Mismatch between Compton profile and atomic relaxation data

[paulromano]

In our photon HDF5 files, we have Hartree-Fock Compton profiles (used for photon Doppler energy broadening) that are obtained from the classic paper by Biggs et al.. The subshell data for these profiles doesn't quite match what appears in the ENDF/B atomic relaxation sublibraries. For example, for carbon:

In [1]: import openmc.data

In [2]: c = openmc.data.IncidentPhoton.from_hdf5('C.h5')

In [3]: c.compton_profiles['num_electrons']
Out[3]: array([2., 2., 2.])

In [4]: c.compton_profiles['binding_energy']
Out[4]: array([291.01,  17.56,   8.99])

In [5]: c.atomic_relaxation.num_electrons
Out[5]: {'K': 2.0, 'L1': 2.0, 'L2': 0.67, 'L3': 1.33}

In [6]: c.atomic_relaxation.binding_energy
Out[6]: {'K': 291.01, 'L1': 17.56, 'L2': 8.99, 'L3': 8.98}

You can see that the Compton profile data lists three shells whereas the atomic relaxation data has four, adding an L3 shell. When we simulate Compton scattering, electrons are kicked out of the atom based on the Compton profile data but the vacancy is filled using the atomic relaxation data, so there is a potential mismatch. Right now, it doesn't crash because as far as I can tell in all cases the atomic relaxation data has more shells listed than the Compton profile data, but it's possible that in some cases we're filling the wrong vacancy.

@amandalund I don't know if we ever had a plan for how to fix this, since it's inherent in the data.