Warning message question

[Kelly Weinersmith]

Hello!

I'm redoing some statistics that I did ages ago, and after updating my packages I'm getting an error that I don't believe I got the last time I ran my code. For example, I ran the model below and received this warning message:

> m6C <- bglmm(log10(CORT) ~ (1|Tank.LetterU/Number.Letter) + Day + Day2 + Mass.c  + EUHADensity.c + RENBDensity.c, data=d, family=gaussian)

Warning message:

In mer_finalize(ans) : false convergence (8)

When I compare the precis() or summary() outputs from this model to models that did not yield this error message I get very similar estimates, so it at least looks like it's getting a reasonable output. 

So, is this a warning message I have to worry about? Probably. If so, what might be going wrong and how would I deal with it? 

Thanks!

Best,

Kelly

--

Kelly Weinersmith, ABD
Graduate Group in Ecology
University of California Davis

www.Weinersmith.com

[Jaime Ashander]

I'm not familiar with bglmm. But I have seen similar (the same?) messages from the underlying package (nlme or lme4 maybe?) that I assume provides mer_finalize. If there's a way to pass an argument about verbosity you may get more info. Look in the bglmm docs for this maybe?

Alternatively, if bglmm isn't supposed to use mer_finalize, you could have multiple packages loaded and some name collisions?

One final suggestion, after looking at the bglmm manual, is to try two things and see if results differ :

1 try a different method 

2 try with OPT = FALSE, which will result in more iterations with the current method

-Jaime 

--
Check out our R resources at http://www.noamross.net/davis-r-users-group.html
---
You received this message because you are subscribed to the Google Groups "Davis R Users' Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to davis-rug+...@googlegroups.com.
Visit this group at http://groups.google.com/group/davis-rug.
For more options, visit https://groups.google.com/d/optout.