Search for Peak Structures in the Stochastic Gravitational-Wave Background in LIGO-Virgo-KAGRA O1-O4a Datasets

TL;DR

The paper develops and applies a Bayesian, model‑independent framework to search for stochastic gravitational‑wave backgrounds with double‑peaked spectral morphologies in LVK data from O1–O4a. By parameterizing the cosmological contribution as a sum of two normalized broken power laws and combining it with the standard CBC background, the analysis constrains peak amplitudes and inter‑peak slopes using cross‑correlation data and the overlap reduction function. No statistically significant double‑peak signal is detected, but the work places quantitative limits on the morphology of potential multi‑peak SGWBs and demonstrates the detector network’s sensitivity to non‑power‑law spectral shapes within the LVK band. This establishes a methodological foundation for future targeted searches in ongoing and next‑generation gravitational‑wave experiments, enabling a more complete exploration of high‑energy cosmology and early‑Universe physics.

Abstract

We present a dedicated search for gravitational-wave backgrounds with nontrivial peak structures using data from the first three and the initial part of the fourth observing runs of the LIGO-Virgo-KAGRA network. The analysis is motivated by a variety of early-Universe models characterized by signals with multiple peaks. We introduce a model independent parameterization of double-peaked spectra based on the superposition of two normalized broken power laws and perform a Bayesian inference study using the LIGO-Virgo-KAGRA isotropic cross-correlation data. While no statistically significant evidence for a multi-peak background is found, the analysis provides constraints on the inter-peak slopes in correlation with the signal amplitude. These results exhibit LIGO-Virgo-KAGRA's ability to probe signals beyond a single peak structure and establish a foundation for future targeted searches for nontrivial gravitational waves background spectral shapes in future observing runs and the advanced detector era.

Figures (7)

• Figure 1: Example of the double-peaked signal, where both peaks are located in the sensitivity range and above the power-law integrated sensitivity curve of the detector. The blue dashed lines indicate the peaks' frequencies. The horizontal purple dashed line marked by $\Omega_{\Delta N_{eff}}$ indicates the constraints coming from CMB and BAO analyses Caprini_2018.
• Figure 2: Example of the double-peaked signal where both peaks are located outside the sensitivity range of the detector, but the valley in between the peaks is above the power-law integrated sensitivity curve. The blue dashed lines indicate the peaks' frequencies. The horizontal purple dashed line marked by $\Omega_{\Delta N_{eff}}$ indicates the constraints coming from CMB and BAO analyses Caprini_2018.
• Figure 3: Posterior distributions obtained when fixing the location of the second peak $f_{*2} = 300$ Hz. As the frequency $f_{*1}$ varies, a region in the $n_2$ vs. $\Omega_{*1}$ parameter space can be excluded.
• Figure 4: Posterior distributions obtained when fixing the location of the first peak $f_{*1} = 100$ Hz. As the frequency $f_{*2}$ varies, a region in the $n_3$ vs. $\Omega_{*2}$ parameter space can be excluded.
• Figure A1: Corner plot results for the Bayesian search using the parameters described above.