Improve waveform classes & derivatives numerical precision

[jacopok]

Thanks for this @AlessandroAgapito23 ! I think the changes are overall good, but there are some rough edges to take care of. See my comments above. Also, you can see that the automated tests are failing - some of them were using GWFish's TaylorF2, and something is going wrong there with the spin variables. You can see the details from the Github interface here, or you can rerun the tests locally (see development installation).

If you want (not compulsory for this to be merged), it would also be great to have some tests for the new PPE waveforms. These could go into the test/test_waveforms.py file, or another one as you prefer. For example, a good and simple one might be to check that the PPE waveforms reduce to the regular TaylorF2 when the PE corrections are zero. You can see what the basic syntax of the tests looks like by looking at the ones that are already present; also you can check out the pytest docs.

Finally, since the waveforms module was originally @bvgoncharov 's child I think we could benefit from his opinion as well.

[bvgoncharov]

Hi @AlessandroAgapito23 , @jacopok , I reviewed the code, please see my comments above

[AlessandroAgapito23]

Hi @bvgoncharov, unfortunately I don't see any comments above. Where can I find them?

[AlessandroAgapito23]

Between line 507 and 590, so regarding "calculate_frequency_domain_strain() function", I changed the following things :

• Definition of the frequency as "frequencyvector" instead of "ff" which is a dimensionless frequency in IMRPhenomD
• Addition of the spin-corrections into the phi_i terms of the psi_TF2 phase, as in IMRPhenomD and, by doing that, I have used the same syntax between the two models
• Addition of a different f_cut (respect to f_cut = 4 * f_isco) in the gw_parms ( default 'cut' : 4)

Note : I was wrong in writing the psi_tot in "TaylorF2", there was a typo in the self.psi definition. Now I think that the "TaylorF2" waveform should work, I'm going to check it.