Functional inference on deviations from General Relativity

TL;DR

This paper introduces GRANITA, a flexible, data-driven framework using Gaussian process regression to detect and quantify parameter-dependent deviations from general relativity in gravitational wave data, including stochastic effects.

Contribution

The paper presents GRANITA, a novel, non-perturbative, theory-agnostic method for analyzing deviations from GR in gravitational wave signals, applicable to real and simulated data.

Findings

Successfully detects waveform deviations across parameter space.

Identifies stochastic, non-deterministic deviations.

Applicable to real gravitational wave events.

Abstract

Extensions of general relativity often predict modifications to gravitational waveform morphology that depend functionally on source parameters, such as the masses and spins of coalescing black holes. However, current analyses of strong-field gravity lack robust, data-driven methods to infer such functional dependencies. In this work, we introduce GRANITA, a non-perturbative, theory-agnostic framework to characterize parameter-dependent deviations from general relativity using Gaussian process regression. Leveraging the flexibility of this method, we analyze both simulated data and real events from the LIGO-Virgo-KAGRA public catalog. We demonstrate the ability of our approach to detect and quantify waveform deviations across the parameter space. Furthermore, we show that the method can identify stochastic (non-deterministic) deviations, potentially arising from environmental effects or subdominant unmodeled physics. As gravitational-wave tests of strong gravity advance in precision, our framework provides a principled approach to constrain modified gravity and to mitigate contamination from astrophysical or instrumental systematics.