[R-sig-phylo] PGLS error

[Ties Coomber]

Hello,

I have a question about an error message that occurs sporadically when I
run multiple PGLS models using the pgls function in geiger.
I am performing all possible models on 15 independent variables (~ 30, 000
models).
Whilst I am looping across the models I get the following error message:

Problem with optim:52ERROR: ABNORMAL_TERMINATION_IN_LNSRCH

There is no pattern to when the error occurs. Sometimes a model will run
and sometimes it won't with the same data.
The problem I think is to do with the linear searching of the Maximum
Likelihood. And I have seen people with the same problem, when only using
the optim function. However, I have the added issue of using another
function around optim (geiger::pgls)

Does any one have any suggestions on why this may be happening? And how we
may solve this error?

Thanks,
Ties

[[alternative HTML version deleted]]

[Liam J. Revell]

Hi Ties.

pgls is in the package 'caper' not 'geiger'. You can fit the same models
using gls in the nlme package using correlation structures from ape.
It's possible that this implementation might better protect you against
regions of the likelihood space that are not defined. (Without a
specific example, though, this is hard to assess.)

- Liam

Liam J. Revell, Assistant Professor of Biology
University of Massachusetts Boston
web: http://faculty.umb.edu/liam.revell/
email: ***@umb.edu
blog: http://blog.phytools.org

Post by Ties Coomber
Hello,
I have a question about an error message that occurs sporadically when I
run multiple PGLS models using the pgls function in geiger.
I am performing all possible models on 15 independent variables (~ 30, 000
models).
Problem with optim:52ERROR: ABNORMAL_TERMINATION_IN_LNSRCH
There is no pattern to when the error occurs. Sometimes a model will run
and sometimes it won't with the same data.
The problem I think is to do with the linear searching of the Maximum
Likelihood. And I have seen people with the same problem, when only using
the optim function. However, I have the added issue of using another
function around optim (geiger::pgls)
Does any one have any suggestions on why this may be happening? And how we
may solve this error?
Thanks,
Ties
[[alternative HTML version deleted]]
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo