[R-sig-ME] GLMM's with quadrature

Ben Bolker bbolker at gmail.com
Thu Dec 11 03:28:39 CET 2014


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 14-12-10 03:58 PM, Boylan, Ross wrote:
> There seems to be a pattern here of the quadrature, or at least
> the
>> 1d quadrature being dropped.  Is it that the previous
> implementations were deemed unreliable?  Quadrature can be
> important in getting accurate results.

  In the case of lme4 it wasn't unreliability, it was just the hassle
of reimplementing the general case.  It's not out of the question,
it's just a matter of time and effort.  Pull requests are welcome!

I have a few notes on vector-valued GH at:

 http://htmlpreview.github.com/?https://github.com/lme4/lme4/blob/master/misc/notes/laplDiag.html

  Dave, can ADMB do G-H with vector-valued random effects?
http://otter-rsch.com/admbre/admbre.pdf says GH applies:

In the situation where the model is separable of type ”block diagonal
Hessian” with only a single random effect in each block (see Section 2.4)

  it sounds to me like this excludes vector-value random effects, but
I could be mistaken ...


> 
> Ross Boylan ________________________________________ From: dave 
> fournier [davef at otter-rsch.com] Sent: Wednesday, December 10, 2014 
> 7:14 AM To: Boylan, Ross Subject: Re: [R-sig-ME] GLMM's with 
> quadrature
> 
> You can do that with ADMB. Unfortunatly the glmmadmb package lost 
> the ability to do gauss-hermite integration when it was rewritten, 
> but the underlying software has that ability for multiple grouping 
> factors.
> 
> _______________________________________________ 
> R-sig-mixed-models at r-project.org mailing list 
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iQEcBAEBAgAGBQJUiQFXAAoJEOCV5YRblxUHkhoH/A8qPdHhRGbNFQLjBRv04o/8
fQztMvn+GQ7f3rm9Dux/uxZ7Qq0KkTSUNrRggumy97DG8irMrPuskXOi9C9Xt4E+
6Sc5t7aEct/Q34XBf3Pr9dvyMGJQuhvnoZUY1zFmiVXyw1xLpz8EhWmupoCglVlw
r5JQcFqF3IIzvcsFiQBFWmGA5HxOC73uYw0Gmso6O1nEpJ2gTMnKqHAcUKHIqeqf
7rO/1NrUhkHnER2ETbcav/aCzLWb388KXnJx7IsQn4z2GVMeUYAIMLittcnmTC5z
KtRaQDcw1oiIpib4zvA3bZKbDmbZVN2LtZoR5r2u9/AVCWNaMHYxx0+4yeb4ZpE=
=1LRH
-----END PGP SIGNATURE-----



More information about the R-sig-mixed-models mailing list