Binary vision: The merging black hole binary mass distribution via iterative density estimation

Binary black hole (BBH) systems detected via gravitational-wave (GW) emission are a recently opened astrophysical frontier with many unknowns and uncertainties. Accurate reconstruction of the binary distribution with as few assumptions as possible is desirable for inference on formation channels and environments. Most population analyses have, though, assumed a power law in binary mass ratio $q$, and/or assumed a universal $q$ distribution regardless of primary mass. Kernel density estimation (KDE)-based methods allow us to dispense with such assumptions and directly estimate the joint binary mass distribution. We deploy a self-consistent iterative method to estimate this full BBH mass distribution, finding local maxima in primary mass consistent with previous investigations and a secondary mass distribution with a partly independent structure, inconsistent with both power laws and with a constant function of $q$. We find a weaker preference for near-equal mass binaries than in most previous investigations; instead, the secondary mass has its own “spectral lines” at slightly lower values than the primary, and we observe an anti-correlation between primary and secondary masses around the ~$10M_{\odot }$ peak.