[R-sig-ME] question about nlmer

voigt.charlotte at web.de voigt.charlotte at web.de
Thu Apr 27 18:39:04 CEST 2017

Hi everyone,

I am using lme4 to fit nonlinear mixed effects models to biological systems.
Those systems describe reaction networks mathematically expressed in ODEs.
To include these ODEs in a mixed model, I wrote a function returning a value vector and a gradient attribute.
I want to use optimizers that also use gradient informations of the residuals, eg the L-BFGS-B from the optimx package.
The nlmer fits are slow in my case. So I checked some things and noticed that the gradient attribute returned by my nonlinear function is not passed to the optimizer.
So I guess the optimizer calculates its own gradients via finite differences.
Is that really the case and is there a possibility to pass the gradient attributes to the optimizer?
I would be grateful for an answer. If you have any further questions regarding the actual problem, please just ask.

More information about the R-sig-mixed-models mailing list