Inquiry on your paper " Zheng P, Barber R, Sorensen R J D, et al. Trimmed Constrained Mixed Effects Models: Formulations and Algorithms[Z]. 2020." and the codes.

[al00014]

Dear Professor Zheng,

I came across your lately published paper "[Zheng P, Barber R, Sorensen R J D, et al. Trimmed Constrained Mixed Effects Models: Formulations and Algorithms[Z]. 2020.](url: https://arxiv.org/abs/1909.10700)", it is astounding and inspiring. However, I have one question for your paper. In Table 1, you suggest that INLA package cannot deal with "Covariates in random effects variance", and has no "Linear constraints". Yet, it seems to me that the f() function within inla() function can exactly deal with covariates in random effects. Besides, the "extraconstr" argument in inla() can also introduce linear constraints to the fitting model. Thus, I have some doubts for your results.

Also, the link of codes provided by your paper is no longer accessible, could you please share with us your experimental codes for your paper?

Much appreciated,
Dr Xiao Lin

[zhengp0]

Hi @al00014,

Thanks for being interested!
I think the constraints functionality INLA has is different than what limetr. According to
https://becarioprecario.bitbucket.io/inla-gitbook/ch-INLAfeatures.html#sec:constraints
INLA has linear equality constraints on the random effects (latent effect). And limetr can do linear inequality constraints (and equality constraints) on fixed effects rather than the random effects.

Hope this helps and please let me know if this makes sense.
Best,
Peng