similar to: Error handling in nlme call

I am having trouble fitting correlation structures within nlme. I would like to fit corCAR1, corGaus and corExp correlation structures to my data. I either get the error "step halving reduced below minimum in pnls step" or alternatively R crashes. My dataset is similar to the CO2 example in the nlme package. The one major difference is that in my case the 'conc' steps are

Hello all. I'm doing a simulation study where I will be making use of the 'nlme' package. I want to loosen up the convergence criteria so that I increase the likelihood of convergence (potentially at the cost of obtaining slightly less than ideal results). The parameters in the function nlmeControl() control the convergence criteria. These default values can be modified to make

Dear Splus and R users: I have a problem in fitting a NLME model, the error message is: "step halving factor reduced below minimum in PNLS step" What does it mean and how to fix the problem, could anyone help me about it? Any suggestion/help would be greatly appreciated. Mei

Hi R-help; In using nlme, I often encounter the following error message: Error in nlme.formula(model = response ~ ......, data = ... : Step halving factor reduced below minimum in PNLS step Resetting the pnlsTol default value does not solve the problem. Any suggestions or workarouds for solving the problem? TIA, Richard

I'm maximising a reasonably complex function using nlme (version 3.1-65, have also tried 3.1-66) and am having trouble with fixed parameter estimates slightly away from the maximum of the log likelihood. I have profiled the log likelihood and it is a parabola but with sum dips. Interestingly changing the parameterisation moves the dips around slightly. Unfortunately the PNLS step is

I would appreciate confirmation that the function vcov(model.nlme) gives the var-cov matrix of the fixed effects in an nlme model. Presumably the random-effects var-cov matrix is given by cov(ranef (model.nlme)? Rob Forsyth

Hello,   I am attempting to use nlme to model the response of 19 groups.  I am able to get a reasonable fit of all of the groups to the observed data using an exponential decay model (a*exp-x*b).  The problem is that when I plot the fitted values to residuals, it demonstrates a pattern of increasing variance.   I attempt to model this variance using   Sig3.nlme <- update(Sig2.nlme, weights =

Hi I have found some postings referring to the fact that one can try and understand why a particular model is failing to solve/converge from the verbose output one can generate when fitting a nonlinear mixed model. I am trying to understand this output and have not been able to find out much: **Iteration 1 LME step: Loglik: -237.4517 , nlm iterations: 22 reStruct parameters: subjectno1

Dear R-user, I am trying to use the R "nlme" function to fit a non linear mixed effects model. The model I wand to fit is an individual somatic growth model with 4 parameters. For all parameters both fixed and random effects have to be estimated, as well as their covariance matrix (see the formula bellow). The data are simulated with the same growth model as in the nlme, with know

I want to extract the variance-covariance matrix of an nlme model of a dataset. The object is to pass this to mvrnorm to create pseudo- replicates of the original data. I note the nlme package has a getVarCov method available for lme objects but not nlme objects. Is the vcov function in the base stats package suitable? If so, why is the additional getVarCov provided? thank you Rob

>>>>> "VW" == Viechtbauer Wolfgang (STAT) <Wolfgang.Viechtbauer at STAT.unimaas.nl> >>>>> on Fri, 26 Sep 2008 18:00:19 +0200 writes: VW> Hi all, I'm trying to fit a marginal (longitudinal) VW> model with an exponential serial correlation function to VW> the Orange tree data set. However, R crashes frequently VW>

Hi List, I have to compare a relationship between y and x for two locations. I found logistic regression fits both datasets well, but I am not sure how to test if relationships for both sites are significantly different. I searched the r site, however no answers exactly match the question. I used Tukey's HSD to compare two means, but the relationship in my study was not simply linear. So I

After loading the RODBC package version 1.1-7, Ctrl-C changes its behaviour and is quitting R and returning to the (unix-)command prompt on the solaris2.8 platform here. Here's what happened before and after loading RODBC > for (i in 1:10^5) rnorm(10) ^C > library(RODBC) > for (i in 1:10^5) rnorm(10) ^C bash-3.00$ platform sparc-sun-solaris2.8 arch

Dear R users; I was trying to fit a nonlinear model using gnls (nlme version 3.1-80, R 2.5.0, WinXP) and I got the following error and warning message: Error in gnls(ht ~ a1 * hd * (1 - a2 * exp(-a3 * (dbh/dq2))), data = hdat, : Step halving factor reduced below minimum in NLS step In addition: Warning message: $ operator is deprecated for atomic vectors, returning NULL in:

I've been searching around for a function for computing the Brown-Forsythe F* statistic which is a substitute for the normal ANOVA F statistic for when there are unequal variances, and when there is evidence of non-normality. A couple of other people have asked this question, the responses I found have been: ?oneway.test However, that function appears to use the Welch W statistic which,

Hello: I have an two factorial random block design. It's a ecology experiment. My two factors are, guild removal and enfa removal. Both are two levels, 0 (no removal), 1 (removal). I have 5 blocks. But within each block, it's unbalanced at plot level because I have 5 plots instead of 4 in each block. Within each block, I have 1 plot with only guild removal, 1 plot with only enfa removal,

Using AIX 5.3 and gcc-4.2.1 for building R-2.7.1 with the following configure flags OBJECT_MODE=64 CC="gcc -maix64" CXX="g++ -maix64" F77="gfortran -maix64" FC="gfortran -maix64" CFLAGS="-g -O3" CXXFLAGS="-g -O3" FFLAGS="-g -O3" FCFLAGS="-g -O3" LDFLAGS="-L/$HOME/usr/local/lib/ppc64

This works fine: fit42<-nls(Vfs~SSlogis(Months,Asym.Int+Asym.Group*Groupdum,xmid,scal), data=df, start=c(Asym.Int=22,Asym.Group=5,xmid=2,scal=6), na.action=na.omit) But this, identical except using gnls, doesn't converge: fit43<-gnls(Vfs~SSlogis(Months,Asym.Int+Asym.Group*Groupdum,xmid,scal), data=df, start=c(Asym.Int=22,Asym.Group=5,xmid=2,scal=6), na.action=na.omit) Error in gnls(Vfs

Hi everyone. I'm working with nlme and I have a question regarding nlme fits that fail because of singularity issues. Specifically, there a way to return an nlme object when the estimation process runs into a singular matrix? For example, can the results up to the point of an error such as "Error in solve.default(pdMatrix(a, fact = TRUE)) : system is computationally singular" or

Dear all, I would like to simulate data from a nlme model using fixed effects and the variances of the random effects distribution. I scanned the help files and the mailing lists with no succes. Thanks for your help! Peter