LinearSolve.jl: High-Performance Unified Interface for Linear Solvers in Julia. Easily switch between factorization and Krylov methods, add preconditioners, and all in one interface.
I think this could be made fast. The * operation can be done lazily so there's really no impediment here other than generalizing whatever computation is getting stuck
Is there any way to use
LinearSolve.jl
algorithms withRecursiveArrayTools.ArrayPartition
? Concrete example:Fails with:
Which - at a first glance - does not seem too horrible to fix. But would this be performant?