Leveraging NRTidalv3 to develop gravitational waveform models with higher-order modes for binary neutron star systems

TL;DR

This work extends the NRTidalv3 tidal model to include higher-order mode corrections for binary neutron star systems and couples it to leading BBH baselines (e.g., IMRPhenomXHM, SEOBNRv5HM_ROM), enabling more accurate HM-inclusive waveforms. The HM-NRTidalv3 models are validated against numerical-relativity simulations in both time and frequency domains, showing improved agreement over HM-free counterparts, with manageable computational cost. Parameter-estimation studies demonstrate robust recovery for comparable-mass systems but reveal degeneracies among mass and spin for high-mass-ratio cases, mitigated by incorporating precession. Reanalysis of GW170817 with HM models yields results consistent with prior analyses, highlighting the practical relevance for future detections and the potential for NSBH modeling and broader EOS exploration.

Abstract

Accurate and reliable gravitational waveform models are crucial in determining the properties of compact binary mergers. In particular, next-generation gravitational-wave detectors will require more accurate waveforms to avoid biases in the analysis. In this work, we extend the recent NRTidalv3 model to account for higher-mode corrections in the tidal phase contributions for binary neutron star systems. The higher-mode, multipolar NRTidalv3 model is then attached to several binary-black-hole baselines, such as the phenomenological IMRPhenomXHM and IMRPhenomXPHM models, and the effective-one-body-based model SEOBNRv5HM_ROM. We test the performance and validity of the newly developed models by comparing them with numerical-relativity simulations and other tidal models. Finally, we employ them in parameter estimation analyses on simulated signals from both comparable-mass and high-mass-ratio systems, as well as on the gravitational-wave event GW170817, for which we find consistent results with respect to previous analyses.

Figures (15)

• Figure 1: Runtime boxplots of the different HM tidal models and their respective $(2,|2|)$-mode counterparts for the aligned spin and precessing models. Horizontal lines between the boxes indicate the median runtimes, whose values can be found in Table \ref{['table: approximants_table']}. For each model, the box contains the middle 50% of the distribution (between the first quartile $Q_1$ or the 25th percentile and the third quartile $Q_3$ or the 75th percentile), while the height of the box itself is the inter-quartile range $\Delta Q = Q_3 - Q_1$. The horizontal line in the middle of the box is the median of the distribution, while the lower and upper whiskers that extend from the box denote $Q_1 - 1.5\Delta Q$ and $Q_3 + 1.5\Delta Q$, respectively. Values higher or lower than the whiskers are denoted with individual points.
• Figure 2: Time-domain dephasing comparisons for the BAM:0130 and BAM:0131 waveforms. For each NR waveform, the upper panel shows the real part of the gravitational wave strain as a function of the retarded time, while the bottom panel shows the phase difference between the waveform model and the NR waveform. The gray band in the bottom panel represents the phase difference between the Richardson-extrapolated phase and the highest-resolution phase of the simulation. For each comparison, we denote the alignment windows by the dashed gray lines, and merger by the solid black line at $t/M = 0$.
• Figure 3: Mismatches between the waveform models and the NR waveforms used in the time-domain comparisons (see Table \ref{['table: bns_td_configs']}). For each model, the marker denotes the mean mismatch over 18 combinations of the inclination $\iota$, reference phase $\phi_0$, and polarization angle $\psi_p$. The upper and lower whiskers denote the maximum and minimum mismatches, respectively. For comparison, for every HM model we also show its $(2,|2|)$-mode counterpart, with the same marker and color but increased transparency in the plot. A smaller variability in the mismatches is observed for the $(2,|2|)$-mode models, and for the equal-mass simulations BAM:0062 and BAM:0101 due to the suppressed HM content.
• Figure 4: Mismatch comparisons of between the different HM tidal waveform models. Each subfigure contains three two-dimensional, density scatter plots of the masses $M_{A,B}$, tidal deformabilities $\Lambda_{A,B}$, and aligned spin components $\chi_{A,B}$, where the log-mismatch $\log_{10}\bar{F}$ is represented by the color bar, with limits set to $\log_{10}\bar{F} \in [-5, -1]$. We note the smaller axis limits for the comparison with SEOBNRv5THM, due to the narrower parameter space used here.
• Figure 5: Mismatch comparisons per mode between XHM_NRT3 and EOBv5HM_NRT3. Each common mode is represented by a subfigure that contains three two-dimensional, density scatter plots of the masses $M_{A,B}$, tidal deformabilities $\Lambda_{A,B}$, and aligned spin components $\chi_{A,B}$, where the log-mismatch $\log_{10}\bar{F}$ is represented by the color bar, with limits set to $\log_{10}\bar{F} \in [-5, -1]$.