[R-sig-ME] lme4

[Safaie, Ebrahim]

Dear Ben Bolker,
Thanks for the quick answer. Yes, I admit that I did not mention  my problem clearly and do apologize for that. It was just because I came across the same error messages in the Qs & As that I did when my package was updated.
I was using the earlier version of lme4 and did not have any problems with it. For instance, for the following code I came to the following calculation without any error messages. (my dependent variable is binary and fixed factors are categorical)
summary(mod.15<-glmer(ErrorRate~1 +cgroup*cgrammaticality*cHeadNoun*cVerbType+(1|itemF)+(1+grammaticality*HeadNoun*VerbType|participantF),data =e3,+ family="binomial",na.action=na.exclude))

Fixed effects:
                                           Estimate Std. Error z value Pr(>|z|)
(Intercept)                                 -1.9745     0.1274 -15.494  < 2e-16 ***
cgroup                                       1.5843     0.1789   8.854  < 2e-16 ***
cgrammaticality                              0.5245     0.2182   2.404   0.0162 *
cHeadNoun                                   -0.2720     0.1853  -1.468   0.1422
cVerbType                                    0.7591     0.2326   3.263   0.0011 **
cgroup:cgrammaticality                       1.5796     0.3586   4.404 1.06e-05 ***
cgroup:cHeadNoun                             0.0475     0.3537   0.134   0.8932
cgrammaticality:cHeadNoun                    0.5368     0.4338   1.237   0.2159
cgroup:cVerbType                            -0.2441     0.3472  -0.703   0.4821
cgrammaticality:cVerbType                   -0.4861     0.4185  -1.162   0.2454
cHeadNoun:cVerbType                         -0.1563     0.3969  -0.394   0.6936
cgroup:cgrammaticality:cHeadNoun             0.2659     0.7161   0.371   0.7104
cgroup:cgrammaticality:cVerbType            -0.4691     0.6945  -0.675   0.4994
cgroup:cHeadNoun:cVerbType                   0.7661     0.6916   1.108   0.2679
cgrammaticality:cHeadNoun:cVerbType          0.9104     0.9147   0.995   0.3196
cgroup:cgrammaticality:cHeadNoun:cVerbType   3.1326     1.3994   2.239   0.0252 *




But as soon as I updated the package to a new version , for the same code I got the following error message and some calculations are not matched with the those with the earlier version (as shown below). I don't know exactly which version was the previous one, but I guess I was using 2013 packages

Warning messages:
1: In commonArgs(par, fn, control, environment()) :
  maxfun < 10 * length(par)^2 is not recommended.
2: In optwrap(optimizer, devfun, start, rho$lower, control = control,  :
  convergence code 1 from bobyqa: bobyqa -- maximum number of function evaluations exceeded
3: In (function (fn, par, lower = rep.int(-Inf, n), upper = rep.int(Inf,  :
  failure to converge in 10000 evaluations
4: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
  Model failed to converge with max|grad| = 0.0928109 (tol = 0.001, component 28)
5: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
  Model failed to converge: degenerate  Hessian with 1 negative eigenvalues




Fixed effects:
                                           Estimate Std. Error z value Pr(>|z|)
(Intercept)                                -1.97217    0.13810 -14.281  < 2e-16 ***
cgroup                                      1.58614    0.18262   8.686  < 2e-16 ***
cgrammaticality                             0.52725    0.24544   2.148   0.0317 *
cHeadNoun                                  -0.28061    0.21350  -1.314   0.1887
cVerbType                                   0.75503    0.25615   2.948   0.0032 **
cgroup:cgrammaticality                      1.57010    0.36695   4.279 1.88e-05 ***
cgroup:cHeadNoun                            0.05736    0.36138   0.159   0.8739
cgrammaticality:cHeadNoun                   0.55238    0.47616   1.160   0.2460
cgroup:cVerbType                           -0.24665    0.35618  -0.692   0.4886
cgrammaticality:cVerbType                  -0.49272    0.45732  -1.077   0.2813
cHeadNoun:cVerbType                        -0.14235    0.44553  -0.319   0.7493
cgroup:cgrammaticality:cHeadNoun            0.24468    0.73223   0.334   0.7383
cgroup:cgrammaticality:cVerbType           -0.45695    0.70627  -0.647   0.5176
cgroup:cHeadNoun:cVerbType                  0.75837    0.70763   1.072   0.2839
cgrammaticality:cHeadNoun:cVerbType         0.88375    0.98856   0.894   0.3713
cgroup:cgrammaticality:cHeadNoun:cVerbType  3.15344    1.42351   2.215   0.0267 *

My suggestion:
Because I am using Rstudio I have just two options when I want to install or to update packages. When I use CRAN and let Rstudio install lme4 automatically it  installs the most recent one. As such, it downloads the new package of lme4 which is problematic as I understand (sorry I might be wrong for that because I don't have any expertise but I'm talking from my observations)
So my suggestion is that let the earlier version of lme4 be on the CRAN such that when users are installing they automatically install the one which was not problematic.
Another option for me to download the earlier version and to install from my pc. But when I use this option from Rstudio, lme4 does not install and come with the following message.

install.packages("~/lme4_1.0-4.tar.gz", repos = NULL, type = "source")
Installing package into 'C:/Users/Azad/Documents/R/win-library/3.0'
(as 'lib' is unspecified)
* installing *source* package 'lme4' ...
** package 'lme4' successfully unpacked and MD5 sums checked
** libs


Sorry for the inconvenience and hope that  I've made things clear now.
Best wishes
Ebrahim


	[[alternative HTML version deleted]]