Accelerating gravitational-wave parameterized tests of General Relativity using a multiband decomposition of likelihood

TL;DR

This paper introduces a multiband likelihood decomposition technique that significantly accelerates parameterized tests of General Relativity using gravitational wave data, reducing computational time by an order of magnitude.

Contribution

The authors adapt and validate a multiband likelihood method to speed up parameterized tests of GR in gravitational wave analysis, making such tests more computationally feasible.

Findings

Speeds up parameterized tests by a factor of about 10 for BNS signals

Maintains accuracy with simulated signals and real data

Reduces analysis time from weeks-years to days or hours

Abstract

The detection of gravitational waves from compact binary coalescence (CBC) has allowed us to probe the strong-field dynamics of General Relativity (GR). Among various tests performed by the LIGO-Virgo-KAGRA collaboration are parameterized tests, where parameterized modifications to GR waveforms are introduced and constrained. This analysis typically requires the generation of more than millions of computationally expensive waveforms. The computational cost is higher for a longer signal, and current analyses take weeks-years to complete for a binary neutron star (BNS) signal. In this work, we present a technique to accelerate the parameterized tests using a multiband decomposition of likelihood, which was originally proposed to accelerate parameter estimation analyses of CBC signals assuming GR by one of the authors. We show that our technique speeds up the parameterized tests of a 1.4 Msun-1.4 Msun BNS signal by a factor of O(10) for a low-frequency cutoff of 20 Hz. We also verify the accuracy of our method using simulated signals and real data.