Constraints on Lorentz Invariance Violation from Fermi-Large Area Telescope Observations of Gamma-Ray Bursts

TL;DR

This study uses Fermi-LAT observations of gamma-ray bursts to set stringent limits on Lorentz invariance violation by constraining the energy dependence of the speed of light, thereby testing quantum gravity theories.

Contribution

The paper presents the first robust constraints on vacuum dispersion due to LIV from multiple analysis techniques applied to GRB data, improving previous limits significantly.

Findings

Limits on LIV energy scale exceed Planck energy for linear dispersion.

Constraints disfavor models with LIV effects below Planck scale.

Improved bounds by a factor of two over previous Fermi and H.E.S.S. results.

Abstract

We analyze the MeV/GeV emission from four bright Gamma-Ray Bursts (GRBs) observed by the Fermi-Large Area Telescope to produce robust, stringent constraints on a dependence of the speed of light in vacuo on the photon energy (vacuum dispersion), a form of Lorentz invariance violation (LIV) allowed by some Quantum Gravity (QG) theories. First, we use three different and complementary techniques to constrain the total degree of dispersion observed in the data. Additionally, using a maximally conservative set of assumptions on possible source-intrinsic spectral-evolution effects, we constrain any vacuum dispersion solely attributed to LIV. We then derive limits on the "QG energy scale" (the energy scale that LIV-inducing QG effects become important, E_QG) and the coefficients of the Standard Model Extension. For the subluminal case (where high energy photons propagate more slowly than lower energy photons) and without taking into account any source-intrinsic dispersion, our most stringent limits (at 95% CL) are obtained from GRB090510 and are E_{QG,1}>7.6 times the Planck energy (E_Pl) and E_{QG,2}>1.3 x 10^11 GeV for linear and quadratic leading order LIV-induced vacuum dispersion, respectively. These limits improve the latest constraints by Fermi and H.E.S.S. by a factor of ~2. Our results disfavor any class of models requiring E_{QG,1} \lesssim E_Pl.