DRAFT: Fix: cache the compute ancillary params at indentation level

[PinkShnack]

As mentioned in #27, the ancillary param computation is currently very slow for extension models that require a model for the initial fit (e.g. KVM Hertz model).

This PR tries to fix that by caching the ancillary params at the Indentation (curve) level per fit. After the fit is complete, the cache is cleared for the next curve.

Perhaps we need a specific test for this with the KVM model? Thoughts?

Todo

• make sure TODO note in core.py is removed if relevant
• add ancillary test that actaully calculates ancillaries
• fix KVM model for numpy version 2
• ensure that pyjibe gives expected results with new cache system in nanite. Show plots here.

[codecov]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[paulmueller]

test_fit_ancillary.py uses a fake model with ancillaries. You can add a test there for caching (no need to include the KVM model AFAIK.

I would initialize the cache with None and invalidate it before the fit is actually called (to prevent possible race conditions).

[PinkShnack]

I added a test. I also checked to see how it looked with the KVM model locally. Turns out my nanite was using numpy 2 and therefore the KVM model extension failed due to a numpy np.argwhere error. Where can I fix the KVM model python code? (where is the code kept?).

[paulmueller]

I added a test. I also checked to see how it looked with the KVM model locally. Turns out my nanite was using numpy 2 and therefore the KVM model extension failed due to a numpy np.argwhere error. Where can I fix the KVM model python code? (where is the code kept?).

I think it is supplementary material to the paper. Not sure whether there is a public repository for it. We could add a curated list of external model as a repository to AFM-analysis?