Improvements to utils/neutrino_astronomy.py

[mlincett]

Some time ago, I had started a review of this module, in particular I would like to tidy up the "Total Flux" and "Flux from the nearest source" calculations.

In the following snippet:

flarestack/flarestack/utils/neutrino_astronomy.py

Lines 81 to 84 in 56810cf

	# Energy requires a 1/(1+z) factor
	zfactor = find_zfactor(lumdist)
	etot = (si * area * e_integral).to(u.erg / u.s) * zfactor

the energy-integrated flux at Earth is multiplied by 1+z as a consequence of the energy range on which the integral takes place being redshifted w.r.t. the rest frame of the source.

From my calculations this should be (1+z)^(1-gamma) that for gamma = 2 should give 1/(1+z). Note that the comment mentions a 1/(1+z) factor while code multiplies by (1+z). For sure this should depend on EnergyPDF and gamma.

Probably my calculations should be double-checked as well, I am attaching here the file that @JannisNe and @robertdstein have already received by mail: nu_astro.pdf

[mlincett]

After a mail exchange with @robertdstein , probably the most convenient option is to remove the "flux from nearest source" calculation. This in light of the fact that:

• the feature requires a revision in any case (at least it should depend on EnergyPDF);
• it has no further use in flarestack and it is more of a convenience for the user;
• calculating the nearest source only is kind of arbitrary (the nearest it's not necessarily the one with the strongest flux).

We can leave luminosity calculation as an "exercise to the reader" and simplify the results dictionary.

[JannisNe]

Hi sorry for the long wait! I went through it and I think you are right. In that case, your and Robert's suggestion sounds good!
Thanks for taking care!

[mlincett]

Thanks for the feedback @JannisNe and @robertdstein .

I drafted PR #242 with the idea of cleaning up the code and removing the controversial calculations.

Unfortunately, it seems that the field Mean Luminosity (erg/s) in the output was used in several past analyses:

flarestack/analyses/benchmarks/benchmark_blazar_flare_search.py:        e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ztf/test_sens_dep.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ztf/test_sensitivity_depth_time.py:                    e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ztf/test_sensitivity_depth.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/tde/compare_fitting_weights.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/tde/calculate_jetted_model_limit.py:e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/tde/compare_spectral_indices.py:                        e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/tde/compare_cluster_search_to_time_integration.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/tde/compare_spectral_indices_individual.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/agn_cores/loop_over_catalogues.py:                        e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ccsn/stasik_2017/calculate_sensitivity_reproduce_stasik_decay.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ccsn/stasik_2017/calculate_sensitivity_reproduce_stasik.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ccsn/necker_2019/decay_sensitivity.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/ccsn/necker_2019/box_sensitivity.py:                e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/fermi/gb6_fermi_blazar_flare_search.py:        e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/txs_0506_056/compare_txs_spectral_indices.py:                    e_key = "Mean Luminosity (erg/s)"
flarestack/analyses/txs_0506_056/INTRO1.py:print(astro_sens["Mean Luminosity (erg/s)"], "erg/s")
flarestack/core/unblinding.py:                        astro_dict["Mean Luminosity (erg/s)"] * inj_time

so I am not sure it is wise to discontinue it right away.

Please advise!

[JannisNe]

In my opinion, we do not have to require past analyses to run with the most recent version of the code. They all mention their respective flarestack version with which we can reproduce them. So I'm in favour of getting rid of it.

[mlincett]

Alright! Will do the last cleanups and proceed with the merge.

[mlincett]

The removed functionality touches the calculate_upper_limits() in the core.unblinding module.

It seems though that calculate_upper_limits() is not covered by unit tests (see #243) nor it is invoked in any of the past analyses (at least the ones stored in the repository).

[mlincett]

The outputs based on "Mean Luminosity (erg/s)" have been removed from calculate_upper_limits().