1: \documentclass{elsart}

2: \journal{Phys. Lett. B}

3: \date{2008 August 27}

4: \usepackage{graphicx}

5: \usepackage{cite}

6: \usepackage{txfonts}

7: 

8: \begin{document}

9: 

10: \begin{frontmatter}

11: 

12: \begin{flushright}

13: {\small {CERN-PH-TH/2007-137, MPP-2007-112, {\tt arXiv:0708.2889v3 [astro-ph]}}}

14: \end{flushright}

15: \vspace*{-.2cm}

16: 

17: \title{Probing quantum gravity using photons from a flare

18:        of the active galactic nucleus Markarian 501 observed by

19:        the MAGIC telescope}

20: 

21: \collab{MAGIC Collaboration}

22: \author[a]{J.~Albert},

23: \author[b]{E.~Aliu},

24: \author[c]{H.~Anderhub},

25: \author[d]{L.~A.~Antonelli},

26: \author[e]{P.~Antoranz},

27: \author[f]{M.~Backes},

28: \author[g]{C.~Baixeras},

29: \author[e]{J.~A.~Barrio},

30: \author[h]{H.~Bartko},

31: \author[i]{D.~Bastieri},

32: \author[f]{J.~K.~Becker},

33: \author[j]{W.~Bednarek},

34: \author[a]{K.~Berger},

35: \author[k]{E.~Bernardini},

36: \author[i]{C.~Bigongiari},

37: \author[c]{A.~Biland},

38: \author[h,i]{R.~K.~Bock},

39: \author[l]{G.~Bonnoli},

40: \author[m]{P.~Bordas},

41: \author[m]{V.~Bosch-Ramon},

42: \author[a]{T.~Bretz},

43: \author[c]{I.~Britvitch},

44: \author[e]{M.~Camara},

45: \author[h]{E.~Carmona},

46: \author[n]{A.~Chilingarian},

47: \author[c]{S.~Commichau},

48: \author[e]{J.~L.~Contreras},

49: \author[b]{J.~Cortina},

50: \author[o,p]{M.~T.~Costado},

51: \author[d]{S.~Covino},

52: \author[f]{V.~Curtef},

53: \author[i]{F.~Dazzi},

54: \author[q]{A.~De Angelis},

55: \author[r]{E.~De Cea del Pozo},

56: \author[o]{C.~Delgado Mendez},

57: \author[e]{R.~de los Reyes},

58: \author[q]{B.~De Lotto},

59: \author[q]{M.~De Maria},

60: \author[q]{F.~De Sabata},

61: \author[s]{A.~Dominguez},

62: \author[a]{D.~Dorner},

63: \author[i]{M.~Doro},

64: \author[b]{M.~Errando},

65: \author[l]{M.~Fagiolini},

66: \author[t]{D.~Ferenc},

67: \author[b]{E.~Fern\'andez},

68: \author[b]{R.~Firpo},

69: \author[e]{M.~V.~Fonseca},

70: \author[g]{L.~Font},

71: \author[h]{N.~Galante},

72: \author[o,p]{R.~J.~Garc\'{\i}a L\'opez},

73: \author[h]{M.~Garczarczyk},

74: \author[o]{M.~Gaug},

75: \author[h]{F.~Goebel},

76: \author[h]{M.~Hayashida},

77: \author[o,p]{A.~Herrero},

78: \author[a]{D.~H\"ohne},

79: \author[h]{J.~Hose},

80: \author[h]{C.~C.~Hsu},

81: \author[a]{S.~Huber},

82: \author[h]{T.~Jogler},

83: \author[c]{D.~Kranich},

84: \author[d]{A.~La Barbera},

85: \author[t]{A.~Laille},

86: \author[l]{E.~Leonardo},

87: \author[u]{E.~Lindfors},

88: \author[i]{S.~Lombardi},

89: \author[q]{F.~Longo},

90: \author[i]{M.~L\'opez},

91: \author[c,h]{E.~Lorenz},

92: \author[k]{P.~Majumdar},

93: \author[v]{G.~Maneva},

94: \author[q]{N.~Mankuzhiyil},

95: \author[a]{K.~Mannheim},

96: \author[d]{L.~Maraschi},

97: \author[i]{M.~Mariotti},

98: \author[b]{M.~Mart\'{\i}nez},

99: \author[b]{D.~Mazin},

100: \author[l]{M.~Meucci},

101: \author[a]{M.~Meyer},

102: \author[e]{J.~M.~Miranda},

103: \author[h]{R.~Mirzoyan},

104: \author[s]{M.~Moles},

105: \author[b]{A.~Moralejo},

106: \author[e]{D.~Nieto},

107: \author[u]{K.~Nilsson},

108: \author[h]{J.~Ninkovic},

109: \author[h,w]{N.~Otte\thanksref{otte}},

110: \author[e]{I.~Oya},

111: \author[o]{M.~Panniello\thanksref{pann}},

112: \author[l]{R.~Paoletti},

113: \author[m]{J.~M.~Paredes},

114: \author[u]{M.~Pasanen},

115: \author[i]{D.~Pascoli},

116: \author[c]{F.~Pauss},

117: \author[l]{R.~G.~Pegna},

118: \author[s]{M.~A.~Perez-Torres},

119: \author[q,x]{M.~Persic},

120: \author[i]{L.~Peruzzo},

121: \author[l]{A.~Piccioli},

122: \author[s]{F.~Prada},

123: \author[i]{E.~Prandini},

124: \author[b]{N.~Puchades},

125: \author[n]{A.~Raymers},

126: \author[f]{W.~Rhode},

127: \author[m]{M.~Rib\'o},

128: \author[y,b]{J.~Rico},

129: \author[c]{M.~Rissi},

130: \author[g]{A.~Robert},

131: \author[a]{S.~R\"ugamer},

132: \author[i]{A.~Saggion},

133: \author[h]{T.~Y.~Saito},

134: \author[d]{M.~Salvati},

135: \author[s]{M.~Sanchez-Conde},

136: \author[i]{P.~Sartori},

137: \author[k]{K.~Satalecka},

138: \author[i]{V.~Scalzotto},

139: \author[q]{V.~Scapin},

140: \author[a]{R.~Schmitt},

141: \author[h]{T.~Schweizer},

142: \author[h]{M.~Shayduk},

143: \author[h]{K.~Shinozaki},

144: \author[b]{N.~Sidro},

145: \author[r]{A.~Sierpowska-Bartosik},

146: \author[u]{A.~Sillanp\"a\"a},

147: \author[j]{D.~Sobczynska},

148: \author[a]{F.~Spanier},

149: \author[l]{A.~Stamerra},

150: \author[c]{L.~S.~Stark},

151: \author[u]{L.~Takalo},

152: \author[d]{F.~Tavecchio},

153: \author[v]{P.~Temnikov},

154: \author[b]{D.~Tescaro},

155: \author[h]{M.~Teshima},

156: \author[k]{M.~Tluczykont},

157: \author[y,r]{D.~F.~Torres},

158: \author[l]{N.~Turini},

159: \author[v]{H.~Vankov},

160: \author[i]{A.~Venturini},

161: \author[q]{V.~Vitale},

162: \author[h]{R.~M.~Wagner\thanksref{cor}},

163: \ead{robert.wagner@mpp.mpg.de}

164: \author[h]{W.~Wittek},

165: \author[m]{V.~Zabalza},

166: \author[s]{F.~Zandanel},

167: \author[b]{R.~Zanin},

168: \author[g]{J.~Zapatero}

169: %\vspace*{-0.7cm}

170: %\collab{(The MAGIC Collaboration)}

171: \collab{and}

172: %

173: \author[z]{John Ellis},

174: \author[aa]{N.~E. Mavromatos},

175: \author[ab,ac,ad]{D.~V. Nanopoulos},

176: \author[c,z]{A.~S. Sakharov},

177: \author[z,ae]{E.~K.~G. Sarkisyan\thanksref{sark}}

178: 

179: \address[a]{Universit\"at W\"urzburg, D-97074 W\"urzburg, Germany}

180: \address[b]{IFAE, Edifici Cn., Campus UAB, E-08193 Bellaterra, Spain}

181: \address[c]{ETH Zurich, CH-8093 Switzerland}

182: \address[d]{INAF National Institute for Astrophysics, I-00136 Rome, Italy}

183: \address[e]{Universidad Complutense, E-28040 Madrid, Spain}

184: \address[f]{Technische Universit\"at Dortmund, D-44221 Dortmund, Germany}

185: \address[g]{Universitat Aut\`onoma de Barcelona, E-08193 Bellaterra, Spain}

186: \address[h]{Max-Planck-Institut f\"ur Physik, D-80805 M\"unchen, Germany}

187: \address[i]{Universit\`a di Padova and INFN, I-35131 Padova, Italy}

188: \address[j]{University of \L\'od\'z, PL-90236 Lodz, Poland}

189: \address[k]{DESY, Deutsches Elektr.-Synchrotron, D-15738 Zeuthen, Germany}

190: \address[l]{Universit\`a di Siena, and INFN Pisa, I-53100 Siena, Italy}

191: \address[m]{Universitat de Barcelona (ICC/IEEC), E-08028 Barcelona, Spain}

192: \address[n]{Yerevan Physics Institute, AM-375036 Yerevan, Armenia}

193: \address[o]{Instituto de Astrofisica de Canarias, E-38200, La Laguna, Tenerife, Spain}

194: \address[p]{Departamento de Astrofisica, Universidad, E-38206 La Laguna, Tenerife, Spain}

195: \address[q]{Universit\`a di Udine, and INFN Trieste, I-33100 Udine, Italy}

196: \address[r]{Institut de Cienci\`es de l'Espai (IEEC-CSIC), E-08193 Bellaterra, Spain}

197: \address[s]{Instituto de Astrofisica de Andalucia (CSIC), E-18080 Granada, Spain}

198: \address[t]{University of California, Davis, CA-95616-8677, USA}

199: \address[u]{Tuorla Observatory, Turku University, FI-21500 Piikki\"o, Finland}

200: \address[v]{Institute for Nuclear Research and Nuclear Energy, BG-1784 Sofia, Bulgaria}

201: \address[w]{Humboldt-Universit\"at zu Berlin, D-12489 Berlin, Germany}

202: \address[x]{INAF/Osservatorio Astronomico and INFN, I-34143 Trieste, Italy}

203: \address[y]{ICREA, E-08010 Barcelona, Spain}

204: \address[z]{Physics Department, CERN, CH-1211 Geneva 23, Switzerland}

205: \address[aa]{King's College London, Department of Physics, Strand, London WC2R 2LS, UK}

206: \address[ab]{Texas A\&M University, College Station, TX-77843, USA}

207: \address[ac]{HARC, Woodlands, TX-77381, USA}

208: \address[ad]{Academy of Athens, Division of Natural Sciences, GR-10679 Athens, Greece}

209: \address[ae]{Physics Department, Universiteit Antwerpen, B-2610 Wilrijk, Belgium}

210: 

211: \corauth[cor]{Corresponding author.}

212: \thanks[otte]{Now at: University of California, Santa Cruz, CA-95064, USA}

213: \thanks[sark]{Now at: University of Texas, Arlington, TX-76019, USA}

214: \thanks[pann]{Deceased.}

215: 

216: \date{Received 18 August 2008; Accepted 27 August 2008}

217: 

218: \begin{abstract}

219: We analyze the timing of photons observed by the MAGIC telescope during a flare

220: of the active galactic nucleus Mkn 501 for a possible correlation with energy,

221: as suggested by some models of quantum gravity (QG), which predict a vacuum

222: refractive index $\simeq 1 + (E/M_{\rm QGn})^n$, $n = 1,2$. Parametrizing the

223: delay between $\gamma$-rays of different energies as $\Delta t =\pm\tau_l E$ or

224: $\Delta t =\pm\tau_q E^2$, we find $\tau_l=(0.030\pm0.012)$~s/GeV at the

225: 2.5-$\sigma$ level, and $\tau_q=(3.71\pm2.57)\times10^{-6}$~s/GeV$^2$,

226: respectively. We use these results to establish lower limits $M_{\rm QG1} >

227: 0.21 \times 10^{18}$~GeV and $M_{\rm QG2} > 0.26 \times 10^{11}$~GeV at the

228: 95\% C.L. Monte Carlo studies confirm the MAGIC sensitivity to propagation

229: effects at these levels. Thermal plasma effects in the source are negligible,

230: but we cannot exclude the importance of some other source effect.

231: \end{abstract}

232: 

233: \begin{keyword}

234: Gamma-Ray Sources (Individual: Mkn 501) \sep 

235: Active Galactic Nuclei \sep 

236: Quantum Gravity \sep 

237: Supersymmetric Models \sep 

238: Particle Acceleration

239: \PACS

240: 98.70.Rz  03.30.+p  04.60.-m  95.85.Pw

241: \end{keyword}

242: 

243: \end{frontmatter}

244: 

245: \section{Introduction}

246: %

247: It is widely speculated that space-time is a dynamical medium, subject to

248: quantum-gravitational (QG) effects that cause space-time to fluctuate on the

249: Planck time and distance scales~\cite{amellis,foam,newstring,gambini,alfaro,kostel,amecam,myers}, 

250: for reviews see~\cite{review}. It has also been suggested that this `foaming' of

251: space-time might be reflected in modifications of the propagation of energetic

252: particles, namely dispersive effects due to a non-trivial refractive index

253: induced by the QG fluctuations in the space-time foam. There are microscopic

254: string-inspired models~\cite{amellis,foam,newstring} that predict only

255: subluminal refraction, and only for photons~\cite{crab}, suppressed either

256: linearly or quadratically by some QG mass scale:

257: %

258: \begin{equation}

259: \label{linear}

260: \frac{\Delta c}{c}=-\frac{E}{M_{\rm QG1}}~, \; {\rm or} \;

261: \frac{\Delta c}{c}=-\frac{E^2}{M^2_{\rm QG2}}.

262: \end{equation}

263: %

264: One might guess that the scale

265: $M_{\rm QG1}$ or $M_{\rm QG2}$ would be related to

266: ${\hat M}_{\rm P}$, where ${\hat M}_{\rm P} =

267: 2.4 \times 10^{18}$~GeV is the reduced Planck mass, but smaller

268: values might be possible in some string theories~\cite{foam,newstring}, or

269: models with large extra dimensions~\cite{merab}. Superluminal

270: modes and birefringence effects are also allowed in some other

271: models~\cite{gambini,alfaro,kostel,amecam,myers}.

272: 

273: A favored way to search for such a non-trivial dispersion relation is to

274: compare the arrival times of photons of different energies arriving on Earth

275: from pulses of distant astrophysical sources~\cite{amellis,mavrik}.

276: The greatest sensitivities may be expected from sources with short pulses, at

277: large distances or redshifts $z$, of photons observed over a large range of

278: energies. In the past, studies have been made of emissions from

279: pulsars~\cite{pulsar}, $\gamma$-ray bursts

280: (GRBs)~\cite{amellis,mavrik,wavegrb,robust,robust2,piranishvilo,merab} and

281: active galactic nuclei (AGNs)~\cite{mkr421,BLM}.

282: In particular, a combined analysis of many GRBs at different redshifts made

283: possible some separation between energy- and source-dependent effects, and

284: yielded a robust lower limit $M_{\rm QG1} > 0.9 \times

285: 10^{16}$~GeV~\cite{robust}. Astrophysical sources that produce very high

286: energy photons in the TeV range or higher could improve significantly the

287: sensitivity to Lorentz violation, if one could distinguish source and

288: propagation effects. Flaring AGNs are celestial objects with the desired

289: properties, and a pioneering study of a flare of the AGN Mkn~421 yielded a

290: sensitivity to $M_{\rm QG1} \sim 4 \times

291: 10^{16}$~GeV~\cite{mkr421}.\footnote{Stronger limits hold in models predicting

292: birefringence~\cite{uvgal,uvgrb}, but these do not apply to stringy models of

293: QG-induced vacuum dispersion~\cite{foam,newstring}, in which birefringence is

294: absent.}

295:  

296: In this Letter we analyze two flares of Mkn~501 ($z = 0.034$) observed by the

297: Major Atmospheric Gamma-ray Imaging Cerenkov (MAGIC) telescope~\cite{magic}

298: between May and July 2005. After applying standard quality checks, data

299: covering a total observation time of 31.6~h spread over 24 nights survived and

300: were analysed~\cite{magicobs}. The data were taken at zenith angles of

301: 10$^\circ-$30$^\circ$, resulting in an energy threshold (defined as the peak of

302: the differential event-rate spectrum after cuts) of $\approx 150$ GeV.

303: The air-shower events were subjected to the standard MAGIC analysis

304: \cite{magican}, which rejects about 99\% of hadronic background events, while

305: retaining 50$-$60\% of the $\gamma$-ray induced showers. The $\gamma$-ray

306: energies are, in a first approximation, proportional to the total amount of

307: light recorded in the shower images; corrections are applied according to

308: further image parameters \cite{hillaspar} obtained from the analysis. The

309: achieved energy resolution is $\sim 25$\% over the range 170~GeV to 10~TeV. The

310: arrival time of each event is obtained with sub-ms

311: precision.

312: 

313: During the observations, variations in the $\gamma$-ray flux by an order of

314: magnitude were observed, with the maximum integrated flux above $\approx$~150

315: GeV exceeding $(11.0\pm0.3) \times 10^{-10}$ cm$^{-2}$s$^{-1}$.

316: In the two nights with the highest flux, high-intensity outbursts of short

317: duration (flares) were recorded with characteristic rise and fall times of

318: 1$-$2 minutes. While the flare of July~9 was clearly visible over the full

319: energy range 0.15$-$10~TeV and reached a peak flux more than a factor two

320: higher than before and after the flare, that seen on June~30 was concentrated

321: at low energies (0.25$-$1~TeV) and less significant. In the analysis below, we

322: applied cuts on the image parameter $ALPHA$ \cite{hillaspar}, describing the

323: gamma shower arrival direction: $|ALPHA| <10^\circ$, and on energy: $E_\gamma

324: >150$~GeV.

325: 

326: \section{Timing analysis}

327: %

328: The spectral time properties of the most intense portions of the flares were

329: quantified in~\cite{magicobs} using four different energy bands with

330: boundaries at 0.15, 0.25, 0.6 and 1.2~TeV, the fourth band extending to

331: infinite energies. In the June~30 flare a signal above a uniform background

332: appeared only in the energy band of 0.25$-$0.6 TeV, which did not permit any

333: conclusion on the time-spectral properties of the signal. For the flare of

334: July~9, a time lag of about 4 minutes was found for the maximum of the time

335: profile envelope for photons in the 1.2$-$10~TeV energy band relative to those

336: in the range 0.25$-$0.6~TeV. The difference between the mean energies in these

337: two bands is $\approx$~2 TeV, which would lead to a naive estimate of a time

338: delay of about 0.12 s for each GeV of energy difference.

339: However, this approach is too simplistic, since the energy range covered by the

340: 1.2$-$10~TeV band is much larger than the energy difference between the two

341: bands, so the binned estimator used in~\cite{magicobs} is inadequate for

342: constraining possible linear or quadratic energy dependences. In view of this

343: and the limited number of photons, we improve here on the binned estimator by

344: analyzing the complete information encoded in the time-energy distribution of

345: {\it individual} photons in the flare, with the aim of probing possible

346: systematic energy-dependent time lags induced by QG vacuum refraction during

347: photon propagation to the Earth, or intrinsic to the source.

348: %

349: \begin{figure}

350: \begin{center}

351: \begin{minipage}{97mm}

352: \includegraphics[width=\linewidth]{f1.eps}

353: \caption{The ECF from one realization of the MAGIC measurements with

354: photon energies smeared by Monte Carlo, for the case of a vacuum refractive

355: index that is linear in the photon energy.}

356: \label{costeqg}

357: \end{minipage}

358: \end{center}

359: \end{figure}

360: %

361: The true shape of the time profile at the source is not known, so we choose the

362: following analysis strategy. In general, the short pulse structure of any

363: flare would be blurred by an energy-dependent effect on photon propagation.

364: Conversely, one may correct for the effects of any given parametric model of

365: photon dispersion, e.g., the linear or quadratic vacuum refractive index, by

366: applying to each photon the appropriate time shift~\cite{robust} corresponding

367: to its propagation in a spatially-flat expanding universe:

368: $\Delta t(E)=H_0^{-1}(E/M_{\rm QG1})\int\limits_0^z(1+z)h^{-1}(z)\mathrm{d}z$

369: or similarly for the quadratic case, where $H_0$ is the Hubble expansion rate

370: and $h(z) = \sqrt{\Omega_{\Lambda} + \Omega_M (1 + z)^3}$. If the correct

371: energy-dependent QG shift is applied, the short pulse structure

372: of the emission profile is restored.

373: 

374: We implement this analysis strategy in two ways. In one analysis, we consider

375: the most active part of the flare, that is distinguished clearly from the

376: uniform background, and the QG shift is varied so as to maximize the total

377: energy in this part. In the other analysis we use the shape of the flare as

378: extracted from untransformed low-energy data.

379: %

380: \begin{figure}

381: \begin{center}

382: \begin{minipage}{97mm}

383: \includegraphics[width=\linewidth]{f2.eps} 

384: \caption{The $\tau_l$ distribution from fits to the ECFs of 1000 realizations

385: of the July~9 flare with photon energies smeared by Monte Carlo.}

386: \label{histas}

387: \end{minipage}

388: \end{center}

389: \end{figure}

390: 

391: \subsection{Energy cost function}

392: %

393: It is well known~\cite{jac} that a pulse of electromagnetic radiation

394: propagating through a linearly-dispersive medium, as postulated here, becomes

395: diluted so that its power (the energy per unit time) decreases.\footnote{The

396: applicability of classical electrodynamics for estimating the low-energy

397: behavior induced by space-time

398: foam~\cite{review,amellis,foam,gambini,alfaro,kostel,amecam,myers,merab} and

399: the corresponding pulse-broadening effect have been discussed elsewhere

400: (see~\cite{mavrik} for details and an explicit example). The dilution effects

401: for the linear or quadratic cases may easily be obtained as described in

402: \cite[Sect 7.9]{jac} by applying the dispersion laws $\omega

403: (k)=k[1-k/(2M_\mathrm{QG1})]$ or $\omega (k)=k[1-k^2/(3M_\mathrm{QG2}^2)]$.}

404: Any transformation of a signal to reproduce the undispersed signal tends to

405: recover the original power of the pulse. If the parameter $M_{\rm QGn}$

406: is chosen correctly, the power of the recovered pulse is maximized.

407: 

408: We implement this observation as follows. First, we choose a time interval

409: $(t_1;t_2)$ containing the most active part of the flare, as determined using a

410: Kolmogorov-Smirnov (KS) statistic~\cite{ks}. The KS statistic is calculated

411: from the difference between the cumulative distribution function estimated from

412: the unbinned data and that of a uniform distribution. The interval $(t_1;t_2)$

413: covers the time range where the value of the KS difference varies from its

414: maximum over the whole time support of the signal down to a negligible value.

415: This procedure determines the proper time-width $t_2-t_1$ of the most active

416: (transient) part of the flare.\footnote{The time interval chosen agrees very

417: well with the spread of a Gaussian fit to the profile of the binned data, as

418: well as the more complicated profile used in~\cite{magicobs}.}

419: Having chosen this window, we scan over the whole support the time-distribution

420: of all photons shifted by $\Delta t(E)$ and sum up the energies of photons in

421: the window. For convenience, we re-parametrize the time shift as 

422: $\Delta t = \pm \tau_l E$ or $ \Delta t = \pm \tau_qE^2$ respectively,

423: with $\tau_l$ and $\tau_q$ in ${\rm s/GeV}$ and ${\rm s/GeV^2}$ units.

424: The transformation is repeated for many values of $\tau_l$ and $\tau_q$, chosen

425: so that the shifts $\Delta t$ match the precision of the arrival-time

426: measurements, and for each $\tau_l$ or $\tau_q$ the scan is performed and the

427: maximal summed energy in a window of width $t_2-t_1$ is obtained. The maximal

428: energies as a function of $\tau_l$ or $\tau_q$ define the `energy cost

429: function' (ECF). The position of the maximum of the ECF indicates the value of

430: $\tau_l$ or $\tau_q$ that best recovers the signal, in the sense of maximizing

431: its power.\footnote{Varying slightly the boundaries of the interval $(t_1;t_2)$

432: has a negligible effect on the position of the maximum. We take into account

433: the difference between the width at the Earth and at the source, also

434: negligible.} This procedure is applied to 1000 Monte Carlo (MC) data samples

435: generated by applying to the measured photon energies the (energy-dependent)

436: Gaussian measurement errors.

437: 

438: Fig.~\ref{costeqg} shows the ECF for one such energy-smeared MC sample. It

439: exhibits a clear maximum, whose position may be estimated by fitting it with a

440: Gaussian profile in the peak vicinity. Fig.~\ref{histas} shows the results of

441: such fits to the ECFs with $\tau_l$ for the 1000 energy-smeared realizations of

442: the July~9 flare. From this distribution we derive the value $\tau_l=(0.030\pm

443: 0.012)$~s/GeV, where $M_{\rm QG1}=1.445\times10^{16}\,{\rm s}/\tau_l$, leading

444: to a lower limit $M_{\rm QG1} > 0.21 \times 10^{18}$~GeV at the 95\%

445: C.L.\footnote{ We propagate the large errors by using the $\pm 1$-$\sigma$

446: range of ${\hat M}_{\rm P}/M_{\rm QGn}$ to estimate the $\mp1$-$\sigma$ range

447: of $M_{\rm QGn}$.} The same procedure applied to the ECF obtained using

448: $\tau_q$ leads to $\tau_q=(3.71 \pm 2.57)\times 10^{-6}$~s/GeV$^2$, where

449: $M_{\rm QG2}=1.222\times10^8\,({\rm s}/\tau_q)^{1/2}$, corresponding to $M_{\rm

450: QG2} > 0.26 \times 10^{11}$~GeV at the 95\% C.L. While our results for the

451: June~30 flare have similar sensitivities and are compatible, they cannot be

452: used to strengthen our results, as this flare is not very

453: significant.

454: 

455: \subsection{Likelihood function}

456: %

457: We have confirmed this result using another technique to study the

458: energy-dependent delay signal in the data. It is motivated by the initial time

459: and energy-binned analysis performed in~\cite{magicobs}, which we used to check

460: that the light-curve is well described by a simple Gaussian profile,

461: superimposed on a time-independent background. We compute a likelihood function

462: ${\cal L}$ based on the probability of a photon to be observed with energy $E$

463: and arrival time $t$, using variables describing the energy spectrum at the

464: source, the time distribution at emission obtained from the measured arrival

465: times of the photons assuming an adjustable energy-dependent propagation delay,

466: and the energy resolution of the detector, which is modelled as a

467: Gaussian~\cite{martinez}. To describe the photon energy at the source a simple

468: power law $\Gamma(E_{\rm s}) \sim E_{\rm s}^{-\beta}$ is taken, with $\beta =

469: 2.7$ for the time-uniform part of the flare and $2.4$ for the flaring

470: part~\cite{magicobs}.

471: %

472: \begin{figure}

473: \begin{center}

474: \begin{minipage}{122mm}

475: \includegraphics[width=\linewidth]{f3.eps}

476: \caption{Comparison of the MAGIC measured lightcurve at low and high

477: energies with the prediction given by the best set of parameters found using

478: the likelihood method, and binning the data and the likelihood function in the

479: same manner.}

480: \label{datafit}

481: \end{minipage}

482: \vspace*{.4cm}

483: \end{center}

484: \end{figure}

485: 

486: The likelihood function is fitted to the July~9 data minimizing $-\log{\cal L}$

487: as a function of four parameters: (i) the energy-dependent delay parameterized

488: in terms of ${\hat M_P}/M_{\rm QG1}$, (ii) the position of the intrinsic

489: maximum of the Gaussian flare, (iii) its width and (iv) the normalization of

490: the time-independent background component in arbitrary units. The best fit

491: yields ${\hat M}_{\rm P}/M_{\rm QG1} = 8.2^{+ 3.7}_{- 3.4}$, corresponding to 

492: $\tau_l = (0.048\pm0.021)$~s/GeV.

493: The shape of the function $\chi^2 \equiv -2 \log {\cal L} + {\rm const}$ around

494: the minimum in these variables is quite parabolic almost up to the 2-$\sigma$

495: level. In view of the correlations with these parameters, the sensitivity to

496: $\tau_l$ would be improved if they were known more precisely.

497: 

498: A similar procedure in the case of a quadratic dependency gives

499: $\tau_q = (4.60\pm5.46)\times10^{-6}$ s/GeV$^2$.

500: 

501: Fig.~\ref{datafit} shows that the ${\cal L}$ function gives a good overall fit

502: to the data: binning in time and energy both the data and the ${\cal L}$

503: function, we find $\chi^2$/NDF $\sim 1$.

504: 

505: \subsection{Crosscheck with Monte Carlo data}

506: %

507: To check the robustness of the ECF and likelihood analyses, we simulated

508: several MC test samples with two components: (a) a time-independent background

509: with the same energy spectrum as the measured data before the flare, and (b) a

510: superposed signal generated at the source with an energy spectrum similar to

511: that observed during the flare and an energy-independent Gaussian time

512: distribution, each with the same numbers of photons as in the measurement. We

513: then calculated the arrival times of all photons using various dispersion

514: models and parameters, taking into account the MAGIC energy resolution. For

515: each dispersion model and parameter, we generated 1000 incarnations, using

516: different random seeds. These samples were then analyzed blindly, and the

517: encoded effects were recovered successfully by the two estimators within the

518: expected uncertainties. In addition, the analysis techniques were applied to MC

519: samples with no energy-dependent dispersive signal encoded, and found no

520: effect, and both techniques also returned null results when applied to Mkn 501

521: data from outside a flare. These tests confirm the numerical sensitivities of

522: the analyses and the estimates of the uncertainties given above. For the

523: likelihood method, additional checks have been performed~\cite{martinez}

524: assuming different flare energy spectra and shapes, besides the Gaussian one

525: discussed here, which also fit reasonably well the binned data (c.f.

526: Fig.~\ref{datafit}).

527: 

528: \section{Conclusions}

529: %

530: The probability of the zero-delay assumption relative to the one obtained with

531: the ECF estimator is $P=0.026$.

532: %

533: The observed energy-dependent delay thus is a likely observation, but does not

534: constitute a statistically firm discovery.

535: %

536: The results of the two independent analyses of the July~9 flare of Mkn~501 are

537: quite consistent within the errors. Their results exhibit a delay between

538: $\gamma$-rays of different energies, $\tau_l=(0.030\pm0.012)$~s/GeV,

539: corresponding to a lower limit $M_{\rm QG1} > 0.21 \times 10^{18}$~GeV at the

540: 95\% C.L. We also find a quadratic delay

541: $\tau_q=(3.71\pm2.57)\times10^{-6}$~s/GeV$^2$, and $M_{\rm QG2} > 0.26 \times

542: 10^{11}$~GeV at the 95\% C.L., far beyond previous limits on a quadratic effect

543: in photon propagation~\cite{mkr421,wavegrb,merab}.

544: %

545: These numbers could turn into a real measurement of $M_{\rm QG1,2}$, if the 

546: emission mechanism at the source were understood and the observed delays were mainly

547: due to propagation.

548: %

549: We cannot exclude, however, the possibility that the delay we find, which is significant

550: beyond the 95\% C.L., is due to some energy-dependent effect at the

551: source.\footnote{Note that if the observed energy-dependent time shift is

552: explained by some source effect, the lower limit on $M_{\rm QG}$ would rise.}

553: However, we can exclude the possibility that the observed time delay may be due

554: to a conventional QED plasma refraction effect induced as photons propagate

555: through the source. This would induce~\cite{plasma} $\Delta t = D (\alpha^2

556: T^2/6q^2) \ln^2(qT/m_e^2)$, where $\alpha$ is the fine-structure constant, $q$

557: is the photon momentum, $T$ is the plasma temperature, $m_e$ is the mass of

558: electron, $D$ is the size of the plasma, and we use natural units: $c, \hbar

559: =1$. Plausible numbers such as $T \sim 10^{-2}$~MeV and $D \sim 10^9$~km

560: (for a review see \cite{hillas}) yield a negligible effect for $q \sim 1$~TeV.

561: Exclusion of other source effects, such as time evolution in the mean emitted

562: photon energy, might be possible with the observation of more flares, e.g., of

563: different AGNs at varying redshifts. Observations of a single flare cannot

564: distinguish the quantum-gravity scenarios considered here from modified

565: synchrotron-self-Compton mechanisms~\cite{magicobs,wagner08}. However, this

566: pioneering study demonstrates clearly the potential scientific value of an

567: analysis of multiple flares from different sources. The most promising

568: candidate for applying the analyses proposed here is the flare

569: from PKS 2155-304 detected recently by H.E.S.S. \cite{hess2155}. 

570: Unfortunately the occurrence of fast flares in AGNs is currently unpredictable,

571: and since no correlation has yet been established with observations in other

572: energy bands that could be used as a trigger signal, only serendipitous

573: detections are currently possible.

574: 

575: 

576: \section*{Acknowledgments}

577: %

578: The MAGIC Collaboration thanks the IAC for the excellent working conditions at

579: the Observatorio del Roque de los Muchachos in La Palma and gratefully

580: acknowledges the support of the German BMBF and MPG, the Italian INFN and the

581: Spanish CICYT. This work was also supported by ETH Research Grant TH 34/04 3

582: and the Polish MNiI Grant 1P03D01028. The work of J.E. and N.E.M. was

583: partially supported by the European Union through the Marie Curie Research and

584: Training Network UniverseNet MRTN-CT-2006-035863, and that of D.V.N. by DOE

585: grant DE-FG02-95ER40917.

586: 

587: \begin{thebibliography}{99}

588: %

589: \bibitem{amellis} G.~Amelino-Camelia, J.~Ellis, N.~Mavromatos, D.~Nanopoulos, S.~Sarkar, Nature 393 (1998) 763.

590: \bibitem{foam} G.~Amelino-Camelia, J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, Int. J. Mod. Phys. A 12 (1997) 607;\\ 

591: J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, Phys.\ Lett.\ B {293} (1992) 37;\\ 

592: J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, Erice Subnucl. Phys. Series 31 (1994) 1; \\ 

593: J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, Chaos Solitons Fractals {10} (1999) 345.

594: \bibitem{newstring} J.R.~Ellis, N.E.~Mavromatos and D.V.~Nanopoulos, Phys. Lett. B 665 (2008) 412.

595: \bibitem{gambini} R.~Gambini, J.~Pullin, Phys.\ Rev.\ D {59} (1999) 124021.

596: \bibitem{alfaro} J.~Alfaro, H.A.~Morales-Tecotl, L.F.~Urrutia, Phys.\ Rev.\ D {65} (2002) 103509.

597: \bibitem{kostel} V.A.~Kostelecky, S.~Samuel, Phys.\ Rev.\ D {39} (1989) 683.

598: \bibitem{amecam} G.~Amelino-Camelia, Int.\ J.\ Mod.\ Phys.\ D {11} (2002) 35.

599: \bibitem{myers} R.C.~Myers, M.~Pospelov, Phys.\ Rev.\ Lett.\ {90} (2003) 211601.

600: \bibitem{review} S.~Sarkar, Mod.\ Phys.\ Lett.\ A {17} (2002) 1025; \\

601: L.~Smolin, arXiv: hep-th/0303185; 

602: \\Y.J.~Ng, Mod.\ Phys.\ Lett.\ A {18} (2003) 1073; \\T.~Piran, Lect. Notes Phys. {669} (2005) 351; \\

603: D.~Mattingly, Living Rev. Rel. {8} (2005) 5.

604: \bibitem{crab} J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, A.S.~Sakharov, Nature {428} (2004) 386.

605: \bibitem{merab} M.~Gogberashvili, A.S.~Sakharov, E.K.G.~Sarkisyan, Phys.\ Lett.\ B {644} (2007) 179.

606: \bibitem{mavrik} J.~Ellis, K.~Farakos, N.E.~Mavromatos, V.A.~Mitsou, D.V.~Nanopoulos, Astrophys. J. {535} (2000) 139.

607: \bibitem{pulsar} P.~Kaaret, Astron. Astrophys. {345} (1999) L32.

608: \bibitem{wavegrb} J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, A.S.~Sakharov, Astron.\ Astrophys.\ {402} (2003) 409.

609: \bibitem{robust} J.R.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, A.S.~Sakharov, E.K.G.~Sarkisyan, Astropart.\ Phys.\ {25} (2006) 402.

610: \bibitem{robust2} J.~Ellis, N.E.~Mavromatos, D.V.~Nanopoulos, A.S.~Sakharov, E.K.G.~Sarkisyan, Astropart.\ Phys.\ {29} (2008) 158.

611: \bibitem{piranishvilo} M.~Rodr\'iguez Mart\'inez, T.~Piran, Y.~Oren, JCAP {0605} (2006) 017.

612: \bibitem{mkr421} S.D.~Biller, et al., Phys. Rev. Lett. {83} (1999) 2108.

613: \bibitem{BLM} O.~Blanch, J.~L{\' o}pez, M.~Mart{\' i}nez, Astropart.\ Phys.\ {19} (2003) 245.

614: \bibitem{uvgal} R.J.~Gleiser and C.N.~Kozameh, Phys.\ Rev.\ D {64} (2001) 083007.

615: \bibitem{uvgrb} Y.Z.~Fan, D.M.~Wei and D.~Xu, Month.\ Not.\ R.\ Astron.\ Soc.\ {376} (2007) 1857.

616: \bibitem{magic} E. Lorenz, MAGIC Collaboration, New Astr. Rev. {48} (2005) 339.

617: \bibitem{magicobs} J.~Albert, et al., MAGIC Collaboration, Astrophys. J. {669} (2007) 862.

618: \bibitem{magican} R.M.~Wagner, et al., MAGIC Collaboration, in Proceedings of 29th International Cosmic Ray Confeference (Pune), {4} (2005) 163.

619: \bibitem{hillaspar} A.M.~Hillas, in Proceedings of 19th International Cosmic Ray Conference (La Jolla), 3 (1985) 445.

620: \bibitem{jac} J.D.~Jackson, Classical Electrodynamics, John Wiley \& Sons, Inc., 1975.

621: \bibitem{ks} D.J.~Sheskin, Handbook of Parametric and Nonparametric Statistical Procedures, Chapman \& Hall/CRC, 2007.

622: \bibitem{martinez} M.~Martinez, M.~Errando, arXiv: 0803.2120 [astro-ph], Nucl. Instrum. Methods A, submitted for publication.

623: \bibitem{plasma} J.I.~Latorre, P.~Pascual, R.~Tarrach, Nucl.\ Phys.\ B {437}, (1995) 60.

624: \bibitem{hillas} L.~Anchordoqui, T.~Paul, S.~Reucroft, J.~Swain, Int. J. Mod. Phys. A~18 (2003) 2229.

625: \bibitem{wagner08} W.~Bednarek, R.M.~Wagner, Astron. Astrophys. 486 (2008) 679.

626: \bibitem{hess2155} F.~Aharonian, et al., H.E.S.S. Collaboration, Astrophys. J. 664 (2007) L71.

627: 

628: \end{thebibliography}

629: 

630: \end{document}

631: 

632: