This seems to be occurring when I try and constrain values across different matrices by label, then use omxAssignFirstParameters to set starting values - omxSetParameters gives the " 'labels' must not contain duplicate values" error. When I use the same label within the same matrix there is no problem. When browsing into the error, I see that indeed, the labels object within omxSetParameters contains duplicates only when they are specified from different matrices.
This should be easy to fix.
Which version of OpenMx are you using? Is this subversion trunk or 2.0 release?
Do you have code that you can share to easily reproduce this problem?
Hi,
I have the same problem (among other problems). I have attached the code and data here: http://openmx.psyc.virginia.edu/thread/3940
Have a nice day,
Ana
Yeah, the problem is fixed and will be corrected in the next release.
As a workaround, you can filter the result by unique names.
Sorry, I should get better at the bug report thing :)
Using the recent release, 2.0.0-0, on 64 bit windows pc's, the following code will show the failure:
test<-matrix(rnorm(100,0,1),ncol=2)
colnames(test)<-c('v1','v2')
testm<-mxModel(type='RAM',
mxData(test,type='raw'),
mxMatrix(name='A',ncol=2,nrow=2),
mxMatrix(name='S',ncol=2,nrow=2,labels=c('freematrix1[1,1]',NA,NA,'freematrix2[1,1]')),
mxMatrix(name='freematrix1',values=3,free=T,nrow=1,ncol=1,labels='freeparam1'),
mxMatrix(name='freematrix2',values=2,free=T,nrow=1,ncol=1,labels='freeparam1'),
mxMatrix(name='M',ncol=2,nrow=1),
mxMatrix(name='F',type='Diag',,values=1,ncol=2,nrow=2,dimnames=list(c('v1','v2'),c('v1','v2'))),
mxExpectationRAM(M='M',dimnames=c('v1','v2')),
mxFitFunctionML()
)
testm<-omxAssignFirstParameters(testm)
fit<-mxRun(testm)
summary(fit)
Wow, I can't believe we could have a bug like that. Oh well, SVN 4070
Add this test to models/passing/startingValues.R please. Prevent future issues.
Cheers guys :)