Constraining weakly interacting massive particles with primordial black holes

Read the paper: arXiv:1905.01238.

Code authors: Adam Coogan, Daniele Gaggero and Bradley Kavanagh.

Reproducing results

The figures can be remade with the following commands:

• Figure 1: python plot_pbh_fraction.py
• Figure 2: python plot_frequentist_limits.py -plot_limits
• Figure 3: python plot_frequentist_limits.py -plot_ps_diff

Computing the point-source limits requires making tables containing p_gamma, the probability that an individual PBH is detectable by Fermi-LAT. The tables used for the analysis in the paper are contained in the directory data/p_gammas/. These may take a few hours to recompute. They can be regenerated with python generate_p_gamma_tables.py -test. With the -test flag, the p_gamma tables will be written to data/p_gammas/test/ instead of overwriting the precomputed tables. The script can also be used to generate p_gamma tables over different (m_dm, sv) grids.

All limit calculations require tabulated posteriors of f_pbh. These are provided in data/posteriors_f but they can be recomputed by running the script ComputePosteriors.sh. This script calls a number of scripts in src/ which calculate the posteriors from gravitational wave observations. Run python src/tabulate_posteriors_O3.py -h and python src/tabulate_posteriors_ET.py -h for more detailed usage of these scripts. Scripts for calculating posteriors from SKA will be added shortly.

bayesian branch

This branch contains code for performing a Bayesian analysis of the cross section limits.

• The notebook plot_bayesian_limits.ipynb generates Bayesian versions of figures 2 and 3 assuming a uniform prior on the cross section.
• submit_posterior_calcs_diff.py generates tables of the posterior over a grid of (m_dm, sv) points for the diffuse analysis. The posterior is very simple: we marginalize over f, assume the background model for the EGB perfectly matches observations, and simply penalize the extragalactic flux from PBHs for exceeding zero using a Gaussian likelihood for each bin.
• The analysis point-source analysis assumes that astrophysical sources can appear in the unassociated point source catalogue with an unknown rate, which we marginalize over assuming a Jeffreys prior. The script submit_posterior_calcs_ps.py generates tables of p_gamma and the posterior over an (m_dm, sv) grid.