[R-sig-ME] lme4 and cholmod warning

[David Duffy]

On Mon, 2 Aug 2010, Marie-Agnes wrote:

> Dears,
>
> I encoutered the following message, when using lme4 :
>
> Warning messages:
> 1: In mer_finalize(ans) :
>  Cholmod warning 'not positive definite' at
> file:../Cholesky/t_cholmod_rowfac.c, line 432
> 2: In mer_finalize(ans) :
>  Cholmod warning 'not positive definite' at
> file:../Cholesky/t_cholmod_rowfac.c, line 432
> 3: In mer_finalize(ans) : singular convergence (7)
>
>
> I saw the same kind of message on your mailing list, but then I could
> not follow up the reply.
>
> My model was:
> > ga1p<-glmer(A1P~T131006+typop*mr+(1|pop/fam), family= poisson)
>
> The dataset is unbalanced:  there are 16 "pop", with a varying number of
> "fam" within pop. In one population, there is only one family.
> The problem is related to T131006 (when omitted, glmer works). When I
> replace this covariate by another one, I have the same message, except
> in one particular case.

Is T131006 continuous or categorical?  If the latter, what do 
table(T131006) and table(pop,T131006) look like?

Cheers, David Duffy
-- 
| David Duffy (MBBS PhD)                                         ,-_|\
| email: davidD at qimr.edu.au  ph: INT+61+7+3362-0217 fax: -0101  /     *
| Epidemiology Unit, Queensland Institute of Medical Research   \_,-._/
| 300 Herston Rd, Brisbane, Queensland 4029, Australia  GPG 4D0B994A v