general machinery?

[bbolker]

I'm still torn about whether we should include more general engines (e.g. Stan, NIMBLE, greta, JAGS/BUGS, rethinking) and interfaces thereto in this Task View. Opinions?

[emitanaka]

I'm in favour of this. I notice there are already references to ASReml-R and INLA so it would be in a similar spirit as these to add others (except perhaps greta which could go under Convenience wrappers or Bayesian).

[bbolker]

There are a few separate (partly orthogonal) categories:

• on CRAN (greta, rstan, cmdstanr, nimble, R2jags etc. [grep("jags|bugs", rownames(available.packages()), value = TRUE, ignore.case=TRUE)] vs not (INLA, ASReml-R)
• FOSS vs not (ASReml-R)
• high-level interface (INLA) vs not (everything else)

The main distinction I was trying to make here is that these frameworks all require you to construct the model from a more basic level than just specifying response variables, fixed effects/RE structures etc.

[emitanaka]

I would say many of the mixed model R packages are in fact high-level interface or wrappers for R (brms for Stan, R2jags for jags, greta for TensorFlow, INLA for INLA, ASReml-R for ASReml standalone, etc). I think it makes more sense to give prominence to packages on CRAN (which I general assume is FOSS), and have well-known, but non-FOSS, ones (e.g. ASReml, SAS) at the bottom including any dependency on these (e.g. asremlPlus).

[bbolker]

closed in c35ccc4