search

SciML / LinearSolve.jl

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.
https://docs.sciml.ai/LinearSolve/stable/
Other
235 stars 51 forks source link

Custom linear solve documentation #502

Open ArnoStrouwen opened 2 months ago

ArnoStrouwen commented 2 months ago

Why are there 2 methods for doing custom linear solves: https://docs.sciml.ai/LinearSolve/dev/advanced/custom/ https://docs.sciml.ai/LinearSolve/dev/advanced/developing/ it is not clear to me when to use one over the other method.

In the latter method, in the function SciMLBase.solve!, it is not clear how the A matrix is passed. It would be better if this was a full runnable example.

There are inplace and outofplace problems. Neither of the two methods seems to interact with this aspect of the problem. It is unclear to me what is actually mutated in an inplace problem with a concrete A.

ChrisRackauckas commented 2 months ago

One is for developing new linear solvers, i.e. devdocs for this package. The other is a shorthand for passing in custom linear solver code by users.