,2,3,1,2,1,2,3,1,2,3,1,2)

#CALCULATE DELTA B
deltaB <- delta(trait0,tree0,lambda0,se,sim,thin,burn)
print(deltaB)##It should be detaB
#and I got a warning:
Warning message:
In sqrt(diag(solve(h))) : NaNs produced
#how to avoid this warning?
#I fellow code of the example to calculate the delta values, and I want to get the result of boxplot, what should I do?
Thanks!

[mrborges23]

Can you please try to change some arguments inside the ace function (this function is used inside the function delta, currently defined as ace(trait,tree,type="discret",method="ML",model="ARD")$lik.anc). This function comes inside the ape library and I suspect that is the matrix exponentiation routine implemented there is returning this errors. Can you please play with the arguments use.expm=T and use.eigen=T (description of these two arguments here: https://www.rdocumentation.org/packages/ape/versions/5.3/topics/ace). These methods should, in principle, give better results. I have been using expm myself in other applications, and I have experienced no issued thus far.