cannot run model without residual error

[JonKnights]

I have a model that runs SAEM with an additive residual error model. If I comment out the add.err term from the ini block and comment out the IPRED ~ add(add.err) line I get the below error:

Error in parse(text = paste(x, collapse = "\n")) :
:2:2: unexpected ')'
1: function(){
2: })
^

How can I run this model without a residual error model? Apologies I cannot provide code snippets.

[mattfidler]

Hi @JonKnights

I am not sure what you are trying to do.

If you are trying to simulate from the model, that is (or will be) supported in rxode2, but not in the current RxODE and nlmixr.

If you are trying to estimate, this is not supported either. nlmixr2 supports fixing the additive error to a very small value which could be sufficient, but as far as I know this is not possible with nlmixr

[JonKnights]

Thanks @mattfidler !! Trying to estimate. We were able to fix an focei run with very small value but when I tried to run with SAEM I got an error that said cannot fix error to zero. I can reproduce if you've never seen this before.

[mattfidler]

Ok. Well saem in nlmixr simply cannot fix the residual error parameters to zero (or any other value). It is unsupported. I have seen it but I cannot do anything about it. This was a new feature in nlmixr2

[JonKnights]

Thanks for the clarification Matt.