Subtracting compact binary foregrounds utilizing anisotropic statistic for third-generation gravitational-wave detectors

TL;DR

Problem: the CBC foreground dominates the 3G GW energy density, masking subdominant stochastic backgrounds. Approach: develop a multi-component likelihood using the anisotropic distribution of subthreshold BNS to constrain the isotropic cosmological GWB, with the angular power spectrum $C_cl$ and Fisher-matrix analysis. Findings: for a realistic network of three 3G detectors, anisotropy from shot noise after loud-signal subtraction is too faint to improve the isotropic $\Omega_{ m GW}^{(\rm c)}$; with 100x angular sensitivity, the improvement is approximately 17.5%. Implications: motivates joint likelihood approaches and highlights potential gains for future space-based detectors like DECIGO, where shot-noise anisotropy is stronger.

Abstract

The astrophysical foreground from compact-binary coalescence signals is expected to be a dominant part of total gravitational wave (GW) energy density in the frequency band of the third-generation detectors. The detection of any other subdominant stochastic GW background (GWB), especially a primordial GWB, will be disturbed by the astrophysical foreground, which needs to be cleaned for further studies of other stochastic GWB. Although previous studies have proposed several cleaning methods, the foreground from subthreshold binary neutron stars (BNS) has been a major obstacle to remove. In this paper, we propose the novel idea to acquire better estimation of the unresolved foreground, by using the information about its anisotropies. We simulate the BNS population and compute its angular power spectrum and shot noise. We find that the shot noise from BNS is too faint to observe after subtracting loud signals due to the limited angular resolution of the third-generation detectors. This justifies the approximation regarding the unresolved foreground as an isotropic component. We also discuss the angular resolution necessary to make our method valid for the foreground subtraction.

Figures (11)

• Figure 1: The complementary cumulative distribution function (CCDF) for SNR catalog of BNS. This CCDF is created from one realization of a BNS population with the model pop Nmed and Umed. It indicates how much fraction of events when we count the events from the highest to the given SNR.
• Figure 2: The violin plot of redshift distribution in each SNR bin. We used one out of 100 realizations of the pop Umed model.
• Figure 3: The energy density spectrum of CBC background for pop Umed model. The blue line shows the total energy density spectrum while other yellow, green, and red line represent the energy density of unresolved background with given SNR threshold.
• Figure 4: The energy density spectrum of CBC background for pop Umed model and Nmed models.
• Figure 5: Left: the ratio of $C_l$ to $C_0$ in a mode $l$ as a function of SNR threshold. Right: angular power spectra $C_l$ at the reference frequency $24.77\,{\rm Hz}$ as a function of SNR threshold for the population model, the pop Umed. We assume the signals whose SNR are above the threshold are subtracted perfectly from the data and only the weaker signals are counted for an anisotropic GWB.