patrick-kidger / patrick-kidger/optimistix

New solvers

Open
#3 2 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

feature
Dominant language
Python
Stars
623
Forks
54
PR merge metrics
No merged PRs in 30d

Description

  • Anderson acceleration
  • LBFGS
  • Affine

Powell's (unconstrained) derivative free optimisers:

  • UOBYQA
  • NEWUOA

On an affine solvers: such systems can be handled with a single linear solve. JAX can detect affine functions via

import jax
import jax.interpreters.partial_eval as pe

def is_affine(f, *args, **kwargs):
    jaxpr = jax.make_jaxpr(jax.jacfwd(f))(*args, **kwargs)
    _, used_inputs = pe.dce_jaxpr(jaxpr.jaxpr, [True] * len(jaxpr.out_avals))
    return all(not x for x in used_inputs)

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the existing LBFGS solver and the solver architecture in the repository, then compare the unchecked Anderson, affine, UOBYQA, and NEWUOA items with the JAX affine-detection sketch in the issue. Done means the selected new solvers are implemented and covered by the project's existing validation or test approach.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.