Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Iterative least squares solvers #85

Open
f0uriest opened this issue Mar 7, 2024 · 1 comment · May be fixed by #86
Open

Iterative least squares solvers #85

f0uriest opened this issue Mar 7, 2024 · 1 comment · May be fixed by #86
Labels
feature New feature

Comments

@f0uriest
Copy link

f0uriest commented Mar 7, 2024

Hi @patrick-kidger , I've really enjoyed playing around with Lineax!

You have several iterative methods for square systems (CG, GMRES, BICGSTAB) but nothing for least squares problems.

I have a JAX port of scipy.sparse.linalg.lsmr that I should be able to get working with lineax, but wanted to check first:

  • is this something lineax users would be interested in?
  • any possible licensing issues with scipy? it's been partially rewritten for different control flow etc, but I'm always a bit paranoid about software copyright stuff
  • What sorts of tests are needed? I've tested it for basic array inputs but not pytrees, vmapping, implicit solves etc.
@patrick-kidger
Copy link
Owner

Definitely something we'd be interested in including!

On licensing issues: if you've used the original and then modified it, then we can probably include it whilst reproducing their own license in the implementation file, and in our own top-level LICENSE.

On tests: pytrees definitely need testing. You should also be able to register it with our jvp/vmap/etc. tests that already exist.

@patrick-kidger patrick-kidger added the feature New feature label Mar 7, 2024
@f0uriest f0uriest linked a pull request Mar 16, 2024 that will close this issue
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
feature New feature
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants