added bin_log_width to EnergyFilter() to help with plotting

[shimwell]

Just wondering if we can add a few lines to the EnergyFilter class that allows one to conveniently access the log bin width.

This can useful when plotting spectra

import matplotlib.pyplot as plt

results = openmc.StatePoint(results_filename)
cell_tally = results.get_tally(name='cell_spectra_tally')
energy_filter = cell_tally.find_filter(filter_type=openmc.filter.EnergyFilter)

bin_boundaries = energy_filter.lethargy_bin_width
flux = cell_tally.mean.flatten()
norm_flux = flux / bin_boundaries

plt.figure()
plt.step(energy_filter.values[:-1], norm_flux)
plt.xscale('log')
plt.yscale('log')
plt.ylabel('Flux per unit lethargy [n/cm2-s]')
plt.show()

[eepeterson]

If you want to add a function like this I'd recommend just a property called lethargy with return np.log(self.bins[:, 1]/self.bins[:, 0])

[paulromano]

@eepeterson I think lethargy is usually defined relative to some reference energy, i.e. something like np.log10(E_reference / energies) where E_reference is a scalar and energies is an array. That is slightly different than what is proposed (ratio of successive energies).

[eepeterson]

@eepeterson I think lethargy is usually defined relative to some reference energy, i.e. something like np.log10(E_reference / energies) where E_reference is a scalar and energies is an array. That is slightly different than what is proposed (ratio of successive energies).

You're right, it could be called bin_lethargy to be clearer, but I think the point is to be able to plot spectra with different group structures on the same plot and have them be comparable. Alternatively, you could just use the highest energy bin boundary as the reference and take the difference of the actual lethargies of the energy bin boundaries.

[shimwell]

Thanks for the feedback, I've renamed the method to bin_lethargy

We could add another method if people want ?

def lethargy(self, E_reference):
    return np.log10(E_reference / self.bins)

[paulromano]

In addition to the line comments, this PR also needs a quick unit test.

[shimwell]

I would be keen to get rid of the need for users to do np.unique(energy_filter.bins)[:-1] when calling plt.step() as well.

What do people think of adding a def bin_edges(self) property that returns np.unique(self.bins) to make things even easier for the user or is that a step too far 🥁

[paulromano]

@shimwell The values property already gives just an array of the energies. I think that's what you're looking for?

[shimwell]

Oh yes, that works, thanks. I shall update the example at the top of this PR.

[paulromano]

Thanks!