Andrew Giuliani

Hi Antoine, I wouldn't say that this is a bug, simply an artifact of using a multithreaded code. The discrepancy is due to OpenMP and goes away when you only...

There are a couple of ways of remedying this. 1. Replacing the critical regions with something that guarantees an ordering. I haven't tested this, but you could check if something...

Hi all, I've been able to SSH into the github runner and reproduce the error. You can do the same by adding ``` - name: Setup tmate session uses: mxschmitt/action-tmate@v3...

I'd suggest proposing the fix in a different PR than #418 to ensure it gets merged more quickly and so that we can see precisely what was changed.

This functionality is already implemented in the shared dofs feature. If you want a new `SurfaceRZFourier` with the same dofs as another, but different set of quadrature points, see the...

If you want to get this working on `SurfaceXYZFourier` and `SurfaceXYZTensorFourier`, I can suggest a more general approach since the user is allowed to change `mpol`, `ntor`, `quadpoints`, `stellsym`, etc......

> Can a test be added in `tests/core/test_derivative.py` to cover this issue using the little optimizable functions there without relying on `BoozerSurface`? This is checked now in `test_as_derivative2`. The issue...

What is the dimension of the array returned by dJ()?

Hi @abaillod, I am happy to dual debug with you, send me a calendar invite

@mishapadidar for future reference, I'm including the figure illustrating what the algorithm is doing (all angles are normalized by 2*pi for consistency with Surface objects in simsopt). The first graph...