[R-sig-ME] interpretation of lme after convergence warning

[David Villegas Ríos]

Dear list,

I’m running some lme models, and some of them do not converge:



Error in lme.formula(value ~ trait * poly(month, 3) - 1, data = long41,  :

  nlminb problem, convergence error code = 1

  message = iteration limit reached without convergence (10)



By using the “control” argument (control=lmeCtlList) where…

lmeCtlList=lmeControl(maxIter=2,msMaxIter=3, tolerance=1e-4,
niter=4,msTol=1e-5, nlmStepMax=5,msVerbose=TRUE ,returnObject=TRUE)



…I’m able to tun the model until a warning is obtained.:



> lme41=lme(value~trait*poly(month,3)-1,data=long41,random=~trait-1|id,weights=varIdent(form=~1|trait),correlation=corAR1(form=~tim|id/trait),control=lmeCtlList)

  0:     32018.004: 0.372405 0.0422295 0.899729  1.28724 0.0267174
0.0599544  2.46838 -0.256311  2.12184 -2.41635  0.00000  0.00000
0.00000  0.00000

  1:     29219.716: 0.373150 0.0430342 0.903878  1.29395 0.0266993
0.0599082  2.47015 -0.256428  2.12293 -2.41762 0.303089 0.793629
-0.323934 -0.416277

  2:     28671.324: 0.293462 0.0160438 0.841610  1.27912 0.0256651
0.0605562  2.47777 -0.257765  2.12967 -2.43187 0.914815 0.813828
-0.436943 -1.19160

  3:     28482.303: 0.298865 0.0294951 0.850191  1.29426 0.0236547
0.0611059  2.47682 -0.257961  2.13046 -2.43399 0.898115 0.650545
-0.538846 -0.935158

  3:     28482.303: 0.298865 0.0294951 0.850191  1.29426 0.0236547
0.0611059  2.47682 -0.257961  2.13046 -2.43399 0.898115 0.650545
-0.538846 -0.935158

Warning message:

In lme.formula(value ~ trait * poly(month, 3) - 1, data = long41,  :

  nlminb problem, convergence error code = 1

  message = iteration limit reached without convergence (10)



But even so, I can get some output of the model with summary(lme41)



My question is: is the output of the summary still valid or should I avoid
interpreting it?



Thank you very much.


David

	[[alternative HTML version deleted]]