Guillaume Dalle
Guillaume Dalle
Great, thanks! I think it would be nice to add a line or two to the docs describing this constructor: it is [briefly mentioned](https://jonniedie.github.io/ComponentArrays.jl/dev/api/#ComponentArrays.ComponentArray) in the API but never demonstrated....
For future reference, @mohamed82008 is on the move again: https://github.com/JuliaNonconvex/NonconvexUtils.jl/pull/6
For future reference, @sethaxen implemented an FD-compatible matrix exponential in https://github.com/sethaxen/ExponentialAction.jl
So what would you recommend to compute the Jacobian of `A -> exp(A)` using ForwardDiff? For lack of a better option I used ExponentialAction with `B = diagm(ones(d))` (it doesn't...
That would be one way of doing it, the other being to ask the user to specify the backend as an argument to every algorithm. In any case, I think...
AbstractDifferentiation depends on many AD packages in a conditional way, thanks to [Requires](https://github.com/JuliaPackaging/Requires.jl). That is also something we could envision for ProximalAlgorithms: a conditional dependency on Zygote (only if the...
My initial motivation for this issue was being able to use ProximalAlgorithms without the very heavy Zygote dependency. Would it be possible to have it as a conditional dependency?
Yes that was precisely my question to them :wink:
Wait, I just found out that the `score` function already exists under the name `gradlogpdf`, but it is not in the docs :exploding_head: ```julia julia> using Distributions julia> gradlogpdf gradlogpdf...