Using artificial neural networks in searches for Lorentz invariance violation

TL;DR

This paper explores the use of artificial neural networks to simultaneously test for multiple Lorentz invariance violation effects in gamma-ray observations, marking a novel approach in the field.

Contribution

It introduces a machine learning framework to jointly analyze modified gamma-ray absorption and photon velocity effects, which were previously tested separately.

Findings

Initial results demonstrate the feasibility of neural networks in LIV testing.

The approach can potentially improve sensitivity to LIV effects.

Machine learning offers a new avenue for complex astrophysical tests.

Abstract

Lorentz invariance violation (LIV) in gamma rays can have multiple consequences, such as energy-dependent photon group velocity, photon instability, vacuum birefringence, and modified electromagnetic interaction. Depending on how LIV is introduced, several of these effects can occur simultaneously. Nevertheless, in experimental tests of LIV, each effect is tested separately and independently. For the first time, we are attempting to test for two effects in a single analysis: modified gamma-ray absorption and energy-dependent photon group velocity. In doing so, we are using artificial neural networks. In this contribution, we discuss our experiences with using machine learning for this purpose and present our very first results.