[R-sig-ME] Message error

[Ben Bolker]

   We probably need more detail to know anything for sure.

  These warnings were introduced in the most recent release of lme4
(1.1-5), and *may* be false positives.  Two things to check:

 (1) is your model fit singular?  If m is your fitted model, try
something like

  lbound <- getME(m,"lower")
  theta <- getME(m,"theta")
  any(lbound==0 & theta<1e-8)

 or fit your model with (g)lmerControl(check.conv.singular="warning")

 (2) if this is a glmer() fit, try glmerControl(optimizer="bobyqa") and
see if that helps.


On 14-03-29 03:25 PM, Neil Collier wrote:
> I'm not sure about the details of the error message but if R is suggesting
> to rescale the variables then you can do this using the scale() function as
> one option. Doing this usually helps the model converge.
> 
> ?scale
> 
> Neil.
> 
> 
> On Fri, Mar 28, 2014 at 1:31 PM, Cintia Akemi Oi <
> CintiaAkemi.Oi at bio.kuleuven.be> wrote:
> 
>> Hello,
>>
>> I'm having this warning messages:
>>
>> Warning messages:
>> 1: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
>>   Model failed to converge with max|grad| = 0.0523852 (tol = 0.001)
>> 2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
>>   Model is nearly unidentifiable: very large eigenvalue
>>  - Rescale variables?
>>
>> Thanks in advance.
>>
>>
>>
>>         [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-mixed-models at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>