Gravitational waves (GWs) have transformed our ability to probe the Universe, opening a new window for studying cosmology. Among their exciting applications is using them as “standard sirens” to measure the Hubble constant, H₀, a critical parameter in understanding the Universe’s expansion. The discovery of the Hubble tension—a discrepancy between local and early-Universe measurements of H₀—has driven researchers to refine these methods, seeking innovative approaches to improve precision and reliability.

In our recent work, we propose a novel approach to infer cosmology from GW observations, leveraging non-parametric methods to study the detector-frame mass distribution of compact binary mergers.

Key Innovations in Our Methodology

Non-Parametric Reconstruction

Traditional methods rely on specific population models, which can introduce biases and limit flexibility. Instead, we reconstruct the observed mass distribution in a non-parametric manner, making no assumptions about its form. This data-driven step enables a model-agnostic exploration of the population features encoded in the observed data.

A Two-Step Framework

1. Reconstruction of the Detector-Frame Distribution: Using hierarchical Dirichlet process Gaussian mixture models ((H)DPGMM), we create a representation of the mass distribution from the GW data.
2. Transformation to Source Frame: We compare this reconstructed distribution with predictions from intrinsic population and cosmological models, transforming the intrinsic distribution to the detector frame while incorporating selection effects.

Testing the Method

We validated our approach using mock data and real observations. By analyzing 70 binary black hole events from the LIGO-Virgo-KAGRA catalogs, we demonstrated that our method is computationally efficient and robust. Notably, it allows for quick testing of different models, reducing the computational burden as the number of GW detections grows.

Our findings show consistency with existing H₀ measurements, and the flexibility of our method makes it a promising tool for future population studies.

Implications and Future Directions

This work highlights the potential of non-parametric methods in cosmological inference. Beyond H₀, our framework could explore joint distributions involving other parameters like mass ratio and redshift. This would enhance our ability to test evolving mass population models and improve the precision of cosmological parameter estimation.

We are excited about the prospect of applying our method to the ever-growing GW datasets. As detectors improve, our approach will help untangle the complex interplay of astrophysical and cosmological phenomena encoded in GW signals.

Read the full paper here: arxiv:2410.23541

This blog post was written with the assistance of OpenAI’s ChatGPT and is based on the paper “Inferring cosmology from gravitational waves using non-parametric detector-frame mass distribution” by Thomas C. K. Ng, Stefano Rinaldi, and Otto A. Hannuksela (arXiv:2410.23541).