There has been some discussion recently about improving scalar root-finding in SciPy, by adding a new root finder called modAB. GH link: ENH: optimize: Implement ModAB Root Finding method · Issue #24585 · scipy/scipy · GitHub
To make it easier for people to experiment with this new solver, I’ve created a small standalone test package.
It can be installed via pip using the instructions in the README.
This repository contains an implementation of the ModAB root-finding algorithm along with examples and tests. The intent is to make it easy for interested users to try it on real problems and evaluate how it behaves compared to existing SciPy methods such as those in scipy.optimize.root_scalar.
If you are interested in performance root finding methods, it would be quite helpful if you’d share feedback.
- If you have code that needs root-solving, is this method faster? Slower? The ModAB solver has been faster than the existing SciPy solvers in benchmarks, but that is not always reflective of real life.
- If you would use this if it were added to SciPy, please weigh in in the associated thread.
- If you observe any bugs, please report them to the associated thread as well.
Real-world tests and feedback are invaluable before adding new features to SciPy, so any experimentation and feedback from the community would be appreciated.