Testing modified gravity with the eccentric neutron star--black hole merger GW200105

TL;DR

This paper tests modified gravity using the eccentric NSBH merger GW200105 by embedding leading-order BD, EdGB, and dCS corrections into an eccentric-precessing waveform model and performing Bayesian inference. It demonstrates that modeling eccentricity eliminates biases in GR tests and substantially tightens bounds on BD and EdGB couplings, with EdGB bound $\alpha^{1/2}_{EdGB} \\lesssim 2.38$ km and BD bound $\\omega_{BD} \\gtrsim 3.48$, while dCS remains weakly constrained due to low spin. The analysis combines the pyEFPE eccentric waveform with TIGER-style tests to assess GR deviations, revealing that apparent deviations in quasi-circular analyses arise from unmodeled eccentricity. The work highlights the importance of eccentricity in GW tests of gravity and projects much stronger future constraints with next-generation detectors that can access higher-eccentricity regimes. Overall, eccentric modeling provides a sharper, more reliable probe of gravity in the strong-field, dynamical regime.

Abstract

Direct detections of gravitational waves offer a unique opportunity to test gravity in the highly dynamical and strong field regime. Current tests are typically performed assuming signals from quasicircular binaries. However, the complex waveform morphology induced by orbital eccentricity can enhance our ability to probe gravity with greater precision. A recent analysis of the neutron star-black hole event GW200105 identified strong evidence for orbital eccentricity. We extend an eccentric-precessing waveform model to test alternative models with this signal by incorporating eccentric corrections induced by Brans-Dicke, Einstein-dilaton-Gauss-Bonnet, and dynamical Chern-Simons gravity at leading post-Newtonian order. We show that analyzing this event with a quasi-circular model leads to a false deviation from general relativity, while the inclusion of eccentricity improves the bounds on the models. Our analysis of GW200105 places tight constraints on Einstein-dilaton-Gauss-Bonnet gravity, $α^{1/2}_{\rm{EdGB}} \lesssim 2.38\,\text{km}$, and Brans-Dicke gravity, $ω_{\rm{BD}} \gtrsim 3.5$, while dynamical Chern-Simons gravity remains unconstrained due to the low spin content.

Figures (4)

• Figure 1: GW200105 results from an inspiral parametrized test of GR using TIGER with the IMRPhenomXPHM model, showing the marginalized posterior distributions of the parametrized deviation in the PN coefficients. We compare the results with a GW200105-like eccentric-precessing injection and recovery with TIGER and the same setup. The horizontal ticks show the 90% credible interval of the posterior. The GR value is marked by a grey dashed line. The full results are shown in Fig. \ref{['fig:violin_tiger_full']}.
• Figure 2: Posterior probability of $\alpha_{\text{EdGB}\xspace}^{1/2}$ (left panel) and $\omega_\text{BD}\xspace$ (right panel) obtained for GW200105 using eccentric (blue) and quasi-circular (orange) waveform models. The dashed-dot lines in left and right panels indicate the 90% upper bound on the $\alpha_{\text{EdGB}\xspace}^{1/2}$ and lower bound on $\omega_\text{BD}\xspace$, respectively. We compare the results with a GW200105-like eccentric injection recovered with a quasi-circular model (green).
• Figure 3: Posterior probability of $\alpha_{\text{dCS}\xspace}^{1/2}$ obtained for GW200105 using eccentric (blue) and quasi-circular (orange) waveform models. We compare the results with a GW200105-like eccentric injection and recovery with a quasi-circular model, as shown in green.
• Figure 4: GW200105 results from a parametrized test of GR using TIGER framework with IMRPhenomXPHM model, showing the marginalized posterior distributions of the inspiral deviation parameters. The horizontal bars show the 90% credible interval of the posterior. The GR value is marked by a red dot-dashed line. For all cases except for 1PN and 1.5PN, the GR is found outside the 90% credible interval. The consistency between the analysis of the GW200105 data and the eccentric-injection study suggests that the apparent deviations arise from biases induced by unmodeled eccentricity in the baseline waveform model.