FieldQuadrupole element based on FieldElement

[agolovanov]

Quadrupole element based on the FieldElement class for tracking a bunch through a quadrupole without approximations.

My current implementation is uniform in longitudinal direction field, without Bz, without the skew component (so orientation only along x or y depending on the sign of foc_strength), and without automatic estimation of the needed timestep. All these missing things are possible improvements to the element.

I am not completely satisfied with the class name FieldQuadrupole (as Quadrupole is already in use), but couldn't come with anything nicer.

I tested it against my own calculation of a thin quadrupole (without Wake-T), but it can't be used as an automated test. Should I design a test against the existing TM quadrupole element for some monoenergetic bunch?

Also, I noticed that docstrings for classes are given after the class. Is this a convention in this project? PEP 257 recommends putting docstrings for the __init__ method instead: The class constructor should be documented in the docstring for its __init__ method.

I also didn't check how to add the new element to the API reference in generated documentation or whether it is done automatically.

[AngelFP]

Hi @agolovanov , thanks a lot for implementing this, and for taking care of the issues in the tests.

Overall it looks good to me. I would however suggest changing the sign of the field, so that a positive value is focusing for electrons on the x plane. This would make it more consistent with the existing Quadrupole element. Do you think this would be fine? Or was there any reason for the other choice?

[AngelFP]

Regaring the docstring of the __init__ method, we are following the numpy style guide https://numpydoc.readthedocs.io/en/latest/format.html#class-docstring, which recommends documenting it in the class docstring. This helps when building the online documentation with sphinx, which shows a class documentation that is ordered better.

[agolovanov]

@AngelFP I used the other sign because that's how it was written in the formula in Wikipedia. You are right that it's better for consistency with already existing code to flip the sign, so I've changed it.

I've also added a new test comparing FieldQuadrupole with the existing Quadrupole.

Thank you for implementing the automatic dt!

[AngelFP]

Looks great! Thanks for making the changes and adding a new test :D