[R] lme convergence

[Spencer Graves]

Dear Dr. Chaudhary:

	  Because your example is not self contained (i.e., your data.frame 
"family" is not something distributed with R), I can only guess.

	  Have you tried things like the following:

	  with(family, table(cengirth, familyid))

	  If each level of familyid is associated with only one level of 
cengirth, then it's clear that your data cannot support the estimation 
of a random perturbation to the slope of the line for "cengirth" 
associated with each level of familyid.  If you have only a very few 
such linkages, I suggest you increase the iteration limit and try again; 
  the lme documentation will tell you how to do that.

	  Also, have you read Pinheiro and Bates (2000) Mixed-Effects Models in 
S and S-Plus?  I've learned some extremely valuable things from Doug 
Bates, including from the contents of this book, and I highly recommend 
it.

	  If it were my problem and I'd done more or less all of the above 
without getting an answer to my question, I might start looking for a 
simpler example that would still replicate the error.  If I could have 
copied a few lines of code out of your email, pasted them into R and 
replicated your error, I almost certainly could have given a more 
precise diagnosis.  If you get a simple example that you still don't 
understand and would like further help from this group, PLEASE do read 
the posting guide! "www.R-project.org/posting-guide.html" and submit a 
follow-on question describing this simple example and what you don't 
understand about it.

	  hope this helps.
	  spencer graves

Chaudhary, Mohammad A. wrote:

> Dear All:
> 
>  
> 
> lme(sbp ~ cengirth, data = family, random= ~ 1 | familyid)
> 
>  
> 
> converges but 
> 
>  
> 
> lme(sbp ~ cengirth, data = family, random= ~ cengirth | familyid)
> 
>  
> 
> does not.
> 
>  
> 
> I get the following message:
> 
>  
> 
> Error in lme.formula(sbp ~ cengirth, data = family, random = ~cengirth |
> : 
> 
>         iteration limit reached without convergence (9)
> 
>  
> 
> The data has 488 rows and 154 familyid levels. For some familyid levels,
> there is only one row and cengirth is 0 as it is a centered variable
> within familyid.
> 
>  
> 
> I appreciate your pointing me to the source of the problem.
> 
>  
> 
> Ashraf 
> 
> _________________________________
> M. Ashraf Chaudhary, Ph.D.
> Department of International Health
> Johns Hopkins Bloomberg School of Public Health
> 615 N. Wolfe St.  Room W5506
> Baltimore MD 21205-2179
> 
> (410) 502-0741/fax: (410) 502-6733
> mchaudha at jhsph.edu <mailto:mchaudha at jhsph.edu> 
> 
>  
> 
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html