Andreas Noack
Andreas Noack
So far, we have aimed for type stability and that wouldn't be possible with the approach you are suggesting. I'd prefer that we just provide an easy way to switch...
I hadn't realized that it showed up like that in the docs. It should indeed be fixed somehow.
I think this broke some stuff via SymbolicUtils https://github.com/JuliaSymbolics/Metatheory.jl/runs/7651826562?check_suite_focus=true
Great. Then our package server will probably pick it up shortly.
Most relevant comments have already been made here but just wanted to point out that it seems that ML algorithms are often expressed and written by observation, i.e. for OLS...
Might be worth moving the implementation to a more upstream package for better visibility. It didn't look like the implementations were using anything MixedModel specific (but I might have missed...
No. There are many that cannot. E.g. `StaticArrays.SMatrix`
Possibly but generally numerical linear algebra is super inconvenient for immutable arrays.
Are we sure it's beneficial to return wrapped types instead of materializing them? The materialized versions should be quite cheap for many of the matrices here, no?
I think the docstring is inaccurate. E.g. I believe `adjoint(::Hermitian)` is still a noop.