[R-sig-ME] Error with nlme and varFixed

[biii m@iii@g oii de@@ey@ws]

Hi Phillip,

Thanks for the confirmation.  I searched and didn't see one line it.  I've submitted it here: https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17712

Thanks,

Bill

-----Original Message-----
From: Phillip Alday <phillip.alday using mpi.nl> 
Sent: Tuesday, February 11, 2020 2:43 PM
To: Bill Denney <wdenney using humanpredictions.com>; r-sig-mixed-models using r-project.org
Subject: Re: [R-sig-ME] Error with nlme and varFixed

This may be a bug -- report it at https://bugs.r-project.org/bugzilla/
(but check to make sure a related bug hasn't already been filed).

Phillip

On 29/01/2020 21:17, Bill Denney wrote:
> Hello,
>
> When trying to fit an NLME model using a vector of varFixed, I get an 
> error that appears to be related to a multiplication issue at this line of code:
>
> https://github.com/cran/nlme/blob/c006dfa23ad390948a74e67978a9828a6d60d89b/R/varFunc.R#L169
>
> Is the below a bug or am I inaccurately applying varFixed (or 
> something else)?
>
> Here is a reproducible example:
>
> library(nlme)
>
> d <-
>   data.frame(
>     obs=rnorm(n=97), # 97 chosen because it's prime and therefore  can't be the size of a rectangular matrix
>     groups=rep(c("A", "B"), each=50)[1:97],
>     wt=abs(rnorm(n=97))
>   )
>
> nlme(
>   obs~b,
>   fixed=b~1,
>   random=b~1|groups,
>   weights=varFixed(~wt),
>   start=c(b=0),
>   data=d
> )
>
> Which gets the error:
>
> Error in recalc.varFunc(object[[i]], conLin) :
>   dims [product 12] do not match the length of object [97]
> In addition: Warning message:
> In conLin$Xy * varWeights(object) :
>   longer object length is not a multiple of shorter object length
>
> Thanks,
>
> Bill