Golden Dark Sirens in GW Cosmology

• Golden dark sirens are gravitational-wave events from compact binaries, defined by sub-degree localization that uniquely associates them with a host galaxy.
• They utilize tidal effects in BNS mergers or statistical cross-matching with galaxy catalogs to infer redshift, supporting precision measurements of cosmological parameters.
• Advanced detector networks and deep spectroscopic surveys drive their ability to bypass EM biases, achieving sub-percent constraints on H0 and dark energy properties.

Golden dark sirens are gravitational-wave (GW) sources—compact binary mergers such as binary neutron stars (BNS), binary black holes (BBH), or neutron star–black hole binaries (NSBH)—detected without electromagnetic (EM) counterparts, whose localization precision is sufficient to enable unique or near-unique host galaxy identification via GW data and/or auxiliary galaxy catalogs. This rigorous localization, achieved through high signal-to-noise ratio (SNR), advanced detector networks, or GW-intrinsic redshift measurements (e.g. via tidal effects), transforms such events into standard sirens capable of delivering precise cosmological parameter inference, most notably the Hubble constant $H_0$, dark energy parameters, and tests of gravity. Golden dark sirens bypass many systematic biases associated with EM-based host identification and open a purely GW-driven route to precision cosmology (Jin et al., 2022, Muttoni et al., 2023, Dang et al., 25 Dec 2025, Alfradique et al., 2023, Cancino-Manríquez et al., 2024, Yang et al., 2021, Chen, 19 May 2025, Matos, 2024, Yu et al., 2023, Yang et al., 2022, Chen et al., 2023, Mukherjee et al., 2020, Mukherjee et al., 2022).

1. Definitions and Selection Criteria

Golden dark sirens are distinguished by localization volumes so small that only a single plausible host galaxy (or, in some strict definitions, a host group) remains after cross-matching with galaxy catalogs, or after direct GW-only redshift inference. Several papers formalize the criterion as requiring the 90–99% credible localization volume $V_\text{loc}$ to satisfy $V_\text{loc}<1/n_g$ for a mean galaxy density $n_g$ (often $0.01$–$0.02~\text{Mpc}^{-3}$), such that the expected number of host galaxies within $V_\text{loc}$ is close to unity (Yang et al., 2021, Yang et al., 2022, Yu et al., 2023). Alternatively, precise sky areas (e.g. $\Delta\Omega_{90}<0.1~\mathrm{deg}^2$) or single-galaxy-dominance in the host-weight posterior ($P_j>90\%$ for some galaxy $j$) serve as criteria (Dang et al., 25 Dec 2025). These events typically arise from high-SNR mergers observed by third-generation (3G) networks (e.g., Einstein Telescope, Cosmic Explorer) capable of sub-square-degree localization and percent-level distance errors (Muttoni et al., 2023, Jin et al., 2022, Matos, 2024).

Crucially, for BNS sources, tidal deformability measurements encode intrinsic mass–redshift information in the GW signal itself, enabling direct GW-only redshift recovery and unique host identification even absent a catalog (Jin et al., 2022, Yu et al., 2023). In BBH or NSBH cases, deep and complete galaxy catalogs are essential for host assignment (Dang et al., 25 Dec 2025, Alfradique et al., 2023, Chen et al., 2023).

Definition criterion	Typical threshold	Host ambiguity
Localization volume ($V_\text{loc}$)	$<1/n_g$, e.g., $<100~\text{Mpc}^3$	Unique
Sky area ($\Delta\Omega_{90}$)	$<0.1~\mathrm{deg}^2$	Unique/near-unique
Probability ($P_j$ for host $j$)	$>90\%$	Dominant single host

2. GW Data Analysis: Tidal Effects and Redshift Inference

For binary neutron star coalescences, measurement of tidal deformability—the dimensionless parameter $\Lambda_i$ determined by the neutron star's Love number, radius, and mass—breaks the degeneracy between intrinsic mass $m_i$ and redshift $z$. The GW phase is given by

$$\tilde{h}(f)\propto f^{-7/6}\exp\left[i\left(2\pi f t_c -\frac{\pi}{4} -2\psi_c +2\Psi_{\rm pm}(f) +\Delta\Psi_{\rm tidal}(f)\right)\right],$$

with $\Delta\Psi_{\rm tidal}(f)$ entering at 5PN order. Tidal effects allow simultaneous inference of $z$ and intrinsic mass from GW data alone, contingent on the neutron star equation of state (EoS) being known (Jin et al., 2022, Yu et al., 2023).

For BBH and NSBH events lacking tidal signatures, redshift is statistically assigned by marginalizing over galaxies within the localization volume, with host probability weights allocated based on sky location, stellar mass, or other astrophysical priors. Full Bayesian posteriors for $H_0$ and other cosmological parameters are constructed as weighted mixtures over potential host galaxies (Muttoni et al., 2023, Dang et al., 25 Dec 2025, Alfradique et al., 2023).

Machine learning methods such as Mixture Density Networks (MDN) supply photometric redshift PDFs in large catalogs (e.g., DELVE, HETDEX), and spectroscopic follow-up can further enhance redshift accuracy and completeness (Dang et al., 25 Dec 2025, Alfradique et al., 2023).

3. Statistical Frameworks and Parameter Estimation

Golden dark siren analyses usually employ hierarchical Bayesian inference combining GW data, host-galaxy redshift information, and selection function modeling. In the GW–only approach (BNS tidal effect), the joint likelihood for cosmological parameters $\Omega$ reads

$$\mathcal{L}(\{d_L^{\rm obs},z^{\rm obs}\}\mid \Omega) = \prod_{i=1}^{N} \int dz~p(d_L^{\rm obs,i}|d_L(z,\Omega))~p(z^{\rm obs,i}|z),$$

with Gaussian error models from Fisher-matrix or posterior sampling (Jin et al., 2022, Yu et al., 2023). For catalog-based analyses,

$$p(H_0|\{x^{(i)}\},\{z_j^{(i)},M_j^{(i)}\}) \propto \prod_i \sum_j L_{GW}(x^{(i)}| \Omega_j^{(i)}, d_L(z_j^{(i)},H_0)) \cdot p_{host}(j),$$

with marginalization over selection effects and host weights (Dang et al., 25 Dec 2025, Muttoni et al., 2023). Tomographic cross-correlation techniques treat GW and galaxy samples as biased tracers of underlying matter, leveraging angular clustering and luminosity-distance tomography to anchor the $d_L(z)$ relation and estimate cosmological parameters (Sala et al., 9 Oct 2025, Mukherjee et al., 2022).

4. Comparative Performance: Precision and Cosmological Implications

Golden dark sirens, especially with 3G detector networks and complete galaxy catalogs, deliver sub-percent-level constraints on $H_0$. Key performance metrics from recent literature include:

• $0.15\%$ fractional error on $H_0$ with $O(10^6)$ BNS events with tidal measurements and perfectly known EoS in three years (Jin et al., 2022).
• $0.8\%$ 90\% confidence interval on $H_0$ from $O(278)$ SNR$>300$ BBH/BNS events per year with ET+CE network (Muttoni et al., 2023).
• $0.2\%$–$0.3\%$ accuracy in $H_0$ from $\sim1\,500$ golden BNS/NSBH dark sirens in five years with CE+ET (Yu et al., 2023).
• 1–2\% error with 5–10 golden BBH/BNS from space-borne atom interferometric detectors (AEDGE) (Yang et al., 2021, Yang et al., 2022).
• 1–2\% precision on $w$ (constant dark energy EoS); $2.04\%$ on $w_0$, $0.13$ on $w_a$ in the CPL parameterization (Jin et al., 2022).
• 0.7% error on $H_0$ via tomographic GW–galaxy cross-correlation in 3G era (Sala et al., 9 Oct 2025).

By contrast, “bright” sirens with EM counterparts are rate-limited ($\lesssim100$ events in 3G era) and constrain $H_0$ at the $1$–$2\%$ level per five years (Yu et al., 2023, Jin et al., 2022, Dang et al., 25 Dec 2025). Silver dark sirens ($\lesssim1~\mathrm{deg}^2$ localization, tens of host candidates) are more common but achieve only few-percent precision (Dang et al., 25 Dec 2025). Space-based GW bright sirens (LISA/ Taiji) yield $5$–$10\%$ $w$ constraints limited by sample size (Jin et al., 2022).

5. Systematic Uncertainties and Methodological Challenges

Key systematics in golden dark siren cosmology include:

• Tidal model uncertainties: e.g., linear fitting for $\lambda(m)$ (BNS EoS) is accurate to $\sim10\%$; errors propagate into $z$ inference and selection fractions (Yu et al., 2023, Jin et al., 2022).
• Galaxy catalog incompleteness: host missing from catalog introduces multi-$\sigma$ bias; completeness corrections (e.g., $f_{comp}$) must be applied (Yu et al., 2023, Dang et al., 25 Dec 2025).
• Weak lensing and peculiar velocities: at low $z$, peculiar velocity dispersion ($v_{\text{rms}}\sim500~\text{km/s}$) significantly affects $d_L(z)$, especially for $z < 0.2$ (Yang et al., 2021, Yang et al., 2022).
• Detector calibration: amplitude and phase calibration must be maintained at $\lesssim1\%$ levels across bands (Muttoni et al., 2023, Jin et al., 2022).
• Machine learning photo-$z$ calibration: biases in galaxy redshift PDFs are non-negligible; spectroscopic follow-up is essential for sub-percent $H_0$ inference (Dang et al., 25 Dec 2025, Alfradique et al., 2023).
• Selection effects: modeling of detection probability, SNR thresholds, and host weighting is required to ensure unbiased cosmological posteriors (Dang et al., 25 Dec 2025, Muttoni et al., 2023, Alfradique et al., 2023).

6. Extensions: Golden Dark Sirens in Fundamental Physics and Joint Probes

Golden dark sirens enable robust tests of dark energy, cosmic anisotropy, and gravitational physics beyond general relativity:

• As an independent $H_0$ anchor, golden dark sirens offer decisive power to arbitrate the current $\sim5\sigma$ Hubble tension between local measurements (SH0ES) and Planck CMB inferences (Jin et al., 2022, Yang et al., 2021, Muttoni et al., 2023, Chen, 19 May 2025).
• Joint dipole $g$ and $H_0$ measurements exploit the directional modulation of luminosity distance; sub–$10^{-3}$ constraints are achievable with tens of golden dark sirens in next-generation networks (Chen, 19 May 2025).
• GW-only measurements decouple $d_L^{GW}$ from $d_L^{EM}$, enabling new constraints on GW propagation (friction term $\Xi(z)$, Horndeski parameter $c_M$, extra-dimensional $D$) and thereby probing modifications to gravity at cosmological scales (Mukherjee et al., 2020, Chen et al., 2023, Matos, 2024).
• Combination with electromagnetic probes (galaxy clustering, SNe Ia, BAO) breaks degeneracies in cosmological parameters, enhances robustness, and allows sub-percent-level joint constraints on $w_0$, $w_a$, and fundamental physics (Sala et al., 9 Oct 2025, Matos, 2024, Muttoni et al., 2023).

7. Prospects and Future Directions

Observation rates, localization capabilities, and catalog completeness for golden dark siren detection will increase markedly with next-generation detector networks (ET, CE) and deep spectroscopic surveys (HETDEX, DESI, Euclid, LSST). Projected yields are:

• $O(10^5-10^6)$ BNS events with tidal effects per 3 years (Jin et al., 2022).
• $\sim278$ golden BBH/BNS events per year with ET+2CE (Muttoni et al., 2023).
• $3$–$4$ golden and $>100$ silver sirens per year at z < 0.2 for LIGO-A#+India (Dang et al., 25 Dec 2025).
• Tens to hundreds of golden dark sirens at z < 0.1 per decade with ET+CE+CE (Chen, 19 May 2025).
• $\mathcal{O}(10^2-10^3)$ golden dark sirens within five years, enabling $\sigma(H_0)\sim$0.3% (Matos, 2024, Sala et al., 9 Oct 2025).

As the GW and galaxy survey infrastructure matures, golden dark sirens will routinely yield sub-percent cosmological parameter constraints, facilitate model-independent cosmology, and enable precision tests of the propagation of gravity. These sources occupy a critical niche between bright sirens and conventional galaxy redshift surveys, offering robust cross-validation and extending cosmological reach into purely gravitational domains.