Mohamed Tarek
Mohamed Tarek
Sounds like a good implementation approach.
Sounds good. I will try to contribute my understanding to the docs.
If you want a default AD backend to be present, I suggest depending on both AbstractDifferentiation and the default AD backend of choice, be it Zygote, ForwardDiff, ReverseDiff, etc. AbstractDifferentiation...
You can have as a conditional dependency if you always tell the users to load some AD package. But that's for the ProximalAlgorithms package authors to decide. They might want...
Sounds good! On the AD ranking, I think some users might want to have control over the package used even when multiple are loaded. Not sure if the `ZygoteFunction` approach...
Check https://github.com/JuliaDiff/AbstractDifferentiation.jl and the associated paper https://arxiv.org/abs/2109.12449.
Feel free to ping me if you have any questions :) I always wanted to add nonconvex optimisation support to ProximalAlgorithms but didn't get a chance to. If I understand...
Fantastic! Exactly what I was hoping for. Thanks!
How large are we talking @ChrisRackauckas?