Mohamed Tarek

Or even better, a package DifferentiableFactorizations.jl.

This showed up in https://github.com/JuliaNonconvex/Nonconvex.jl/issues/130 so just giving it a bump.

any update here?

Awesome!

Since these functions are different functions, I think having a method that takes in the output type as an argument would be a nice stop-gap solution right now for when...

I think storing the entire state could be useful then documenting how to get the invH for BFGS using the state.

We also needed this over in https://github.com/gdalle/ImplicitDifferentiation.jl.

My implementation even works for structs if you pass in the constructor, see the tests. Struct support needs more infrastructure though compared to the simple vec/reshape needed for vector/matrix support....

I mentioned to @gdalle before, every feature of NonconvexUtils should probably be its own package :)

ForwardDiffChainRules?