ImplicitDifferentiation.jl icon indicating copy to clipboard operation
ImplicitDifferentiation.jl copied to clipboard

Published 20 hours ago • verified publisher icon JuliaDecisionFocusedLearning

Fully mutating linear solver

Open gdalle opened this issue 2 years ago • 4 comments

At the moment, AbstractDifferentiation does not have a fully mutating API, and benchmarks show allocations are non-negligible when multiplying by A and B. Opening this to track https://github.com/JuliaDiff/AbstractDifferentiation.jl/issues/14 and see what we can do

gdalle avatar May 24 '23 07:05 gdalle

The problem is that most AD packages don't actually support mutation.

mohdibntarek avatar May 24 '23 10:05 mohdibntarek

Yeah I know, it would be useless for ChainRules, but not for ForwardDiff?

gdalle avatar May 24 '23 10:05 gdalle

Trying to solve this with https://github.com/gdalle/DifferentiationInterface.jl

gdalle avatar Feb 21 '24 13:02 gdalle

This is now possible with DifferentiationInterface but

  • it won't work with every outer backend
  • it requires a mutating version of the conditions conditions!(c, x, y)

I hope to make it part of v0.7

gdalle avatar Apr 10 '24 13:04 gdalle