search

stan-dev / gmo

Inference on marginal distributions using gradient-based optimization
BSD 3-Clause "New" or "Revised" License
13 stars 4 forks source link

gmo_lmer, gmo_glmer, gmo_nlmer #10

Open dustinvtran opened 7 years ago

dustinvtran commented 7 years ago

Implement specialized algorithms for mixed effects models (see Appendix G in write-up). Take functions internal to lme4. See the feature/lmer branch.

(this is a big pain as lme4 doesn't expose everything)

dustinvtran commented 7 years ago

@bob-carpenter writes:

We can't use lme4's code for Stan itself because of licensing. It's GPL and we're using BSD.

That's a good point. If we decide to implement these algorithms in Stan, we'll most likely have to write lme's internal functions from scratch—which is fine.

At a higher level, there's likely discussion to be made on whether we should implement algorithms specialized to certain models in Stan (both on whether philosophically it should be done, and how it can be done implementation-wise). I propose postponing that discussion until we have a clearer idea of the usefulness of these algorithms, and when we start putting GMO in Stan.