[R-sig-ME] Error in a negative binomial, zero-inflated glmmADMB with random effect

[Holly G]

I apologize for the formatting issues that occurred with my previous email, so I am sending the same message again as plain text:

I have been working with a dataset that is large, overdispersed, non-normally distributed, non-linear, spatially autocorrelated, and zero-inflated. So far, I have been able to account for everything except for the zero-inflation using GAMMs in R. To my knowledge, there is no general software code available yet to incorporate zero- inflation into negative binomial GAMMs, please correct me if I am wrong. As far as GLMMs, I have modeled the data with glmmPQL (accounting for random effects and autocorrelation structure), and zeroinfl{pscl} (accounting for zero-inflation). For a zero-inflated model that includes random effects, I am attempting glmm.admb (the example from the documentation ran smoothly), but have encountered a few errors, all a variation of “The function maximizer failed” (below).  I am not sure if these errors are due to the nature of the data or to a mistake that I overlooked, so I provide details below. I would appreciate any advice.
Thank you,
Holly

> glmmadmbZINBrLogCRATdens5fixGrpSrvy <- glmm.admb(
+ fixed=round(LogCRATdens)~
+ Dist + 
+ DiffST1 +
+ LogTnDens +
+ LogDlDens +
+ Sal,
+ group="fSurveyID",
+ data= Data110217noFlNaRVfallsub, 
+ zeroInflation=TRUE,
+ family="nbinom")

0 30.22 1 14.145 0.17 0 0 30.24 1 14.861 0.08 0 0 30.25 1 14.861 0.21 0 0 30.29 1 15.665 0.22 0 0 30.4 $
size error in get_choleski
Error in glmm.admb(fixed = round(LogCRATdens) ~ Dist + DiffST1 + LogTnDens +  : 
  The function maximizer failed
In addition: Warning messages:
1: running command 'C:\Windows\system32\cmd.exe /c "C:/Users/Yo/Documents/R/win-library/2.12/glmmADMB/bin/windows/nbmm.exe" -maxfn 500 ' had status 1 
2: In shell(cmd, invisible = TRUE) :
  '"C:/Users/Yo/Documents/R/win-library/2.12/glmmADMB/bin/windows/nbmm.exe" -maxfn 500 ' execution failed with error code 1

I have tried a variety of alternatives to this model, varying the fixed effects, the group, the random effect, the response variable, and zeroInflation, to no avail. Changing imaxfn values anywhere from 10 to 1000000 produces the same error.

Using either non-integers or counts (integers) for the response variable, I get the following error instead of the “size error in get_choleski”:

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 $
Incompatible array bounds in dvar_vector  operator * (_CONST dvar_matrix& m,_CONST dvar_vector& x)

Using 3 of the variables, I don’t get either of the above errors, but simply: “The function maximizer failed”:  

> glmmadmbZINBrLogCRATdens3fixGrpSrvyRand1 <- glmm.admb(
+ fixed=round(LogCRATdens)~
+ Dist + 
+ DiffST1 +
+ LogTnDens,
+ group="fSurveyID",
+ random =~1,
+ data= Data110217noFlNaRVfallsub, 
+ zeroInflation=TRUE,
+ family="nbinom")
1 143.521 0.02 0 1 143.13 4.54 0 1 143.636 4.54 0 1 145.986 3.3 0 1 146.859 0.27 0 1 147.734 0.27 0 1 $
Error in glmm.admb(fixed = round(LogCRATdens) ~ Dist + DiffST1 + LogTnDens,  : 
  The function maximizer failed
In addition: Warning messages:
1: running command 'C:\Windows\system32\cmd.exe /c "C:/Users/Yo/Documents/R/win-library/2.12/glmmADMB/bin/windows/nbmm.exe" -maxfn 500 ' had status 1 
2: In shell(cmd, invisible = TRUE) :
  '"C:/Users/Yo/Documents/R/win-library/2.12/glmmADMB/bin/windows/nbmm.exe" -maxfn 500 ' execution failed with error code 1

 > str(Data110217noFlNaRVfallsub)
'data.frame':   9559 obs. of  18 variables:
 $ LogCRATdens  : num  0 0 0 0 0 0 0 0 0 0 ...
 $ CRATdens     : num  0 0 0 0 0 0 0 0 0 0 ...
 $ CRATcount    : int  0 0 0 0 0 0 0 0 0 0 ...
 $ TransBegDate : int  39674 39674 39674 39674 39674 39674 39674 39674 39674 39674 ...
 $ fTransBegDate: Factor w/ 126 levels "38979","38980",..: 34 34 34 34 34 34 34 34 34 34 ...
 $ x            : num  1867 1867 1867 1867 1867 ...
 $ y            : num  320 319 318 317 316 ...
 $ fSurveyID    : Factor w/ 8 levels "DEL0615","DEL0706",..: 4 4 4 4 4 4 4 4 4 4 ...
 $ fSurveyLeg   : Factor w/ 16 levels "EMaug08a","EMaug08b",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ Dist         : num  26.1 26.9 27 27.8 28.6 ...
 $ ST           : num  20.9 20.9 20.9 20.8 20.4 ...
 $ DiffST1      : num  0.03 0.06 0.07 0.38 0.38 0.22 0.22 0.16 0.12 0.12 ...
 $ Sal          : num  31.5 31.5 31.5 31.6 31.6 ...
 $ LogTnDens    : num  0 0 0 0 0 0 0 0 0 0 ...
 $ TnDens       : num  0 0 0 0 0 0 0 0 0 0 ...
 $ LogDlDens    : num  0 0 0 0 0 0 0 0 0 0 ...
 $ DlDens       : num  0 0 0 0 0 0 0 0 0 0 ...
 $ EW           : num  9.49 8.87 8.55 9.17 9.35 9.11 8.47 8.42 8.7 9.24 ...

> sapply(Data110217noFlNaRVfallsub, function(i) length(table(i)))
LogCRATdens    CRATdens    CRATcount    TransBegDate    fTransBegDate
51    51    41    83    126
x    y    fSurveyID    fSurveyLeg    Dist
9559    9559    8    16    5044
ST    DiffST1    Sal    LogTnDens    TnDens
1265    266    1445    11    12
LogDlDens    DlDens    EW        
30    33    3481