Gravitational waves from eccentric binary neutron star mergers: Systematic biases and inadequacy of quasicircular templates

TL;DR

This work demonstrates that using quasicircular gravitational-wave templates to analyze eccentric binary neutron star mergers leads to significant biases in inferred parameters, including the chirp mass $\mathcal{M}$, mass ratio $q$, effective spin $\chi_{\rm eff}$, and tidal parameters $\tilde{\Lambda}$, with fractional SNR losses up to about ${\sim}25\%$ for $e_{0}$ up to $0.1$ and $\rho_{inj} \gtrsim 12$. By performing full Bayesian mock analyses with eccentric injections generated by TEOBResumS-Dalí and recovered with quasicircular templates (both nonprecessing and spin-precessing), the study reveals that biases intensify with eccentricity and that $\mathcal{M}_{\rm ecc}$ can explain small-$e$ biases, but breaks down for $e>0.05$ due to strong degeneracies with $q$. Spin precession does not mimic eccentricity, and biases in $\chi_{\rm eff}$ and tidal parameters persist, complicating source identification and neutron-star physics in high-SNR data. The results underscore the need for standardized, accurate eccentric waveforms for current and next-generation detectors, and discuss computational cost and potential speedups to enable practical analyses with higher-fidelity models including higher modes.

Abstract

The use of quasicircular waveforms in matched-filter analyses of signals from eccentric binary neutron star mergers can lead to biases in the source's parameter estimation. We demonstrate that significant biases can be present already for moderate eccentricities $e_{0} \gtrsim 0.05$ and signals detected by LIGO-Virgo-KAGRA with signal-to-noise ratio $\gtrsim 12$. We perform systematic Bayesian mock analyses of unequal-mass nonspinning binary neutron star signals up to eccentricities $e_0 \sim 0.1$ using quasicircular effective-one-body waveforms with spins. We find fractional signal-to-noise ratio losses up to tens of percent and up to 16$σ$ deviations in the inference of the chirp mass. The latter effect is sufficiently large to lead to an incorrect (and ambiguous) source identification. The inclusion of spin precession in the quasicircular waveform does not capture eccentricity effects. We conclude that high-precision observations with advanced (and next generation) detectors are likely to require standardized, accurate, and fast eccentric waveforms.

Figures (12)

• Figure 1: Injected waveforms for different values of eccentricity, as reported in the legend. The injections have been performed with TEOBResumS-Dalí model, with $\mathcal{M}=1.1977\,{\rm M_{\odot}}$, $q=1.5$, $\Lambda_1=400$, $\Lambda_2=600$, $D_L=40\,{\rm Mpc}$, and $\iota = 45^\circ$.
• Figure 2: Injected and reconstructed waveforms in the spin-aligned case for different values of eccentricity, as reported in the legend. The injections, plotted with solid lines, have been performed with TEOBResumS-Dalí model, with $\mathcal{M}=1.1977\,{\rm M_{\odot}}$, $q=1.5$, $\Lambda_1=400$, $\Lambda_2=600$, $D_L=40\,{\rm Mpc}$, and $\iota = 45^\circ$. The shaded regions are the $90\%$ credibility regions of the reconstructed spectrum and the dash-dotted line represents the amplitude spectral density of the GW170817 event.
• Figure 3: Recovered snr (top panel) and snr loss (bottom panel) as function of the injected snr for spin-aligned pe. Different colors refer to different eccentricities as reported in the legend. Round markers show the results of the median values of matched-filtered snr, star markers of the maximum values of the snr, both for the zero noise case, whereas triangular markers show the median values of the snr in the case of Gaussian noise.
• Figure 4: Posterior distribution of chirp masses and mass ratios for spin-aligned pe with zero noise, for two different luminosity distances $D_L=40\,{\rm Mpc}$ ($\rho_{\rm inj} = 31.7$) and $D_L=100\,{\rm Mpc}$ ($\rho_{\rm inj} = 12.7$). Injected values are identified by the black lines.
• Figure 5: Recovered $\mathcal{M}$ as function of the eccentricity for spin-aligned pe. Different colors refer to different injection distances as reported in the legend. The black dots show the leading order approximation for $\mathcal{M}_{\rm ecc}$ as per Eq. \ref{['eq:mchirp-ecc']}