eq.compute doesn't correctly override grids to compute quantities that require evaluating functions along field lines

[rahulgaur104]

In the compute function in equilibrium.py if the user provides a grid, the function should set override_grid = False.

https://github.com/PlasmaControl/DESC/blob/f1689c85c022d5e8073325b7c057f6ea253923b4/desc/equilibrium/equilibrium.py#L794C1-L800C14

However, override_grid still stays True. This is a simple fix.

[unalmis]

This logic is intentional, see #440

[f0uriest]

See also #688 and related PR

[rahulgaur104]

Well, it definitely throws an error on the ballooning branch when I create a separate grid before calculating the growth rate using eq.compute.

I get assertion error and see this code

            data1dr = compute_fun(
                self,
                dep1dr,
                params=params,
                transforms=get_transforms(dep1dr, obj=self, grid=grid1dr, **kwargs),
                profiles=get_profiles(dep1dr, obj=self, grid=grid1dr),
                data=None,
                **kwargs,
            )
            # need to make this data broadcast with the data on the original grid
            data1dr = {
                key: grid.expand(
                    grid1dr.compress(val, surface_label="rho"), surface_label="rho"
                )
                for key, val in data1dr.items()
                if key in dep1dr
            }

[f0uriest]

It sounds like you might just need to pass override_grid=False when you call eq.compute if you have some custom grid you want to use and don't care about averages etc (because you're precomputing those on a different grid). That's the purpose of the override_grid option.

If you can post some code that reproduces the issue I can look into it more.

[rahulgaur104]

Yeah, with override_grid=False, it worked fine. I was just thinking that it could be a potential issue for people doing calculations on a custom grid. It also only becomes an issue when I do this calculation in a jupyter notebook.

Ok, I guess we can close this for now.

[unalmis]

At some point, can you show where you have needed to use override_grid=False so that I can check your code? I believe it indicates that your code has a bug. Edit: actually this isn't always true so nevermind.

For example, I recently fixed a bug in bounce branch where this was an issue with an inconsistent use of normalization between the toroidal flux label and desc's radial label. (This can lead to computations being done across different flux surfaces than expected). In this case, setting this to override_grid=False will temporarily supress the error, but it causes bugs that will pop up elsewhere.