Simulations of black-hole binaries with unequal masses or non-precessing spins: accuracy, physical properties, and comparison with post-Newtonian results

TL;DR

This paper presents accurate numerical gravitational waveforms for black-hole binaries with unequal masses or non-precessing spins, compares them with post-Newtonian predictions, and analyzes the physical properties of the merger.

Contribution

It provides new numerical waveforms for specific black-hole binary configurations and compares their accuracy with various post-Newtonian approximants.

Findings

3.5PN TaylorT4 approximant is most accurate for nonspinning binaries.

TaylorT1 with 2.5PN spin terms performs best for equal-mass spinning binaries.

Including high-order amplitude corrections increases the amplitude by 2-4%.

Abstract

We present gravitational waveforms for the last orbits and merger of black-hole-binary (BBH) systems along two branches of the BBH parameter space: equal-mass binaries with equal non-precessing spins, and nonspinning unequal-mass binaries. The waveforms are calculated from numerical solutions of Einstein's equations for black-hole binaries that complete between six and ten orbits before merger. Along the equal-mass spinning branch, the spin parameter of each BH is $\chi_i = S_i/M_i^2 \in [-0.85,0.85]$, and along the unequal-mass branch the mass ratio is $q =M_2/M_1 \in [1,4]$. We discuss the construction of low-eccentricity puncture initial data for these cases, the properties of the final merged BH, and compare the last 8-10 GW cycles up to $M\omega = 0.1$ with the phase and amplitude predicted by standard post-Newtonian (PN) approximants. As in previous studies, we find that the phase from the 3.5PN TaylorT4 approximant is most accurate for nonspinning binaries. For equal-mass spinning binaries the 3.5PN TaylorT1 approximant (including spin terms up to only 2.5PN order) gives the most robust performance, but it is possible to treat TaylorT4 in such a way that it gives the best accuracy for spins $\chi_i > -0.75$. When high-order amplitude corrections are included, the PN amplitude of the $(\ell=2,m=\pm2)$ modes is larger than the NR amplitude by between 2-4%.