update math function unit test with catch2

[lhchg]

Description

Some interfaces do not need to be exposed to the python API, except for unit testing, so move math unit tests to C++ unit tests, which implemented in python before

Fixes #2943

Checklist

• I have performed a self-review of my own code
• I have run clang-format (version 15) on any C++ source files (if applicable)
• I have followed the style guidelines for Python source files (if applicable)
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)

[lhchg]

Sorry, seems some error will be introduced. I'm checking

[paulromano]

Thanks for tackling this @lhchg! I haven't reviewed all the changes yet but noticed the test failure and see that it's related to the issue below:

[gridley]

Awesome!

One benefit of this change is that calc_pn_c function at a among a few others requires temporary space to be allocated while particles are in flight. That's not really a great code design, since we could be using the recursive formula instead for scoring purposes. It's nice that this PR removes dependence on that behavior for testing. It would be more straightforward to just test the values (regardless of how they're obtained) rather than having the intermediate API that assumes a certain output format.

One thing worth addressing: is it possible any users are relying on these math functions in the C API?

[lhchg]

Hi @paulromano @gridley

I updated the code to use hardcode for the reference part.
And I think if included a python script to generate the reference part, maybe this would introduce new python package dependencies when running the catch2 unit tests?

What do you think of this hardcode approach?

If you think it would be better to include a python script, I will continue to modify it.

[paulromano]

@lhchg Thanks for the updates on this! It looks good to me now.

@gridley Do you have any further requests on this one?

[gridley]

Awesome changes! Nice to have the code simplified some like this.

[lhchg]

There seems to be some dependency on the python math api

self = <openmc.polynomial.ZernikeRadial object at 0x7fe6bfe0dc90>, r = 0.196

    def __call__(self, r):
        import openmc.lib as lib
        if isinstance(r, Iterable):
            return [np.sum(self._norm_coef * lib.calc_zn_rad(self.order, r_i / self.radius))
                    for r_i in r]
        else:
>           return np.sum(self._norm_coef * lib.calc_zn_rad(self.order, r / self.radius))
E           AttributeError: module 'openmc.lib' has no attribute 'calc_zn_rad'

openmc/polynomial.py:82: AttributeError

I have fixed it