[R] lme convergence error
gunter.berton at gene.com
Tue Jul 26 15:37:20 CEST 2011
1. This thread belongs on the r-sig-mixed-models list ,
2. I would bet that for some of the data sets, the model is
unbalanced/overspecified, causing te convergence issues (imagine
climbing a narrow ridge by zig-zagging back and forth across it).
On Tue, Jul 26, 2011 at 12:56 AM, ONKELINX, Thierry
<Thierry.ONKELINX at inbo.be> wrote:
> Dear Ben,
> Maybe the model converges more slowy than other models. Running more iterations might solve the problem. Have a look at ?lme and ?lmeControl.
> Best regards,
> PS R-sig-mixedmodels is a better list for questions on lme().
> ir. Thierry Onkelinx
> Instituut voor natuur- en bosonderzoek
> team Biometrie & Kwaliteitszorg
> Gaverstraat 4
> 9500 Geraardsbergen
> Research Institute for Nature and Forest
> team Biometrics & Quality Assurance
> Gaverstraat 4
> 9500 Geraardsbergen
> tel. + 32 54/436 185
> Thierry.Onkelinx at inbo.be
> To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of.
> ~ Sir Ronald Aylmer Fisher
> The plural of anecdote is not data.
> ~ Roger Brinner
> The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data.
> ~ John Tukey
>> -----Oorspronkelijk bericht-----
>> Van: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
>> Namens Ben Grannan
>> Verzonden: maandag 25 juli 2011 23:00
>> Aan: r-help at r-project.org
>> Onderwerp: [R] lme convergence error
>> Hello, I am working from a linux 64 machine on a server with R-2.12 (I can't
>> update to 2.13). I am iterating through many linear mixed models for
>> longitudinal data and I occasionally receive the following convergence
>> > BI.lme <- lme(cd4 ~ time + genBI + genBI:time + C1 + C2 + C11 + C12,
>> random =~ 1 + time | IID, data = d)
>> Error in lme.formula(cd4 ~ time + genBI + genBI:time + PC1 + :
>> nlminb problem, convergence error code = 1
>> message = false convergence (8)
>> >From iteration to iteration, genBI is the only variable which changes
>> >and it
>> is a numerical variable equal to 1, 2, or 3 for a given IID. When I remove the
>> interaction term genBI:cd4years there is no problem with convergence.
>> Also, I have played around with scaling time (days to years, eg) which can solve
>> the problem but then leads to convergence issues in other iterations.
>> When I shift the time (say in units of years by +2) the model also converges. So
>> it seems like the genBI:time is the issue. Do you have any suggestions for
>> exploring why I get these convergence errors only on certain iterations? I really
>> appreciate any suggestions!
>> Here is a summary of the included variables just in case it is helpful:
>> IID (808 groups) cd4 time ( in years)
>> ACTG_617 : 26 Min. : 0.0 Min. :0.0000 Min. :-0.032700
>> ACTG_1055: 25 1st Qu.: 288.0 1st Qu.:0.4603 1st Qu.: 0.001600
>> ACTG_1590: 25 Median : 434.0 Median :1.1534 Median : 0.004300
>> ACTG_200 : 23 Mean : 461.3 Mean :1.2107 Mean : 0.002139
>> ACTG_420 : 23 3rd Qu.: 599.0 3rd Qu.:1.8630 3rd Qu.: 0.006100
>> ACTG_454 : 23 Max. :2022.0 Max. :2.7808 Max. : 0.011300
>> (Other) :11992
>> C2 C11 C12
>> Min. :-0.105800 Min. :0.000 Min. :0.0000 Min. :1.000
>> 1st Qu.: 0.002100 1st Qu.:0.000 1st Qu.:0.0000 1st Qu.:1.000
>> Median : 0.007900 Median :0.000 Median :0.0000 Median :1.000
>> Mean : 0.001803 Mean :0.367 Mean :0.3020 Mean :1.095
>> 3rd Qu.: 0.011700 3rd Qu.:1.000 3rd Qu.:1.0000 3rd Qu.:1.000
>> Max. : 0.030800 Max. :1.000 Max. :1.0000 Max. :2.000
>> Min. :1.000
>> 1st Qu.:1.000
>> Median :1.000
>> Mean :1.369
>> 3rd Qu.:2.000
>> Max. :3.000
>> [[alternative HTML version deleted]]
>> R-help at r-project.org mailing list
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
> R-help at r-project.org mailing list
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
"Men by nature long to get on to the ultimate truths, and will often
be impatient with elementary studies or fight shy of them. If it were
possible to reach the ultimate truths without the elementary studies
usually prefixed to them, these would not be preparatory studies but
-- Maimonides (1135-1204)
Genentech Nonclinical Biostatistics
More information about the R-help