Douglas Bates

An alternative for the parametric bootstrap is just to introduce a new property, say `tbl`, which has the `Table`-compatible structure and deprecate the `allpars` property.

Another couple of things to throw in the hopper for a breaking version is to change the name and default value of the argument to `fit` that causes the `fitlog`...

Exactly. I am getting to the point where I keep coming up with the same ideas, over and over, and considering them to be novel approaches. I suppose in a...

The `"goldstein"` test set in `test/pirls.jl` is an example where `init_from_lmm` provides much different, and superior (in the sense of non-trivially smaller deviance), parameter estimates. ```julia julia> using MixedModels, PooledArrays...

When I initially looked at this I thought that the starting values were from an unweighted LMM fit to the same formula/data combination as the GLMM, which made me wonder...

You have to scroll to the right to see the calls to `fit` but the bottom line is that without `init_from_lmm` the fit takes about 3.6 seconds and with it...

Also a 4-fold increase in speed for `init_from_lmm` on the `verbagg` example. ```julia julia> @b fit(MixedModel, @formula(r2 ~ 1 + anger + gender + btype + situ + (1|subj) +...

Here's a collection of model fits, including one with vector-valued random effects. ```julia julia> runglbmk(gltbl; init_from_lmm=(:β, :θ)) Table with 4 columns and 4 rows: bmk dsnm dist frm ┌───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── 1...

code is in `bench/runbenchmarks.jl` in the `db/RegressionTests` branch

I tried using juliac as described in https://github.com/TheCedarPrince/InteroperableJuliaBinaries but I am on an M1 Mac and ran into the problems described in Issue 1 of that repository.