Limits on an Energy Dependence of the Speed of Light from a Flare of the Active Galaxy PKS 2155-304

In the past few decades, several models have predicted an energy dependence of the speed of light in the context of quantum gravity. For cosmological sources such as active galaxies, this minuscule effect can add up to measurable photon-energy dependent time lags. In this Letter a search for such time lags during the High Energy Stereoscopic System observations of the exceptional very high energy flare of the active galaxy PKS 2155-304 on 28 July 2006 is presented. Since no significant time lag is found, lower limits on the energy scale of speed of light modifications are derived.

Figure 1

Black points show the integral flux VHE light curves measured on July 28 from PKS 2155-304 by H.E.S.S. between 200–800 GeV (upper panel) and $>800\mathrm{GeV}$ (lower panel), binned in two-minute time intervals. The zero time point is set to MJD 53944.02. Gray points show the oversampled light curve, for which the 2 min. bins are shifted in units of 5 sec. The inlay in the upper panel illustrates this in a zoom, where the horizontal error bar shows the duration of the bin in the original light curve.

Figure 2

Left: MCCF of the light curves in Fig. 1. The black line shows the best fit of a Gaussian plus first-degree polynomial. The peak of the fitted function is located at ${\tau }_{\mathrm{peak}}=20\mathrm{s}$. Right: CCPD obtained from 10 000 simulated light curves. The shaded area shows the range of the CCPD for ${\tau }_{\mathrm{peak}}\le 0$, corresponding to 21% of the total area. The dotted line shows the position of ${\tau }_{\mathrm{peak}}$ from the left panel. The CCPD is slightly asymmetric, with a mean of 25 s and an rms of 28 s

Figure 3

Mean of the CCPD as a function of the dispersion injected in the H.E.S.S. data. The points have been shifted by the mean value of the CCPD of the original data shown in Fig. 2 to emphasize the relative time shifts. Each CCPD is derived from ten thousand simulated light curves. The error bars show the rms of distributions. The solid line shows the calibration curve used to transform time shifts into dispersion. For comparison, the dotted line shows the linear response function expected from the mean energy of the correlated light curves (see text).