1: %% Beginning of file 'sample.tex'

2: %%

3: %% Modified 2005 June 21

4: %%

5: %% This is a sample manuscript marked up using the

6: %% AASTeX v5.x LaTeX 2e macros.

7: 

8: %% The first piece of markup in an AASTeX v5.x document

9: %% is the \documentclass command. LaTeX will ignore

10: %% any data that comes before this command.

11: 

12: %% The command below calls the preprint style

13: %% which will produce a one-column, single-spaced document.

14: %% Examples of commands for other substyles follow. Use

15: %% whichever is most appropriate for your purposes.

16: %%

17: \documentclass[12pt,preprint]{aastex}

18: %\documentclass[preprint]{aastex}

19: 

20: %% manuscript produces a one-column, double-spaced document:

21: 

22: %%\documentclass[manuscript]{aastex}

23: %\usepackage[]{graphicx}

24: 

25: %% preprint2 produces a double-column, single-spaced document:

26: 

27: %%\documentclass[preprint2]{aastex}

28: 

29: %% Sometimes a paper's abstract is too long to fit on the

30: %% title page in preprint2 mode. When that is the case,

31: %% use the longabstract style option.

32: 

33: %% \documentclass[preprint2,longabstract]{aastex}

34: 

35: %% If you want to create your own macros, you can do so

36: %% using \newcommand. Your macros should appear before

37: %% the \begin{document} command.

38: %%

39: %% If you are submitting to a journal that translates manuscripts

40: %% into SGML, you need to follow certain guidelines when preparing

41: %% your macros. See the AASTeX v5.x Author Guide

42: %% for information.

43: 

44: %\newcommand{\vdag}{(v)^\dagger}

45: %\newcommand{\myemail}{skywalker@galaxy.far.far.away}

46: 

47: %% You can insert a short comment on the title page using the command below.

48: 

49: %\slugcomment{Not to appear in Nonlearned J., 45.}

50: 

51: %% If you wish, you may supply running head information, although

52: %% this information may be modified by the editorial offices.

53: %% The left head contains a list of authors,

54: %% usually a maximum of three (otherwise use et al.).  The right

55: %% head is a modified title of up to roughly 44 characters.

56: %% Running heads will not print in the manuscript style.

57: 

58: \shorttitle{A universal GRB energy-luminosity relation}

59: \shortauthors{Willingale et al.}

60: 

61: %% This is the end of the preamble.  Indicate the beginning of the

62: %% paper itself with \begin{document}.

63: 

64: \begin{document}

65: 

66: %% LaTeX will automatically break titles if they run longer than

67: %% one line. However, you may use \\ to force a line break if

68: %% you desire.

69: 

70: \title{A universal GRB photon energy-peak luminosity relation}

71: 

72: %% Use \author, \affil, and the \and command to format

73: %% author and affiliation information.

74: %% Note that \email has replaced the old \authoremail command

75: %% from AASTeX v4.0. You can use \email to mark an email address

76: %% anywhere in the paper, not just in the front matter.

77: %% As in the title, use \\ to force line breaks.

78: 

79: \author{R. Willingale\altaffilmark{1},

80: P.T. O'Brien\altaffilmark{1},

81: M.R. Goad\altaffilmark{1},

82: J.P. Osborne\altaffilmark{1},

83: K.L. Page\altaffilmark{1},

84: N.R. Tanvir\altaffilmark{1}

85: %%O. Godet\altaffilmark{1},

86: %%D.N. Burrows\altaffilmark{2},

87: %%B. Zhang\altaffilmark{4},

88: %%E. Rol\altaffilmark{1},

89: %%N. Gehrels\altaffilmark{3},

90: %%G. Chincarini\altaffilmark{5}

91: }

92: 

93: %% Notice that each of these authors has alternate affiliations, which

94: %% are identified by the \altaffilmark after each name.  Specify alternate

95: %% affiliation information with \altaffiltext, with one command per each

96: %% affiliation.

97: 

98: \altaffiltext{1}{Department of Physics and Astronomy, University of Leicester,

99: LE1 7RH, UK}

100: %%\altaffiltext{2}{Department of Astronomy and Astrophysics, Pennsylvania State

101: %%University, University Park, PA 16802, USA}

102: %%\altaffiltext{3}{NASA Goddard Space Flight Center, Greenbelt, Maryland, 20771, USA}

103: %%\altaffiltext{4}{Department of Physics, University of Nevada, Las Vegas, NV 89154, USA}

104: %%\altaffiltext{5}{INAF, Osservatorio Astronomico di Brera, Via E. Bianchi 46,

105: %%I-23807 Merate (LC), Italy}

106: 

107: %% Mark off your abstract in the ``abstract'' environment. In the manuscript

108: %% style, abstract will output a Received/Accepted line after the

109: %% title and affiliation information. No date will appear since the author

110: %% does not have this information. The dates will be filled in by the

111: %% editorial office after submission.

112: 

113: \begin{abstract}

114: The energetics and emission mechanism of GRBs are not well understood.

115: Here we demonstrate that the instantaneous peak flux or equivalent isotropic

116: peak luminosity, $L_{iso}$ ergs s$^{-1}$, rather than the integrated

117: fluence or equivalent isotropic energy, $E_{iso}$ ergs,

118: underpins the known high-energy correlations.

119: Using new spectral/temporal parameters calculated for

120: 101 bursts with redshifts from {\em BATSE}, {\em BeppoSAX}, {\em HETE-II}

121: %{\em Konus-WIND},{\em INTEGRAL},

122: and {\em Swift} we describe a parameter space which

123: characterises the apparently diverse properties of the prompt emission.

124: We show that a source frame characteristic-photon-energy/peak luminosity

125: ratio, $K_{z}$, can be constructed

126: which is constant within a factor of 2

127: for all bursts whatever their duration, spectrum,

128: luminosity and the instrumentation used to detect them.

129: %This confirms that what really distinguishes short from long is not

130: %the peak luminosity but the duration and total energy.

131: The new parameterization embodies the Amati relation but

132: indicates that some correlation between $E_{peak}$ and $E_{iso}$ follows

133: as a direct mathematical inference from the Band function and that

134: a simple transformation of $E_{iso}$ to $L_{iso}$

135: yields a universal high energy correlation for GRBs.

136: The existence of $K_{z}$ indicates that the mechanism responsible

137: for the prompt emission from all GRBs is probably predominantly thermal.

138: \end{abstract}

139: 

140: %% Keywords should appear after the \end{abstract} command. The uncommented

141: %% example has been keyed in ApJ style. See the instructions to authors

142: %% for the journal to which you are submitting your paper to determine

143: %% what keyword punctuation is appropriate.

144: 

145: \keywords{Gamma Rays: bursts --- radiation mechanisms: non-thermal ---

146: ISM: jets and outflows}

147: 

148: %% From the front matter, we move on to the body of the paper.

149: %% In the first two sections, notice the use of the natbib \citep

150: %% and \citet commands to identify citations.  The citations are

151: %% tied to the reference list via symbolic KEYs. The KEY corresponds

152: %% to the KEY in the \bibitem in the reference list below. We have

153: %% chosen the first three characters of the first author's name plus

154: %% the last two numeral of the year of publication as our KEY for

155: %% each reference.

156: 

157: %% Authors who wish to have the most important objects in their paper

158: %% linked in the electronic edition to a data center may do so by tagging

159: %% their objects with \objectname{} or \object{}.  Each macro takes the

160: %% object name as its required argument. The optional, square-bracket 

161: %% argument should be used in cases where the data center identification

162: %% differs from what is to be printed in the paper.  The text appearing 

163: %% in curly braces is what will appear in print in the published paper. 

164: %% If the object name is recognized by the data centers, it will be linked

165: %% in the electronic edition to the object data available at the data centers  

166: 

167: \section{Introduction}

168: 

169: The energetics of the central engine which powers the explosion

170: responsible for a GRB are both intriguing and fundamental to our

171: understanding of these cosmic events. The isotropic energy outflow at source,

172: estimated using the integrated gamma-ray fluence, is enormous, up to

173: $E_{iso}\sim10^{54}$

174: ergs, and even if the outflow is collimated in jets the total

175: energy involved is still huge,

176: $E_{\gamma}\sim10^{51}$ ergs. The possibility that the explosion taps

177: a standard energy resevoir has been pursued by many authors following

178: the initial suggestion from Frail et al. (2001). If this total energy

179: available were, indeed, roughly constant (or predictable through other means)

180: and we could reliably estimate the collimation,

181: then GRBs could be used as a cosmological probe to very high redshifts,

182: Bloom et al. (2003), Ghirlanda et al. (2004).

183: 

184: Early on it was noted that, based on analysis of {\em BATSE} data,

185: there was a correlation between $E_{p}$, the peak of

186: $E.F(E)$ where $F(E)$ ergs cm$^{-2}$ keV$^{-1}$ is the observed

187: spectrum, and the fluence (Mallozzi et al. 1995,

188: Lloyd et al. 2000). When redshifts

189: became available for long bursts

190: the isotropic energy, $E_{iso}$, could be estimated from 

191: the fluence and the peak energy could be transformed into the source

192: frame, $E_{pz}$, the so-called Amati relation, a correlation

193: between $E_{iso}$ and $E_{pz}$ in the sense that more energetic

194: bursts have a higher $E_{pz}$, was discovered using data from

195: {\em BeppoSAX}, (Amati et al. 2002). This correlation has subsequently been

196: confirmed and extended although there remain many significant

197: outliers, including all short bursts.

198: The physical origin of the correlation may be associated

199: with the emission mechanisms operating in the fireball but the theoretical

200: details are far from settled

201: (see the discussion by Amati (2006) and references therein).

202: More recently a tighter correlation between $E_{iso}$, $E_{pz}$

203: and the jet break time, $t_{break}$, measured in the optical afterglow

204: has been reported (Ghirlanda et al. 2004).

205: This is explained in terms of a modification to the Amati relation in which

206: $E_{iso}$ is corrected to a true collimated energy, $E_{\gamma}$,

207: using an estimate of the collimation angle derived from $t_{break}$.

208: The details of the 

209: collimation correction depend on the density and density profile

210: of the circumburst medium, Nava et al. (2006) and references therein.

211: Multivariable regression analysis was performed by Liang \& Zhang (2005)

212: to derive a model-independent relationship,

213: $E_{iso}\propto E_{pz}^{1.94}t_{zbreak}^{-1.24}$, indicating that the

214: rest-frame break time of the optical afterglow, $t_{zbreak}$

215: was indeed correlated with the prompt emission parameters.

216: 

217: Other studies have concentrated on the properties of the

218: isotropic peak (maximum) luminosity, $L_{iso}$

219: ergs s$^{-1}$, measured over some short time scale $\approx1$ s,

220: rather than the time integrated isotropic energy, $E_{iso}$.

221: Yonetoku et al. (2004) noted a correlation between $L_{iso}$ and

222: $E_{pz}$ for 16 GRBs with firm redshifts. A correlation

223: between $L_{iso}$ and the spectral lag was first identified by Norris et al.

224: (2000) and explained in terms of the evolution of $E_{peak}$ with time.

225: The shocked material responsible for

226: the gamma-ray emission is expected to cool at a rate proportional

227: to the gamma-ray luminosity and it has been suggested that

228: $E_{peak}$ traces the cooling (Schaefer 2004). 

229: A similar

230: correlation between $L_{iso}$ and the variability of the GRB ($V$) was

231: described by Reichart et al. (2001). The origin of the $L_{iso}-V$ 

232: relation is likely to be related to the physics of the relativistic shocks

233: and the bulk Lorentz factor of the outflow.

234: It could be that 

235: high $\Gamma_{outflow}$ results in high $L_{iso}$ and $V$ while

236: lower luminosity and variability are expected if $\Gamma_{outflow}$ is low

237: (see, for example, M\'{e}sz\'{a}ros et al. 2002).

238: A rather bizzare correlation involving $L_{iso}$, $E_{pz}$ and variability

239: was found by Firmani et al. (2006). They employed the ``high signal'' time,

240: $T_{45}$, as formulated by Reichart et al. (2001) in their study of

241: variability, and showed that $L_{iso}\propto E_{pz}^{1.62}T_{45}^{-0.49}$

242: for 19 GRBs with a spread much narrower than that of the Amati relation.

243: There is currently no explanation for such a correlation although

244: it may be connected with the spectral lag and variability

245: correlations and the Amati relation.

246: 

247: The correlation between $E_{iso}$ and $E_{pz}$ supplemented by additional

248: empirical information can be

249: used in pseudo redshift indicators, for example Atteia (2003), 

250: Pelangeon \& Atteia (2006), but the intrinsic spread in the correlation

251: and uncertainty about the underlying physical interpretation

252: introduce errors, typically

253: of a factor $\sim2$. It may be possible to reduce the errors by

254: simultaneous application of several independent luminosity/energy 

255: correlations, and extension of the Hubble Diagram to high redshifts

256: using GRBs has been attempted, see for example Schaefer (2007).

257: However, it is not clear that the correlations briefly described above

258: are truly independent and there may be some underlying principle or

259: mechanism which connects them all together. Recently, and more controversially,

260: Butler et al. (2007) have raised serious doubts about the validity

261: of these correlations suggesting that it is likely that they

262: are introduced by observational/instrumental bias and have nothing to

263: do with the physical properties of the GRBs and hence they conclude

264: that GRBs are probably

265: useless as cosmological probes. Here we take a new look at the source

266: frame spectral and temporal properties of a large number of GRBs for

267: which we have redshifts in order

268: to try and understand what really correlates with what and whether or

269: not this can provide useful intrinsic information about the GRBs and

270: what drives them. In this analysis we include the short-duration

271: GRBs which may share a similar emission mechanism with long bursts

272: despite probably having different progenitors.

273: 

274: \section{Source frame spectra of the prompt emission}

275: 

276: The profile of the prompt energy spectrum of all GRBs is well represented by

277: a Band function (Band et al. 1993),

278: \[

279: B(E)=E^{-\beta_{X}} \exp(-E/E_{c}),\:\: E \le E_{c}(\beta_{\gamma}-\beta_{X})

280: \]

281: \begin{equation}

282: B(E)=E^{-\beta_{\gamma}} \exp(\beta_{\gamma}-\beta_{X})

283: [(\beta_{\gamma}-\beta_{X})E_{c}]^{\beta_{\gamma}-\beta_{X}},\:\:

284: E \ge E_{c}(\beta_{\gamma}-\beta_{X})

285: \label{eq1}

286: \end{equation}

287: where $\beta_{X}$ and $\beta_{\gamma}$ are the spectral power law indices

288: at low (X-ray) and high ($\gamma$-ray)

289: energies respectively and $E_{c}$ keV is the high cut-off energy. Note

290: that in the original formulation of Band et al. (1993) photon indices were

291: used and the profile described the photon number density

292: (because these are the parameters which most closely describe the detected

293: count spectrum which is fitted). Here we choose to

294: use an energy density profile and energy spectral indices.

295: The observed total fluence is

296: \begin{equation}

297: F_{tot}=\int_{E_{1}}^{E_{2}}F(E)dE=

298: N_{tot}\int_{E_{1}}^{E_{2}}B(E)dE

299: \label{eq2}

300: \end{equation}

301: ergs cm$^{-2}$, where $N_{tot}$ is the normalisation in

302: ergs cm$^{-2}$ keV$^{-1}$ at 1 keV and $E_{1}$ to $E_{2}$ is the

303: observed energy band. Spectral fitting of the observed count spectrum

304: will yield values for $\beta_{X}$, $\beta_{\gamma}$, $E_{c}$

305: and $N_{tot}$. The cut-off energy, $E_{c}$, is often converted

306: to the peak energy of the $E.F(E)$

307: spectrum which is given by $E_{p}=(1-\beta_{X})E_{c}$ and the

308: normalisation may be expressed as the fluence, $F_{tot}$, rather than the energy

309: density at 1 keV, $N_{tot}$. However, the separation of the fluence into a

310: normalisation term and a spectral integral is central to the development

311: of the argument which follows.

312: Table \ref{tab1} gives the spectral

313: parameters for 101 GRBs for which we have redshift values and a prompt

314: light curve.

315: The spectral

316: parameters for bursts detected by {\em BATSE}, {\em BeppoSAX}, {\em HETE-2}

317: and {\em Konus/WIND}

318: %and {\em INTEGRAL}

319: were taken from the references cited. The values for {\em Swift} bursts

320: were derived from the {\em BAT} spectra supplemented by detections

321: by {\em INTEGRAL} and {\em Konus/WIND} where available. Many of the

322: {\em Swift} spectra ($\approx 40$) are adequately fitted by a simple

323: power law or a cut-off power law with $E_{c}$ fixed. For these

324: bursts a cut-off power law model was used with $E_{c}=150$ keV

325: (corresponding to the upper limit of the BAT energy band). Providing

326: the fitted $\beta_{X}<1$ the fitted function has a peak in $E.F(E)$

327: and a value for the peak energy can then be estimated.

328: The spectra of 7 very soft {\em Swift} bursts with redshifts

329: (GRB050406, GRB050416A, GRB050824, GRB051016B, GRB060512, GRB060926 and

330: GRB070419A)

331: gave $\beta_{X}\ge 1$ and these were discarded because, for such spectra,

332: we have no meaningful estimate of $E_{p}$.

333: Such GRBs are normally designated as X-ray flashes (XRFs) and

334: the exceptionally high $\beta_{X}$ values may arise because we are actually

335: observing the high energy tail ($\beta_{\gamma}$) and not the lower energy

336: power law in the Band function. Alternatively it may be that such

337: soft spectra are the result of a second soft X-ray component which

338: dominates in these objects.

339: 

340: The equivalent isotropic energy from the source is given by

341: \begin{equation}

342: E_{iso}=\frac{4\pi d_{L}^{2} N_{tot}}{(1+z)^{2-\beta_{X}}}

343: I_{bol}(E_{pz},\beta_{X},\beta_{\gamma})

344: \label{eq3}

345: \end{equation}

346: ergs, where $d_{L}$ is the luminosity distance corresponding to the

347: redshift $z$ under some cosmology,

348: $I_{bol}(E_{pz},\beta_{X},\beta_{\gamma})$

349: is the bolometric integral of the spectral energy profile in the source

350: frame, $B_{z}$, taken over the wide energy band 1 keV to 10 MeV

351: \begin{equation}

352: I_{bol}(E_{pz},\beta_{X},\beta_{\gamma})=\int_{1}^{10^{4}} B_{z}(E)dE

353: \label{eq4}

354: \end{equation}

355: and $E_{pz}=E_{p}(1+z)$ is the peak energy in the source frame.

356: The first term in Equation \ref{eq3}

357: is the equivalent isotropic energy density, $Q_{z}$

358: ergs keV$^{-1}$ at 1 keV in the source frame.

359: \begin{equation}

360: Q_{z}=\frac{4\pi d_{L}^{2} N_{tot}}{(1+z)^{2-\beta_{X}}}.

361: \label{eq5}

362: \end{equation}

363: A factor $1/(1+z)^{1-\beta_{X}}$ arises because we have shifted the

364: normalisation from 1 keV in observer frame to 1 keV in the source frame.

365: The remaining factor of $1/(1+z)$ accounts for the time-dilation of the

366: duration over which the bursts are seen.

367: It is pertinent to transform this to the isotropic energy density at the peak

368: energy, $E_{pz}$ keV, in the source frame,

369: \begin{equation}

370: Q_{pz}=Q_{z}\exp[(1-\beta_{X})(E_{pz}^{-1}-1)] E_{pz}^{-\beta_{X}}

371: \label{eq6}

372: \end{equation}

373: ergs keV$^{-1}$ so that

374: the spectrum normalisation is specified at a characteristic

375: energy in or close to the

376: observed $\gamma$-ray energy band.

377: We can then write Equation \ref{eq3} as

378: \begin{equation}

379: E_{iso}=Q_{pz}E_{wz},

380: \label{eq7}

381: \end{equation}

382: where

383: \begin{equation}

384: E_{wz}=

385: \exp [(\beta_{X}-1)(E_{pz}^{-1}-1)] 

386: E_{pz}^{\beta_{X}}

387: I_{bol}(E_{pz},\beta_{X},\beta_{\gamma})

388: \label{eq8}

389: \end{equation}

390: keV is a {\em characteristic photon energy} which depends on the profile

391: of the energy spectrum and the limits adopted for the integration

392: and it serves to convert from an energy density ($Q_{pz}$ ergs keV$^{-1}$)

393: at the peak of the $E.F_{z}(E)$

394: spectrum to the total isotropic energy ($E_{iso}$ ergs).

395: The isotropic energy spectrum in the source frame is given  by

396: \begin{equation}

397: F_{z}(E)=Q_{z}B_{z}(E)=

398: Q_{pz} \exp [(\beta_{X}-1)(E_{pz}^{-1}-1)]

399: E_{pz}^{\beta_{X}}B_{z}(E)

400: \label{eq9}

401: \end{equation}

402: ergs keV$^{-1}$.

403: The source frame spectra of the GRBs listed in Table \ref{tab1}

404: are shown in Figure \ref{fig1} with the spectral energy density $Q_{pz}$ marked

405: at energy $E_{pz}$ keV.

406: In the majority of spectra the high energy spectral index is not measured but

407: set to $\beta_{\gamma}=1.3$ which is the approximate average

408: found by {\em BATSE}. 

409: Figure \ref{fig2} shows the corresponding

410: $E.F_{z}(E)$ spectra in ergs.

411: We assumed a cosmology with $H_{0}=71$ km s$^{-1}$

412: Mpc$^{-1}$, $\Lambda=0.27$ and $\Omega=0.73$ to calculate the 

413: luminosity distance $d_{L}$.

414: 

415: \section{The Amati relation}

416: 

417: The Amati relation is a correlation between $E_{pz}$ and $E_{iso}$,

418: first reported by Amati et al. (2002), and subsequently shown to be obeyed

419: by the majority of long GRBs although there is a fairly large scatter.

420: The top left panel of Figure \ref{fig3} shows the histogram of

421: isotropic energy values, $E_{iso}$, calculated using Equation \ref{eq7} using

422: the spectral parameters in Table \ref{tab1} and redshift in Table \ref{tab3}.

423: A large range of values for $E_{iso}$ is produced because

424: of the spread in the isotropic

425: energy density at the peak $Q_{pz}$, the peak energy $E_{pz}$

426: and the bolometric integral $I_{bol}$. The top right panel of

427: Figure \ref{fig3} shows the peak energy values, $E_{pz}$, plotted

428: against the characteristic energy, $E_{wz}$. There is a tight correlation

429: between these 2 parameters because of the form of the Band function.

430: To a first approximation $E_{pz}=0.23 E_{wz}$ (the solid line

431: in Figure \ref{fig3}) although

432: the best fit correlation is a little steeper ($E_{pz}\propto E_{wz}^{1.14}$)

433: and the small scatter

434: evident in Figure \ref{fig3} is introduced by differences in the spectral

435: indices, $\beta_{X}$ and $\beta_{\gamma}$.

436: In fact, the bolometric integral is well approximated by a function of the form

437: \begin{equation}

438: I_{bol}\approx I_{fit}=E_{pz}^{1+c_{1}}

439: \exp(c_{0}+c_{2}\beta_{X}+c_{3}\beta_{\gamma})

440: \label{eq10}

441: \end{equation}

442: where the coefficients $c_{0},c_{1},c_{2},c_{3}$ can be found by a

443: least squares fitting procedure. A comparison of $I_{fit}$ and $I_{bol}$

444: for the GRBs listed in Table \ref{tab1} is shown in the bottom left panel of

445: Figure \ref{fig3} together with the best fit coefficients.

446: We can use $I_{fit}$ in place of $I_{bol}$ and

447: estimate $E_{iso}\approx E_{fit}$. The distribution of the

448: ratio $E_{iso}/E_{fit}$ is

449: shown in the bottom right-hand panel of Figure \ref{fig3}. For the majority

450: of objects the estimation, $E_{fit}$, is within $\pm10\%$ of the value obtained

451: by numerical integration. There are a few GRBs with a larger discrepancy

452: but all are within $\pm20\%$ which is a very small perturbation in comparison

453: with the dynamic range of the $E_{iso}$ values.

454: 

455: Using $I_{bol}\approx I_{fit}$  we can express $E_{iso}$ as an explicit

456: function of $E_{pz}$:

457: \begin{equation}

458: E_{iso}\approx Q_{pz} E_{pz}^{\beta_{X}+1+c_{1}}

459: \exp(1+c_{0}+(c_{2}-1)\beta_{X}+c_{3}\beta_{\gamma}).

460: \label{eq11}

461: \end{equation}

462: The immediate origin of the Amati relationship is now clear.

463: Given Equation \ref{eq11} some degree of correlation between $E_{pz}$ and

464: $E_{iso}$ is guaranteed. The nature and spread of this correlation

465: will depend on the relationship between

466: the flux density, $Q_{pz}$, and the peak energy, $E_{pz}$, and the distribution

467: of spectral index $\beta_{X}$. It could be that $Q_{pz}$ and $E_{pz}$ are

468: correlated in such a way to cancel the apparent dependence on $E_{pz}$ but

469: this is highly unlikely.

470: This correlation arises because the GRB spectral profile has the form of 

471: Band function (Equation \ref{eq1}) with a particular range of values

472: for the spectral indices, $\beta_{X}$, $\beta_{\gamma}$,

473: and the energy $E_{c}$.

474: So understanding where the Amati

475: relation comes from is really the same as understanding why the spectra

476: have this functional form in the first place.

477: 

478: Figure \ref{fig4} shows the Amati relationship

479: for the GRBs in Table \ref{tab1}.

480: Here and subsequently we use the exact form for $E_{iso}$, calculated

481: from $I_{bol}$, and not the

482: approximation involving $I_{fit}$ which was only introduced to derive

483: Equation \ref{eq11}.

484: The correlation line shown (derived

485: ignoring the obvious outliers) is $E_{pz}\propto E_{iso}^{0.46}$ 

486: consistent with Amati 2006, $E_{pz}\propto E_{iso}^{0.5}$.

487: All the short bursts are outliers with

488: low $E_{iso}$ values compared with the long bursts of similar

489: $E_{pz}$ value. The other notable outliers are GRB980425

490: and GRB060218 (see Amati 2006, Campana et al. 2006).

491: The XRFs (characterised by the hardness ratio of the low energy spectra, see

492: below) all fall on the lower edge of the correlation with low $E_{pz}$ 

493: compared with $E_{iso}$.

494: A more fundamental

495: relationship is that between the flux density $Q_{pz}$ and the

496: characteristic energy $E_{wz}$ which is also shown in Figure \ref{fig4}.

497: It appears that, disregarding the short bursts, the Amati correlation is

498: tighter than this new relationship but this is deceptive.

499: Unlike $E_{iso}$ and $E_{pz}$, $Q_{pz}$ and $E_{wz}$

500: are independent and their product provides the isotropic

501: energy $E_{iso}$ (Equation \ref{eq7}). We now have a correlation which

502: goes beyond the simple fact that GRB spectra have the Band function profile.

503: Crudely, $Q_{pz}$ is a measure of the

504: height of the spectrum as plotted in Figure \ref{fig1} and $E_{wz}$

505: (which is itself a function of $E_{pz}$, $\beta_{X}$ and $\beta_{\gamma}$)

506: is a measure of the characteristic photon energy.

507: There is a weak correlation between these two

508: quantities, $E_{wz}\propto Q_{pz}^{0.3}$, as can be seen

509: in Figures \ref{fig4} and \ref{fig1}.  However,

510: the pattern of outliers is the same as for the Amati relationship.

511: The short bursts, and the sub-luminous long burst, GRB980425,

512: have significantly low $Q_{pz}$ values but $E_{wz}$ values

513: which are comparable to the gamut of long bursts. 

514: The bursts designated as XRFs (see below)

515: all lie in the low tail of the $E_{wz}$ range

516: but have $Q_{pz}$ values which are similar to many long bursts.

517: 

518: \section{The rate profile and luminosity time of the prompt emission}

519: 

520: The analysis above has highlighted the well known problems associated with

521: the Amati relation and other correlations involving $E_{iso}$.

522: We now consider a way of converting $E_{iso}$ into a characteristic

523: luminosity to see if this can improve the situation.

524: The variety of time variablity in the prompt emission from GRBs is 

525: astonishing. Some bursts consist of a single Fast Rise Exponential Decay

526: (FRED) profile, other have multiple peaks, some are very spikey with

527: rapid variations while others have a smoother profile. The luminosity

528: is continually varying between bright, short peaks and low troughs

529: and in some cases the flux drops below the detection threshold for a while

530: before flaring up again. With such a range of behaviour defining some

531: characteristic luminosity and/or duration

532: is tricky. Reichart et al. (2001) showed that the peak

533: luminosity correlated with a variability measure $V$ 

534: computed by taking the difference between the light

535: curve and a smoothed version of the light curve

536: where the smoothing or correlation time was the 

537: time taken to emit the brightest fraction $f$ of the flux,

538: $T^{E}_{f}$. They showed that the most robust correlation was obtained for

539: $f\approx0.45$. The correlation of the peak luminosity with $T_{45}$

540: has been adopted by subsequent authors, for example Guidorzi et al. (2005),

541: Firmani et al. (2006), but in all cases

542: the peak luminosity must be defined using some small arbitrary bin size

543: (typically 1 second) and the only connection between the total fluence

544: and the peak luminosity is indirect, through the $T_{45}$ value.

545: 

546: The variability measure $V$ depends on the correlation of structures

547: (peaks, troughs etc.) in the light curves. Here we try a different

548: approach in which the sequence of features or events in the light curves

549: is abandoned completely. We identify the time periods in which 

550: significant flux is measured and then construct a {\em rate profile}

551: by sorting the sequence of count rate samples from these time periods

552: into descending order to produce,

553: for every GRB, a monotonically decreasing function, $f_{s}(t_{s})$, where

554: $t_{s}$ is sorted time. The total sum of

555: all the samples should be the total count fluence and the profile

556: is normalised by dividing by this fluence so that the integral under

557: the profile is unity. Such a rate profile shows what fraction of the

558: burst is spent at what fraction of the peak rate and has the general

559: form shown schematically in Figure \ref{fig5}.

560: Examples of these

561: rate profiles are shown in Figures \ref{fig6} and \ref{fig8}.

562: The time periods in which significant flux is detected were found by

563: successive correlation with boxcar functions of increasing width. It

564: doesn't matter if the total duration of these

565: periods is a little larger than required to capture the total fluence

566: because the small excess of samples in the tail can be dropped and

567: the rest of the profile is unchanged.

568: Remarkably the shape of these rate profiles is surprisingly similar

569: for {\em all} GRBs and is insensitive

570: to the time bin size used as long as it is not too large or

571: too small. If the bin size is too large

572: then there may be too few samples defining the profile, but we found that

573: a number of bins $>20$ was fine. Using excessively large time bins can

574: also hide significant real structure in the fluctuations of the light curve

575: and this should be avoided.

576: At the other extreme, if the bins are

577: too small the number of counts per bin may drop to single figures and

578: the profile shape is again compromised. In practice all long bursts are

579: well represented using $\approx64$ ms bins while short bursts require 

580: $\approx4$ ms bins or something similar.

581: 

582: The influence of statistical fluctuations (noise) on the rate

583: profiles is rather strange. Because the integral is normalised to unity

584: statistical fluctuations on the total fluence are not included.

585: The profile reflects the distribution of the detected flux over a

586: range of brightness but is not influenced by uncertainties in the total flux.

587: The sorting of bins into decreasing brightness order also ensures the profiles

588: are always smooth with the larger errors or distortion due to noise

589: accumulating at the start and end of the profile. This is often most noticable

590: as a slight increase in gradient or curl over at the end of the profile.

591: Although errors can be estimated for each of the samples, $d_{i}$,

592: Chi-squared minimization using

593: these errors cannot be employed for any function fitting because

594: the sorting operation destroys the meaning of the errors. i.e. the scatter of

595: the sorted data values about the fitted function is not governed directly by

596: the errors on $d_{i}$.

597: 

598: Most profiles are well represented by an empirical function of the form

599: \begin{equation}

600: f_{s}(t_{s})=f_{0}\left(

601: 1-\left( \frac{t_{s}}{T_{E}}\right)^{1/C_{L}}\right)^{C_{L}}+f_{E}

602: \label{eq12}

603: \end{equation}

604: where $T_{E}$ is the total emission time or duration of the profile,

605: $f_{E}$ is the level of the profile at $T_{E}$ and represents the

606: minimum detectable flux (or luminosity) and

607: $C_{L}$ is a luminosity index which describes the curvature.

608: This function is illustrated in Figure \ref{fig5}.

609: Because the profile integral

610: is normalised to unity the peak value at the start is

611: $f_{0}+f_{E}=1/T_{L}$, where

612: $T_{L}$ is a {\em luminosity time} in seconds. The peak flux is then given by

613: the fluence divided by the luminosity time, $F_{tot}/T_{L}$ cts s$^{-1}$

614: or, perhaps more intuitively, the peak flux multiplied by $T_{L}$

615: is the total fluence. Because $T_{L}$ is derived from the functional

616: fit of all the data it does not depend strongly

617: on the time bin size (as discussed

618: above) and therefore the peak flux calculated using $T_{L}$ is also not

619: dependent on the binning.

620: If $C_{L}=1$ the

621: profile is linear and if $C_{L}>1$ the profile is concave and the fraction

622: at high rate is smaller.

623: If $C_{L}<1$ the curvature would be negative

624: but this is not seen for any GRBs. So $C_{L}$ is a measure

625: of the sharpness or spikiness of the profile.

626: 

627: We fitted all rate profiles with the function $f_{s}$ given

628: by Equation \ref{eq12} finding the best fit values for the parameters

629: $T_{L}$, $C_{L}$ and $f_{E}$ using a least squares statistic

630: \begin{equation}

631: \Sigma=100N\Delta t^{2} \sum(d_{i}-f_{si})^{2}

632: \label{eq13}

633: \end{equation}

634: where $N$ is the total number of samples, $d_{i}$, of time width $\Delta t$.

635: Note $T_{E}$ is fixed as the cumulative duration of all the significant

636: samples detected, $T_{E}=N\Delta t$.

637: The $\Sigma$ statistic has properties similar to

638: reduced Chi-Squared, with

639: typical values in the range 0.5-2.0 (set by the scaling factor of 100),

640: independent of the number of samples $N$ or the sample size $\Delta t$.

641: Table \ref{tab2} provides a 

642: complete list of all the temporal parameters. This table also includes

643: the instrument and a GRB classification using the usual observational

644: definitions:

645: Short bursts if $T_{90}<2$ s and X-ray Flashes (XRFs) if

646: fluence(1-30 keV)/fluence(30-500 keV)$>1$.

647: 

648: Figure \ref{fig6} shows examples of typical fits. Note that

649: sorted time is scaled by $1/T_{E}$ and the $f_{s}$ values by $T_{L}$ so

650: that both axes take the range 0-1. The top-right panels show

651: GRB021211 which is a typical FRED burst and has a low curvature index,

652: $C_{L}=1.28$. The top-left and bottom right panels show

653: GRB990510 and GRB070521 which have more complicated flaring structure but

654: are well fitted with $C_{L}$ values of 2.48 and 1.69 respectively.

655: The remaining objects have short bright spikes and extended low

656: level emission, a class discussed by

657: Norris \& Bonnell (2006). GRB050724 and GRB051221A are essentially short bursts

658: followed by a low level, extended tail and the combination of these

659: features produces large $C_{L}$ values, 3.17 and 2.85

660: respectively.

661: For GRB051221A, $\Sigma=3.22$ which is rather high. In this

662: case the short spike followed by the extended tail produces an extra feature

663: or wiggle in the rate profile which is not fitted by the

664: simple $f_{s}$ function, Equation \ref{eq12}.

665: For these and similar bursts a sample size of 4 ms was used to

666: accomodate the profile of the initial short spike.

667: The left-hand panel of

668: Figure \ref{fig7} shows the distribution of $\Sigma$ and $T_{L}$

669: values for all GRBs in Table \ref{tab2}. There is no correlation between

670: the goodness of fit measured by $\Sigma$ and the luminosity time, $T_{L}$.

671: The same is true for $\Sigma$ and the luminosity index $C_{L}$.

672: Figure \ref{fig8} shows the worst fits

673: of rate profiles with large $\Sigma$ values. In all these GRBs

674: the peak value, $1/T_{L}$, is a good approximation to the data peak

675: but the fit is compromised by undulating features. GRB990705 and

676: GRB061007 represent a small group of bursts which have flares that

677: rise fast, are reasonably flat at the top and decay fast. These produce

678: a characteristic S-feature in the profile.

679: Only 12 rate profiles (out of 101) have $\Sigma>2$ and

680: only 3 of these have a substantial mis-match, GRB990705, GRB061007 and

681: GRB061210. The latter is an extreme example of a short burst, $T_{L}=0.03$ s,

682: which has an extended low flux tail giving $T_{90}=85.3$ s. We note that

683: GRB991216 has a faint pre-cursor just visible on the lightcurve plot.

684: 

685: The combination of luminosity time, $T_{L}$, and curvature index,

686: $C_{L}$, gives us information closely related to $T_{45}$. The right-hand

687: panel of Figure \ref{fig7} shows the correlation between $C_{L}$ and the

688: ratio of $T_{45}$ calculated directly from the sample values $d_{i}$ and

689: $T_{L}$ from the fitted function. $T_{45}$ could be calculated

690: by integration of the fitted function using the parameters

691: $T_{L}$, $C_{L}$, $f_{E}$ and $T_{E}$ and this would produce a smooth curve of

692: $C_{L}$ vs. $T_{45}/T_{L}$ if $f_{E}$ were zero or constant.

693: The parameter $C_{L}$, for example,

694: could be replaced by $T_{45}$ and the fitted function would still be

695: uniquely defined. The scatter in Figure \ref{fig7}

696: results from the small differences between the data and the fitted function

697: and the value of $f_{E}$ which is generally much smaller than

698: $1/T_{L}$ but different for each GRB.

699: Error ranges for $T_{L}$ and $C_{L}$ were estimated assuming

700: the statistic $\Sigma$ has properties similar to reduced Chi-Squared.

701: The errors so derived are not statistically correct, because of

702: the odd statistical nature of the sorted rate profile, and in some

703: cases they are an over estimate as is evident from the

704: scatter in Figure \ref{fig7}.

705: 

706: Although the minimum flux level, $f_{E}$, was included in the fitting it

707: is a measure of the instrument sensitivity rather than some intrinsic

708: property of the rate profile. If the noise level were lower the

709: number of significant samples detected would increase, $T_{E}$ would

710: get bigger and $f_{E}$ would decrease. The instrument would detect

711: a slightly larger fluence, $F_{tot}$, and the fitted value of

712: $T_{L}$ would increase a little, however, the peak flux level,

713: $F_{tot}/T_{L}$ would remain unchanged and $C_{L}$ would be

714: essentially the same. The analysis of the rate profile

715: described above provides a robust estimate of the peak flux (or peak

716: luminosity) using all the available light curve data and is

717: not biased by the instrument sensitivity providing the burst

718: detection significance is secure in the first instance.

719: The error on the peak flux so estimated is dominated by the error on the

720: fluence rather than any error associated with estimating the

721: luminosity time, $T_{L}$. It is also unchanged by the choice of sample

722: size, $\Delta t$, providing the number of samples is sufficient to

723: capture the details of the emission profile as already discussed above.

724: We can never be sure that resampling a light curve with a smaller

725: $\Delta t$ will not reveal a very short, bright, isolated spike which was

726: hidden by the previous binning and this would compromise the shape

727: of the profile, but such has not been seen in any of the GRB light

728: curves analysed so far (about 250 including all {\em Swift} bursts to date).

729: 

730: Using the redshift, $z$,

731: we can calculate the luminosity time in the source frame,

732: $T_{Lz}=T_{L}/(1+z)$ and $90\%$

733: duration in the source frame, $T_{90z}=T_{90}/(1+z)$.

734: The peak luminosity multiplied by the $T_{Lz}$ gives the

735: isotropic energy, $L_{iso}T_{Lz}=E_{iso}$ ergs. This simple

736: property of $T_{Lz}$ makes it a highly significant measure of the burst

737: duration and is why we chose to call it the luminosity time. Such a time

738: is often introduced in theoretical dicussions, see for example $t_{j}$

739: in Thompson et al. (2007) or $t_{burst}$ in Ghirlanda et al. (2007).

740: Above we have described a method to calculate this time for every GRB.

741: 

742: \section{Characterisation of the prompt emission in the source frame}

743: 

744: The prompt emission of each GRB in the source frame is characterised

745: by the peak energy density,

746: $Q_{pz}$ ergs keV$^{-1}$, the characteristic photon energy, $E_{wz}$ keV

747: (which embodies the spectral indices $\beta_{X}$, $\beta_{\gamma}$ and the

748: peak energy $E_{pz}$, Equation \ref{eq8}),

749: the luminosity time, $T_{Lz}$ s, and the luminosity curvature, $C_{L}$.

750: Figure \ref{fig9} shows $T_{Lz}$ plotted against the standard measure of

751: burst length $T_{90z}$ where the dashed line shows equality.

752: For bursts consisting

753: of a single smooth pulse then $T_{Lz}\approx T_{90z}$. If there is more

754: structure in the light curve and, in particular, if there are periods when

755: the flux drops to zero then $T_{Lz}<T_{90z}$. In some cases a short

756: precursor pulse is followed by a long time gap before the main burst

757: starts and then $T_{Lz}<<T_{90z}$. So the ratio of the two times is a

758: crude measure of the variability but this

759: includes all time scales and long periods when no flux is detected

760: and is not equivalent to the short time scale variability defined by

761: Reichart et al. (2001).

762: The top-right panel

763: of Figure \ref{fig9} shows the distribution of $T_{Lz}$. Two

764: peaks containing the short-bursts,

765: centred around 0.05 seconds, and long-bursts centred

766: at 5 seconds, are clearly visible.

767: The distribution of $C_{L}$ is shown in the lower left-hand panel of

768: Figure \ref{fig9}. Most bursts are contained in a symmetrical peak

769: centred on $C_{L}=1.6$. The few bursts with $C_{L}>2.2$ include

770: the short bursts which have a long weak tail and bursts which exhibit

771: several very short spikes on top of a more generally smooth emission.

772: The bottom right-hand panel of Figure \ref{fig9} shows the distribution

773: of the Band lower energy spectral index,

774: $\beta_{X}$. Hard bursts have $\beta_{X}<0$ and softer bursts have

775: $\beta_{X}>0$. We do not show the distribution of the high energy

776: spectral index, $\beta_{\gamma}$,

777: because this parameter is only available for a few bursts and in most

778: cases it was set to $\beta_{\gamma}=1.3$ which is the approximate average

779: found by {\em BATSE}.

780: 

781: The distributions of the remaining parameters,

782: the characteristic photon energy, $E_{wz}$, and the peak energy density,

783: $Q_{pz}$, are shown at the top of the Figure \ref{fig10}.

784: $E_{wz}$ stretches over two decades from 100 keV to 10000 keV.

785: $Q_{pz}$ has a much larger spread with a main peak spanning three

786: decades and a low energy tail covering another three.

787: Since the product of the two gives us $E_{iso}$ the

788: range of isotropic energy is very large, as is evident from Figure \ref{fig3}

789: and the Amati relation plotted in Figure \ref{fig4}.

790: The peak energy density, $Q_{pz}$, is correlated with the luminosity

791: time $T_{Lz}$ as demonstrated by the bottom left-hand panel of Figure

792: \ref{fig10}.

793: Short bursts have $Q_{pz}<10^{48.5}$ while in general

794: long bursts have larger $Q_{pz}$ values. The two notable exceptions

795: are, as before, GRB980425 and GRB060218 which are long bursts with very

796: low luminosity. Five short bursts with long tails that are classified

797: as long because their $T_{90}>2$,

798: GRB050603, GRB050724, GRB061006, GRB061210 and GRB070714B

799: have $T_{Lz}$ of 0.22, 0.38, 0.33, 0.02 and 0.48 s

800: respectively and these sit below the main long grouping along with the

801: shorts.

802: The XRFs tend to have lower

803: $Q_{pz}$ and lower $T_{Lz}$ values within the long burst population.

804: The peak luminosity density of a burst

805: is given by $Q_{pz}/T_{Lz}$ ergs keV$^{-1}$ s$^{-1}$. This has

806: a much narrower distribution than either $Q_{pz}$ or $T_{Lz}$ with

807: a $90\%$ range just over 2 decades, $3.1\times10^{47}-7.8\times10^{49}$

808: ergs keV$^{-1}$ s$^{-1}$, as is

809: clear from the histogram in the bottom right-hand panel of Figure \ref{fig10}.

810: Both short and long bursts have similar values of peak luminosity

811: density (the short bursts are shown as the white histogram, all bursts

812: are shown in the grey histogram).

813: 

814: We have used Principal Components Analysis (PCA) to investigate the

815: scatter within the parameter space described above

816: ($T_{Lz}$, $C_{L}$, $Q_{pz}$ and $E_{wz}$). This analysis confirms that

817: there is indeed a correlation between $T_{Lz}$ and $Q_{pz}$ and the

818: best fit is very close to proportionality with index $0.89$ but there is

819: considerable scatter with Pearson's correlation coefficient is $r=0.5$,

820: Kendall's $\tau=0.3$, $4.5\sigma$ (see Figure \ref{fig10}).

821: If we fix this index to unity then the only other significant

822: correlation is between $Q_{pz}/T_{Lz}$ and the characteristic energy

823: $E_{wz}$.

824: This is shown in the left-hand panel of Figure \ref{fig11} including

825: the best fit correlation,

826: $E_{wz}\propto (Q_{pz}/T_{Lz})^{0.25}$ which has Pearson's 

827: correlation coefficient $r=0.55$ and Kendall's $\tau=0.32$, significance

828: $4.8\sigma$.

829: The critical difference between this

830: plot and the right-hand panel of Figure \ref{fig4} (the Amati

831: relation) is that the 

832: peak energy density $Q_{pz}$ has been converted to a peak luminosity

833: density by dividing by the time $T_{Lz}$. The large difference between

834: the short and long bursts has disappeared and most bursts are now 

835: clustered in a small area on the energy-luminosity plane.

836: It seems that all correlations

837: involving the properties of GRBs must have outliers and this is no

838: exception; GRB980425 still refuses to conform but the remaining 100 bursts come

839: into line.

840: 

841: The correlation shown in Figure \ref{fig11} between the peak luminosity density

842: and the characteristic photon energy in the source frame

843: is the first GRB relationship to unify the short and the long

844: bursts.

845: If the peak luminosity density is multiplied by $E_{wz}$ the x-axis

846: becomes the peak isotropic luminosity, $L_{iso}$ ergs s$^{-1}$.

847: The correlation of $E_{wz}$ vs. $L_{iso}$ is shown in the

848: right-hand panel of Figure \ref{fig11} along with the best fit

849: \begin{equation}

850: \frac{E_{wz}}{381\:{\rm keV}}

851: =\left(\frac{L_{iso}}{10^{50}\:{\rm ergs\:s^{-1}}}\right)^{0.25}

852: \label{eq14}

853: \end{equation}

854: which has a Pearson's correlation coefficient of $r=0.75$, Kendall's 

855: $\tau=0.54$, significance $8.0\sigma$. 

856: Thus Figure \ref{fig11} encapsulates a major result of this work, showing

857: a high quality correlation of characteristic photon energy with

858: peak isotropic luminosity for 101 GRBs including 9 short bursts and 7 XRFs.

859: The correlation between $E_{wz}$ and $L_{iso}$ is similar to those

860: reported by Yonetoku et al. (2004) and

861: Firmani et al. (2006) but there are important differences. Here we have

862: estimated the peak isotropic

863: luminosity from the rate profile so we are not

864: restricted to long bursts or a particular time bin size, and the

865: peak energy, $E_{pz}$, is replaced by the characteristic photon energy,

866: $E_{wz}$. We note that the correlation derived by Yonetoku et al. (2004)

867: is significantly steeper, $E_{pz}\propto L_{iso}^{0.5\pm0.1}$

868: but they used a rather small sample of 16 GRBs.

869: We can identify 13 of these objects

870: in our sample and we find they give $E_{pz}\propto L_{iso}^{0.41\pm0.06}$ with

871: Pearson's correlation coefficient $r=0.81$, consistent with their result.

872: The same 13 objects also give 

873: $E_{wz}\propto L_{iso}^{0.31\pm0.04}$ with $r=0.88$

874: so using $E_{wz}$ in place of $E_{pz}$ gives a slightly tighter correlation 

875: with a shallower slope which is consistent with our result obtained

876: from the full sample of 101 bursts.

877: Unlike the Firmani et al. relationship the present

878: correlation does not contain $T_{45z}$. We tried including $T_{45z}$ in the

879: PCA but found no significant correlation or reduction in the scatter.

880: If we replace $E_{wz}$ by $E_{pz}$ in the PCA of the complete sample then

881: $E_{pz}\propto L_{iso}^{0.27}$ with $r=0.71$ and Kendall's $\tau=0.52$,

882: significance $7.7\sigma$ so, again, using $E_{wz}$ yields a tighter

883: correlation with a shallower slope compared to $E_{pz}$. The small change in

884: slope arises because the correlation of $E_{pz}$ with $E_{wz}$ is not quite

885: unity (see Figure \ref{fig3}).

886: 

887: For each burst we calculate $K_{z}$ which is a measure of its displacement

888: perpendicular from the the best fit correlation line in the right-hand panel

889: of Figure \ref{fig11}.

890: \[K_{z}

891: =\left(\frac{E_{wz}}{1320\:{\rm keV}}\right)^{0.97}

892: \left(\frac{L_{iso}}{1.45\times10^{52}\:{\rm ergs}\:{\rm s}^{-1}}\right)^{-0.24}\]

893: \begin{equation}

894: =\left(\frac{E_{wz}}{1320\:{\rm keV}}\right)^{0.74}

895: \left(\frac{Q_{pz}}{2.08\times10^{49}\:{\rm ergs}\:{\rm keV}^{-1}}\:

896: \frac{1.89\: {\rm s}}{T_{Lz}}\right)^{-0.24}

897: \label{eq15}

898: \end{equation}

899: This is a function of the ratio of the characteristic photon energy to the peak

900: isotropic luminosity.

901: The constants quoted in this definition are the mean values of the parameters

902: so they represent the centre of the clustering of objects within the

903: parameter space.

904: The distribution of $K_{z}$ is plotted in Figure \ref{fig12}.

905: The mean value is $\log_{10}(K_{z})=0$ or, equivalently,

906: $K_{z}=1$.

907: Hard-dim bursts (including most short bursts)

908: have $K_{z}>1$, soft-bright bursts (including all XRFs)

909: have $K_{z}<1$. The distribution is approximately log-normal

910: (the best fit Gaussian profile is shown in Figure \ref{fig12})

911: and has a rms width of

912: $\sigma[\log_{10}(K_{z})]=0.19$. $90\%$ of the GRBs (90 objects) are

913: contained in the range $0.45<K_{z}<1.95$.

914: The obvious outlier is GRB980425/SN1998bw which is either very 

915: sub-luminous or has an exceptionally high peak energy for such a dim

916: burst.

917: Under the hypothesis

918: that $K_{z}$ is constant, $\chi^{2}=421$ with 99 degrees of freedom

919: and the mean of the estimated errors on $\log{10}(K_{z})$ is 0.12

920: so there is clear evidence for intrinsic scatter in $K_{z}$ with an

921: estimated $90\%$ range of $0.57<K_{z}<1.75$.

922: The largest uncertainties arise from the estimation of $E_{wz}$ because

923: this depends on $E_{pz}$ and the spectral indices $\beta_{X}$, $\beta_{\gamma}$.

924: The mean value of $K_{z}$ for the pre-{\em Swift} bursts is

925: $-0.05$ and for {\em Swift} bursts is $0.02$ so they are statistically

926: indistinquishable.

927: The distribution for pre-{\em Swift} bursts, plotted as the white 

928: histogram in Figure \ref{fig12}, sits symmetrically within the total 

929: distribution.

930: The right-hand panel of Figure \ref{fig12} shows $\log_{10}(K_{z})$

931: as a function of redshift, $z$. There is no obvious trend. The objects

932: with the smallest errors that contribute most to the high $\chi^{2}$ show

933: no dependence on redshift. Table \ref{tab3} provides a complete listing

934: of the rest frame parameters and associated errors.

935: 

936: \section{Discussion}

937: 

938: Within the new parameterisation of the temporal and spectral properties

939: of the prompt GRB emission the three important quantities are the

940: characteristic energy in the source frame, $E_{wz}$ keV (Equation \ref{eq8}),

941: the energy density at the peak of the $E.F_{z}(E)$ spectrum,

942: $Q_{pz}$ ergs kev$^{-1}$ (derived from  the total fluence, Equations \ref{eq2},

943: \ref{eq5} and \ref{eq6}) and the luminosity

944: time, $T_{Lz}$ s, derived from the rate profile.

945: The ratio $Q_{pz}/T_{Lz}$ gives us the peak luminosity density in

946: ergs keV$^{-1}$ s$^{-1}$ where ``peak'' corresponds to both the

947: maximum in the $E.F_{z}(E)$ spectrum and the maximum flux level in

948: the light curve.

949: $E_{wz}$ and $Q_{pz}/T_{Lz}$ are correlated and it is this correlation

950: which gives rise to the Amati relation (Amati et al. 2002).

951: The instantaneous maximum brightness

952: of the prompt emission is characterised by a function of the

953: photon energy/peak luminosity ratio $K_{z}$ given

954: in Equation \ref{eq15}. This is not a constant but it covers a remarkably

955: small dynamic range compared with the constituent parameters, $E_{wz}$,

956: $Q_{pz}$ and $T_{Lz}$. Given the measurement errors it is difficult to 

957: make an accurate estimate of the intrinsic dynamic range but it is 

958: certainly less than $0.5<K_{z}<2.0$ and this holds for 100 GRBs in the

959: sample of 101 we have analysed including long, short and XRFs, the

960: exception being GRB980425.

961: 

962: \subsection{Is $K_{z}$ intrinsic?}

963: 

964: We might wonder whether the narrow range in $K_{z}$ is an artefact of the

965: observational data or something instinsic to the nature of the GRB emission?

966: An artificial tightness of the energy-luminosity correlation could arise

967: in several ways; the observed quantities may be correlated by some property 

968: of the instrumentation/measurement,

969: the measured positions of GRBs in the energy-luminosity

970: plane could be incorrect because of some systematic error/bias

971: or GRBs from certain areas in the plane

972: may be selectively missed. The measured quantities in the observer frame which

973: map to $E_{wz}$ and $Q_{pz}/T_{Lz}$ are the peak energy, $E_{p}$ keV, and

974: the spectral energy density at the peak

975: \begin{equation}

976: f_{p}=\frac{N_{tot}}{T_{L}}\exp[(1-\beta_{X})(E_{p}^{-1}-1)] E_{p}^{-\beta_{X}}

977: \label{eq16}

978: \end{equation}

979: ergs cm$^{-2}$ keV$^{-1}$ s$^{-1}$. These are plotted in the top left-hand

980: panel of Figure \ref{fig13}. There is no tight clustering or significant

981: correlation. Pearson's correlation coefficient is $r=0.27$

982: and Kendall's $\tau=0.13$, significance $2.0\sigma$.

983: The $90\%$ range of $f_{p}$ is $4.1\times10^{-10}-2.8\times10^{-8}$

984: ergs cm$^{-2}$ keV$^{-1}$ s$^{-1}$ 

985: and the $90\%$ range of the observed peak energy is

986: $34-412$ keV, both one to two orders of magnitude. Using the

987: redshift to transform these into $Q_{pz}/T_{L}$ and $E_{wz}$ produces the

988: distributions shown in Figure \ref{fig10}. These quantities have slightly

989: narrower distributions in the source frame with $90\%$ ranges of

990: $3.1\times10^{47}-7.8\times10^{49}$ ergs keV$^{-1}$ s$^{-1}$ and

991: $511-3450$ keV respectively.

992: Finally they combine in $K_{z}$ which has a rather narrow

993: $90\%$ range of $0.45-1.95$ and some of this is attributable to the

994: measurement errors. It is very unlikely that some systematic error or bias

995: in the measured quantities which have a large dynamic range and are not

996: correlated

997: conspires to give such a tight correlation and we conclude that $K_{z}$

998: encodes real, useful, information about the source frame properties of

999: the prompt emission. The rate profile fitting not only provides us with 

1000: the peak flux density level but also 

1001: the minumum detected flux density

1002: \begin{equation}

1003: f_{m}=N_{tot}f_{E}\exp[(1-\beta_{X})(E_{p}^{-1}-1)] E_{p}^{-\beta_{X}}

1004: \label{eq17}

1005: \end{equation}

1006: ergs cm$^{-2}$ keV$^{-1}$ s$^{-1}$. The top

1007: right-hand panel of Figure \ref{fig13}

1008: shows $f_{m}$ vs. $f_{p}$. There is some clustering of the weaker bursts

1009: along the line $f_{p}\approx5f_{m}$ and clearly the area above this

1010: line in the top left corner is below the threshold.

1011: It could be that we are preferentially missing hard-dim bursts

1012: while soft-dim bursts are detected

1013: but redshift works in our favour because the distant dim bursts

1014: are redshifted into the lower observation energy band where the

1015: sensitivity is higher and time dilation stretches the light curve so

1016: we have longer to detect the emission. We are undoubtedly

1017: missing low luminosity bursts especially at high redshift.

1018: The lower panels of Figure \ref{fig13} shows $L_{iso}$ and $E_{wz}$

1019: plotted vs. redshift.

1020: The hard (high $E_{wz}$) and most luminous

1021: (high $L_{iso}$) sources are seen at all redshifts

1022: while the softer, weaker sources are only seen at low redshifts,

1023: entirely as one would expect,

1024: but as we can see from Figure \ref{fig12} there is no obvious

1025: difference in the distribution of $K_{z}$ as a function of $z$. There is

1026: no reason to suspect that absence of dim bursts too weak to detect

1027: is biasing the distribution in the photon energy-luminosity plane.

1028: 

1029: Our conclusions are somewhat different from Butler et al. (2007).

1030: The combination of recent {\em Swift} detections with pre-Swift results

1031: confirms the general correlation between $E_{pz}$ and $E_{iso}$ (the Amati

1032: relation) but the spread is indeed large and many bursts, including 

1033: all the short bursts, are extreme outliers from the bulk correlation of

1034: the long bursts. Such a correlation is, in part, a simple consequence of

1035: the shape of a typical GRB spectrum (the Band function)

1036: but the spread and presence of many outliers renders this correlation 

1037: insensitive in testing of cosmological world models.

1038: There is a clustering of events

1039: in the ratio of fluence/duration (effectively luminosity), however, we think 

1040: the root of the problem is not observational bias

1041: or sensitivity thresholding but rather that the Amati relation

1042: (and similar correlations involving $E_{iso}$) are looking at the

1043: wrong parameter space. Yes, this is the demise of the existing pre-{\em Swift}

1044: high-energy correlations, but if we re-cast them in terms of the

1045: instantaneous peak luminosity and we replace $E_{pz}$ by $E_{wz}$, which

1046: combines the spectral parameters $E_{pz}$, $\beta_{X}$ and $\beta_{\gamma}$,

1047: then they reappear in a new light. The quantities in the observer frame

1048: are not correlated, the short and dim bursts are no longer

1049: outliers in the source frame

1050: correlation and there is no difference between {\em Swift}

1051: and pre-{\em Swift} detections.

1052: 

1053: \subsection{Correlations involving evolution of parameters}

1054: 

1055: All the analysis presented here involves average spectral and temporal

1056: properties of the prompt emission.

1057: $E_{wz}$ and $Q_{pz}$ are derived from the time-integrated

1058: spectra and $T_{Lz}$ is estimated from the full energy band light curves.

1059: We know that GRB spectra evolve with time and the light curves are different

1060: in different energy bands. In general the spectra soften as the burst proceeds,

1061: $\beta_{X}$ increases and $E_{p}$ decreases with time (e.g. Goad et al. 2007,

1062: Page et al. 2007). The light curves

1063: are shorter and more spikey at high energies than they are at low energies

1064: (Reichart et al. 2001).

1065: The lag-luminosity (Norris et al. 2000) and variability-luminosity

1066: (Reichart et al. 2001) correlations are testament to this temporal-spectral

1067: evolution. In this work

1068: we have estimated the hardness and brightness of just the peak

1069: emission. If instrumentation could follow the evolution of the

1070: characteristic energy

1071: and luminosity through the light curve each burst would form a track on the

1072: energy-luminosity density plane which may run from top right to bottom

1073: left with $K_{z}\sim constant$.

1074: The lag and variability correlations may provide a means by which

1075: scatter can be introduced

1076: in $K_{z}$ although physical reasons for this are not immediately

1077: apparent. Further analysis and better quality data are

1078: required to explore the evolution of $K_{z}$ through individual bursts.

1079: 

1080: \subsection{Emission processes}

1081: 

1082: The correlation between the hardness and brightness of GRB spectra, 

1083: previously in the form of the Amati relation and now the correlation

1084: between $E_{wz}$ and $Q_{pz}/T_{Lz}$, is a challenge to theoretical

1085: modelling of the prompt emission. Within the standard fireball picture

1086: there are many variants involving internal and external shocks in which

1087: synchrotron emission, inverse Compton scattering and photospheric

1088: emission feature, and the fireball itself may be dominated by kinetic energy

1089: or magnetic energy (Poynting flux). The initial problem is to predict

1090: a spectrum which has the general form of the Band function with a spectral

1091: break or curvature characterised by some energy, $E_{c}$, $E_{p}$ or $E_{wz}$,

1092: and the second problem is to

1093: predict the coupling between the characteristic energy or hardness of the

1094: spectrum and the luminosity (see the review by Zhang \& M\'{e}sz\'{a}ros 2002).

1095: 

1096: With a kinetic energy dominated

1097: outflow and a simple synchrotron model generated by internal shocks,

1098: incorporating a peak in the electron energy

1099: one expects $E_{pz}\propto \Gamma^{-2}t_{var}^{-1}L^{1/2}$ where

1100: $\Gamma$ is the bulk Lorentz factor, $t_{var}$ is the typical variability

1101: time scale associated with the internal shocks

1102: and $L$ is the luminosity (Zhang \& M\'{e}sz\'{a}ros 2002).

1103: This is consistent with the Amati

1104: relation if $L\propto E_{iso}$ (which is not the case if we include

1105: both short and long bursts as shown above)

1106: and there is a constancy of both $\Gamma$ and $t_{var}$

1107: across all bursts, which seems unlikely (Rees \& M\'{e}sz\'{a}ros 2005). 

1108: The relationship $E_{wz}\propto L_{iso}^{0.25}$

1109: derived here and shown to hold for all bursts

1110: is significantly flatter than the Amati relation.

1111: If the Lorentz factor depends on luminosity, $\Gamma\propto L^{\beta}$, then

1112: we can choose $\beta\approx1/8$ to match the observed correlation

1113: providing $t_{var}$ is independent of luminosity and approximately

1114: constant for all bursts. It is not obvious why the Lorentz factor should

1115: have such a specific and low dependence on luminosity and, again,

1116: why $t_{var}$ should be constant when the burst durations ($T_{90z}$ or

1117: $T_{Lz}$) have such a large dynamic range. Within this model the radius

1118: at which the emission occurs is given by $r\sim ct_{var}\Gamma^{2}$ so if

1119: we assume typical values of

1120: $t_{var}\sim0.01$ s and $\Gamma\sim300$, $r\sim3\times10^{13}$ cm.

1121: Furthermore,

1122: by considering the onset of X-ray afterglows observed by {\em Swift}

1123: Kumar et al. (2007) estimate that emission originates at much larger

1124: radii, between $10^{15}$ and $10^{16}$ cm, and suggest that 

1125: synchrotron/inverse Compton parameters cannot account for the prompt 

1126: emission.

1127: 

1128: Alternatively, we can can consider a thermal origin for the peak in

1129: the $E.F(E)$ spectrum and the correlation of the characteristic energy

1130: with luminosity, see for example Rees \& M\'{e}sz\'{a}ros (2005), Ryde (2005)

1131: and Ghirlanda et al. (2007). If the photosphere of the expanding fireball has

1132: radius $R_{0}$, Lorentz factor $\Gamma_{0}$, a blackbody spectral component

1133: with temperature $T_{bb}$ and isotropic luminosity fraction

1134: $\varepsilon_{bb}$ of the total isotropic luminosity $L_{iso}$,

1135: then the observed temperature,

1136: $T_{obs}=(4/3)\Gamma_{0}T_{bb}$, is given by

1137: \begin{equation}

1138: \frac{T_{obs}}{1460\:{\rm keV}}=

1139: \left(\frac{\Gamma_{0}\:10^{7}\:{\rm cm}}{R_{0}}\right)^{1/2}

1140: \left(\frac{\varepsilon_{bb} L_{iso}}{10^{52}\:{\rm ergs\:s^{-1}}}\right)^{1/4}

1141: \label{eq18}

1142: \end{equation}

1143: Thompson (2006). This is just the Stefan-Boltzmann law modified to account

1144: for the relativistic expansion rate of the photosphere and it matches

1145: the observed correlation if $E_{wz}\propto T_{obs}$ and

1146: $\varepsilon_{bb}^{1/2}\Gamma_{0}/R_{0}$

1147: is approximately constant for all bursts. The observed spectrum

1148: is not a single temperature blackbody. If $\varepsilon_{bb}<<1$

1149: then any single temperature blackbody component is diluted by non-thermal

1150: (possibly power law or inverse Compton)

1151: components which combine to give the Band function.

1152: If $\varepsilon_{bb}\sim 1$ the observed spectrum must result from 

1153: the summation of a large number of thermal components with

1154: a spread of temperatures that give an average of $T_{obs}$,

1155: (for example, in a manner similar to that described by Ruffini et al. 2004).

1156: We can re-arrange Equation \ref{eq18} as

1157: \begin{equation}

1158: \left(\frac{\varepsilon_{bb}^{1/2}\Gamma_{0}\:10^{7}\:{\rm cm}}

1159: {R_{0}}\right)^{1/2}=

1160: \left(\frac{T_{obs}}{1460\:{\rm keV}}\right)

1161: \left(\frac{L_{iso}}{10^{52}\:{\rm ergs\:s^{-1}}}\right)^{-1/4}

1162: \label{eq19}

1163: \end{equation}

1164: If $T_{obs}\sim E_{wz}$ this is essentially the same as the

1165: definition of the photon energy-

1166: luminosity quasi-constant, $K_{z}$, and so, under this interpretation,

1167: the scatter in the correlation

1168: arises from variations in the fireball dimension, $R_{0}$, the Lorentz

1169: factor, $\Gamma_{0}$, or the blackbody luminosity fraction, $\varepsilon_{bb}$.

1170: If $K_{z}=1$ then we have an average fireball with

1171: $R_{0}/(\varepsilon_{bb}^{1/2}\Gamma_{0})\approx10^{7}$ cm.

1172: If $K_{z}>1$ the fireball has

1173: a higher than average $\varepsilon_{bb}^{1/2}\Gamma_{0}$

1174: and/or a smaller radius, $R_{0}$.

1175: If $K_{z}<1$ the fireball has a low $\varepsilon_{bb}^{1/2}\Gamma_{0}$

1176: and/or large

1177: radius. With $\Gamma_{0}=300$ and $\varepsilon_{bb}=1$

1178: then $R_{0}\approx3\times10^{9}$ cm which

1179: is the thermalization radius (i.e. the radius of the jet or fireball

1180: photosphere) estimated by Thompson, M\'{e}sz\'{a}ros \&

1181: Rees (2007). This radius is much smaller than estimates arising from

1182: the internal shock model or recent estimates involving the onset of

1183: the X-ray afterglow (see above) so independent estimates

1184: of the radius of the prompt emission and/or the Lorentz factor may help to

1185: discriminate between thermal and internal shock models.

1186: We also note that baryonic photospheres

1187: are governed by physical argument (M\'{e}sz\'{a}ros et al. 2002)

1188: such that the ratio $R_{0}/\Gamma_{0}$ is constrained and the relationship

1189: between $\varepsilon_{bb}^{1/2}\Gamma_{0}/R_{0}$

1190: and $K_{z}$ may not be as simple as we have indicated.

1191: 

1192: \subsection{$K_{z}$ as a cosmological probe}

1193: 

1194: The $E_{wz}-Q_{pz}/T_{Lz}$ correlation should be useful as a pseudo

1195: redshift indicator 

1196: but there is evidence for intrinsic scatter in $K_{z}$, the

1197: errors in determining the characteristic energy and luminosity density are

1198: large and the correlation gradient is rather low, 0.24.

1199: In particular, reliable estimation of 

1200: $E_{wz}$ requires an accurate measurement of the

1201: broadband spectrum and always involves some extrapolation to cover the

1202: source frame energy band 1-10000 keV. Without a good measurement of

1203: the peak energy, $E_{p}$, we can't calculate a good estimate of $Q_{pz}$

1204: from $Q_{z}$. The launch of {\em GLAST} in the near

1205: future will hopefully provide excellent high-energy spectral measurements

1206: which will tie down the spectral parameters more precisely.

1207: The intrinsic scatter may arise in several ways. If the 

1208: dominant emission mechanism is non-thermal then the coupling of 

1209: the Lorentz factor of the expansion with the luminosity and the 

1210: variability time associated with the internal shocks may be the root

1211: cause. If thermal processes dominate then the ratio $R_{0}/\Gamma_{0}$ may

1212: vary as discussed above. Because we can identify classes which

1213: fall predominately at $K_{z}>1$ (shorts) and $K_{z}<1$ (XRFs) there is

1214: some hope that additional parameters which distinguish these

1215: classes may serve to narrow the distribution. In the above

1216: discussion we made no mention of collimation or beaming of the

1217: outflow. Since the Amati relation involving $E_{iso}$ has been transformed

1218: into the present universal correlation involving $L_{iso}$ the simple

1219: beaming argument that underpinned the Ghirlanda relation 

1220: (Ghirlanda et al. 2004) is not directly applicable and currently there

1221: is no simple physical model which links collimation to scatter

1222: in peak luminosity or the characteristic photon energy.

1223: However, it is not

1224: unreasonable to suppose that collimation may introduce scatter in

1225: the peak luminosity, and that correlation of $K_{z}$ with afterglow parameters

1226: such as optical jet break times $t_{break}$ or the time of the

1227: start of the final X-ray afterglow, $T_{a}$  (Willingale et al. 2007),

1228: may be fruitful. 

1229: In its present form $K_{z}$ is not a sensitive cosmological probe

1230: but the signs are that it may be in the future.

1231: 

1232: \section{Conclusion}

1233: 

1234: The equivalent isotropic energy, $E_{iso}$ ergs, of a GRB can be expressed

1235: as the product of two source frame terms, a characteristic photon energy,

1236: $E_{wz}$ keV, calculated from

1237: the shape of the spectrum across the range 1-10000 keV and the energy

1238: density at the

1239: peak of the $E.F_{z}(E)$ spectrum, $Q_{pz}$ ergs keV$^{-1}$. The correlation

1240: trend between $E_{wz}$ and $Q_{pz}$ gives rise to the Amati relation.

1241: By stacking the samples of a GRB light curve into

1242: descending order we can construct a rate profile. 

1243: The functional form of such rate profiles is

1244: common to the vast majority of bursts.

1245: Fitting the profile gives us a

1246: luminosity time, $T_{Lz}$ s, a measure of the burst duration

1247: which can be used to convert the energy

1248: density at the peak to a luminosity density at peak, $Q_{pz}/T_{Lz}$ ergs

1249: keV$^{-1}$ s$^{-1}$. We can calculate the peak equivalent isotropic 

1250: luminosity as a product $L_{iso}=E_{wz}Q_{pz}/T_{Lz}=E_{iso}/T_{Lz}$

1251: ergs s$^{-1}$.

1252: 

1253: $E_{wz}$ is a characteristic photon 

1254: energy or a measure of the colour or hardness of the burst

1255: and $Q_{pz}/T_{Lz}$ is a measure of the instantaneous peak brightness.

1256: We have gathered and analysed

1257: sufficient spectral and temporal data from 101 bursts to produce

1258: the relation between $E_{wz}$ vs. $Q_{pz}/T_{Lz}$ and $E_{wz}$ vs. $L_{iso}$,

1259: shown in Figure \ref{fig11},

1260: which constitutes the closest thing we have to 

1261: an intrinsic colour-magnitude diagram for the peak emission from GRBs,

1262: $E_{wz}\propto L_{iso}^{0.25}$.

1263: All bursts are clustered such that we can construct a intrinsic

1264: colour-magnitude quasi constant $K_{z}$, which is a function of the 

1265: source frame characteristic photon energy/peak luminosity ratio

1266: given by Equation \ref{eq15}.

1267: The range of equivalent isotropic energy that drives the expanding fireball

1268: is very large, 6 orders of magnitude (Figure \ref{fig3}),

1269: but the instantaneous hardness/brightness

1270: of the peak emission covers a very small intrinsic dynamic range, $\approx4$.

1271: 

1272: The existence and form of $K_{z}$ indicates that the physical mechanism

1273: for the Gamma-ray production at the photosphere of the fireball is

1274: common to all bursts and is probably thermal although many other

1275: possibilities are not ruled out. If the prompt spectra

1276: are dominated by thermal photons the scatter in $K_{z}$ may be

1277: attributed to variations in the size and/or Lorentz factor of the fireball.

1278: XRFs have low $\Gamma_{0}$ and/or large radii. Short bursts

1279: have high $\Gamma_{0}$ and/or small radii.

1280: The relation between $T_{Lz}$ vs. $Q_{pz}$ clearly separates

1281: short from long, but both classes have the same instantaneous peak

1282: hardness/brightness. 

1283: 

1284: \acknowledgments

1285: 

1286: We gratefully acknowledge funding for {\em Swift} at the

1287: University of Leicester by STFC. We thank the authors of the

1288: {\em BATSE} ($cossc.gsfc.nasa.gov/docs/cgro/batse$),

1289: {\em HETE} ($space.mit.edu/HETE$) and

1290: {\em BeppoSAX} ($www.asdc.asi.it/grb\_wfc$) websites which gave us

1291: access to the prompt lightcurves of pre-{\em Swift} bursts. We also thank

1292: B. Zhang for valuable comments/discussions.

1293: 

1294: %% To help institutions obtain information on the effectiveness of their

1295: %% telescopes, the AAS Journals has created a group of keywords for telescope

1296: %% facilities. A common set of keywords will make these types of searches

1297: %% significantly easier and more accurate. In addition, they will also be

1298: %% useful in linking papers together which utilize the same telescopes

1299: %% within the framework of the National Virtual Observatory.

1300: %% See the AASTeX Web site at http://www.journals.uchicago.edu/AAS/AASTeX

1301: %% for information on obtaining the facility keywords.

1302: 

1303: %% After the acknowledgments section, use the following syntax and the

1304: %% \facility{} macro to list the keywords of facilities used in the research

1305: %% for the paper.  Each keyword will be checked against the master list during

1306: %% copy editing.  Individual instruments or configurations can be provided 

1307: %% in parentheses, after the keyword, but they will not be verified.

1308: 

1309: %% Appendix material should be preceded with a single \appendix command.

1310: %% There should be a \section command for each appendix. Mark appendix

1311: %% subsections with the same markup you use in the main body of the paper.

1312: 

1313: %% Each Appendix (indicated with \section) will be lettered A, B, C, etc.

1314: %% The equation counter will reset when it encounters the \appendix

1315: %% command and will number appendix equations (A1), (A2), etc.

1316: 

1317: 

1318: \begin{thebibliography}{}

1319: \bibitem[Atteia 2003]{atteia03}Atteia J.-L., 2003, A\&A 407, L1-L4

1320: \bibitem[Atteia et al. 2005]{atteia04}Atteia J.-L., 2005, ApJ 626, 292

1321: \bibitem[Amati et al. 2002]{amat02}Amati L. et al., 2002, A\&A, 390, 81

1322: \bibitem[Amati 2006]{amat06}Amati L., 2006, MNRAS 372, 233

1323: \bibitem[Amati et al. 2006]{amat06b} Amati et al., 2006b, astro-ph/0611189

1324: \bibitem[Amati et al. 2007]{amat07}Amati L., Della Valle, M., Frontera F.,

1325: Malesani D., Guidorzi C., Montanari E. and Pian E., 2007, A\&A, 463, 913

1326: \bibitem[Band et al. 1993]{band93}Band D. et al., 1993, ApJ 413, 281

1327: \bibitem[Butler et al. 2007]{butl07}Butler N.R., Kocevski D., Bloom J.S.,

1328: Curtis J.L., 2007, arXiv0706.1275B

1329: \bibitem[Barthelmy et al. 2005]{bart05}Barthelmy S.D. et al., 2005,

1330: Nature, 438, 994

1331: \bibitem[Bloom et al. 2003]{bloo03}Bloom J.S., Frail D.A. and Kulkarni S.R.,

1332: 2003, ApJ 594, 674

1333: \bibitem[Burrows et al. 2006]{burr06}Burrows D.N. et al., 2006, ApJ 653, 468

1334: \bibitem[Campana et al. 2006]{camp06}Campana S. et al, A\&A, 454, 113

1335: \bibitem[Firmani et al. 2006]{firm06}Firmani C., Ghisellini G.,

1336: Avila-Reese V., Ghirlanda G., 2006, MNRAS, 370, 185

1337: \bibitem[Frail et al. 2001]{frai01}Frail D.A. et al.,

1338: 2001, ApJ, 562, L55

1339: \bibitem[Friedman \& Bloom]{frie05}Friedman A.S. \& Bloom J.S., 2005, ApJ 627, 1

1340: \bibitem[Galassi et al. 2004]{gala04}Galassi M. et al., 2004, GCN 2770

1341: \bibitem[Ghirlanda et al. 2004]{ghir04}Ghirlanda G., Ghisellini G. and

1342: Lazzati D., 2004, ApJ 616, 331

1343: \bibitem[Ghirlanda et al. 2007]{ghir07}Ghirlanda G., Bosnjak Z., Ghisellini G.,

1344: Tavecchio F., Firmani C., 2007, MNRAS, 379, 73

1345: \bibitem[Golenetskii et al. 2005]{gole05} Golenetskii S. et al., 2005, GCN 3474,

1346: GCN 3518, GCN 4150, GCN 4238, GCN 4394, GCN 5264

1347: \bibitem[Guidorzi et al. 2005]{guid05} Guidorzi C., Frontera F., 

1348: Montanari E., Rossi F., Amati L., Gomboc A., Hurley K., Mundell C.G.,

1349: 2005, MNRAS, 363, 315

1350: \bibitem[Goad et al. 2007]{goad07}Goad M.R. et al., 2007, A\&A, 468, 103

1351: \bibitem[Liang \& Zhang 2005]{lian05}Liang E. \& Zhang B., 2005, ApJ 633, 611

1352: \bibitem[Kumar et al. 2007]{kuma07}Kumar P., McMahon E., Panaitescu A.,

1353: Willingale R., O'Brien P., Burrows D., Cummings J., Gehrels N., Holland S.,

1354: Pandey S.B., Vanden Berk D., Zane S., 2007, MNRAS, 376, 57

1355: \bibitem[Lloyd et al. 2000]{lloyd00}Lloyd N.M., Petrosian V. \&

1356: Mallozzi R.S., 2000, ApJ 534, 227

1357: \bibitem[M\'{e}sz\'{a}ros et al. 2002]{mesz02}M\'{e}sz\'{a}ros P.,

1358: Ramirez-Ruiz E., Rees M.J. \& Zhamg B., 2002, ApJ, 578, 812

1359: \bibitem[Mallozzi et al, 1995]{mall95}Mallozzi R.S., Paciesas W.S.,

1360: Pendleton G.N., Briggs M.S., Preece R.D.,

1361: Meegan C.A., Fishman G.J., 1995, ApJ, 454, 597

1362: \bibitem[Nava et al. 2006]{nava06}Nava L., Ghisellini G., Ghirlanda G.,

1363: Tavecchio F., Firmani C., 2006, A\&A, 450, 471

1364: Cabrera J.I., Firmani C. and Avila-Reese V., 2007, MNRAS, submitted

1365: \bibitem[Norris et al. 2000]{norr00}Norris J.P., Marani G.F., Bonnell J.T.,

1366: 2000, ApJ, 534, 248

1367: \bibitem[Norris \& Bonnell 2006]{norr06}Norris J.P. \& Bonnell J.T., 2006

1368: ApJ, 643, 266

1369: \bibitem[Page et al. 2007]{page07}Page K.L. et al. 2007, ApJ, 663, 1125

1370: \bibitem[Pelangeon \& Atteia 2006]{pelan06}Pelangeon A. \& Atteia J.-L.,

1371: 2006, GCN 4442

1372: \bibitem[Piro et al. 2005]{piro05}Piro L. et al., 2005, ApJ, 623, 314

1373: \bibitem[Rees \& M\'{e}sz\'{a}ros 2005]{rees05}Rees M.J. \&

1374: M\'{e}sz\'{a}ros P., 2005, ApJ, 628, 847

1375: \bibitem[Reichart et al. 2001]{reic01}Reichart D.E., Lamb D.Q., Fenimore E.E,

1376: Ramirez-Ruiz E., Cline T.L., Hurley K., 2001, ApJ 552, 57

1377: \bibitem[Romano et al. 2006]{roma06} Romano P. et al., 2006, A\&A, 456, 917

1378: \bibitem[Ruffini et al. 2004]{ruff04} Ruffini R., Bianco, C.L., Xue 

1379: S-S., Chardonnet P., Fraschetti, F., Gursadyan V., Int.J.Mod.Phys.D, 13, 843

1380: \bibitem[Ryde 2005]{ryde05} Ryde F., 2005, ApJ, 625, L95

1381: \bibitem[Schaefer 2004]{scha04}Schaefer B.E., 2004 ApJ, 602, 306

1382: \bibitem[Schaefer 2007]{scha07}Schaefer B.E., 2007, ApJ, 660, 16

1383: \bibitem[Sakamoto et al. 2005]{saka05}Sakamoto T. et al., 2005, ApJ 629, 211

1384: \bibitem[Thompson 2006]{thom06}Thompson, C., 2006, ApJ, 651, 333

1385: \bibitem[Thompson et al. 2007]{thom07}Thompson, C., M\'{e}sz\'{a}ros P.,

1386: Rees M.J., 2007, ApJ, 666, 1012

1387: \bibitem[Villasenor et al. 2005]{vill05} Villasenor J.S. et al., 2005,

1388: Nature 437, 855

1389: \bibitem[Willingale et al. 2007]{willi07}Willingale R. et al., 2007, ApJ 662,

1390: 1093

1391: \bibitem[Yonetoku et al. 2004]{yone04}Yonetoku D., Murakami T., Nakamura T.,

1392: Yamazaki R., Inoue A.K., Ioka K., 2004, ApJ, 609, 935

1393: \bibitem[Zhang \& M\'{e}sz\'{a}ros 2002]{zhan02}Zhang B. \&

1394: M\'{e}sz\'{a}ros P., 2002, ApJ, 581, 1236

1395: 

1396: %\bibitem[Arnaud 1996]{arna96} Arnaud K., 1996, in Jacoby G., Barnes J., eds,

1397: %Astronomical Data Analysis Software and Systems,

1398: %ASP Conf. Series Vol 101, p17

1399: %\bibitem[Aslan et al. 2006]{asla06}Aslan Z. et al., 2006, GCN 3896

1400: %\bibitem[Blustin 2006]{Blus06} Blustin A.J. et al., 2006, ApJ 637, 901

1401: %\bibitem[Burrows et al. 2005]{burr05}Burrows D.N. et al., 2005, Sp. Sc. Rev.,

1402: %in press (astro-ph/0508071)

1403: %\bibitem[Campana et al. 2006]{camp06b} Campana, S. et al., Nature, 442, 1008

1404: %\bibitem[Cenko et al. 2006]{cenk06}Cenko S.B. et al., 2006, astro-ph/0608183

1405: %\bibitem[Crew et al. 2005]{crew05} Crew G. et al., GCN 4021, 2005

1406: %\bibitem[Cummings et al. 2005]{cumm05} Cummings J. et al., GCN 3479, 2005

1407: %\bibitem[Eichler \& Granot 2006]{eich06}Eichler D. \& Granot J., 2006,

1408: %ApJ 641, L5-L8

1409: %\bibitem[Galama et al. 1998]{gala98}Galama T.J. et al., 1998, Nature 395, 670

1410: %\bibitem[Gehrels et al. 2004]{gehr04}Gehrels, N. et al., 2004, ApJ 611, 1005

1411: %\bibitem[Godet et al. 2006]{gode06a}Godet O. et al., 2006a, ApJ, submitted

1412: %\bibitem[Godet et al. 2006]{gode06b}Godet O. et al., 2006b, ApJ, in prep

1413: %\bibitem[Golenetskii et al. 2006]{gole06} Golenetskii S. et al., 2006, GCN 4599,

1414: %GCN 4989

1415: %\bibitem[Grupe et al. 2006]{grup06}Grupe D. et al., 2006, ApJ, in prep

1416: %\bibitem[Harrison et al. 2001]{harr01}Harrison F.A. et al., 2001, ApJ 559, 123

1417: %\bibitem[Hjorth et al. 2003]{hjor03}Hjorth J. et al., 2003, Nature, 423, 847

1418: %\bibitem[Jakobsson et al. 2006]{jako06}Jakobsson P. et al., 2006, GCN 5298

1419: %\bibitem[Kaneko et al. 2006]{kane06}Kaneko Y. et al., 2006, ApJS 166, 298

1420: %\bibitem[Kulkarni et al. 1998]{kulk98}Kulkarni et al. 1998, Nature, 395, 663

1421: %\bibitem[Kumar \& Panaitescu 2000]{kuma00}Kumar P. \& Panaitescu A., 2000,

1422: %ApJ, 541, L51

1423: %\bibitem[Liang \& Zhang 2005]{lian05}Liang E. \& Zhang B., 2005, ApJ 633,

1424: %611-623

1425: %\bibitem[Mangano et al. 2006]{mang06}Mangano V. et al., subitted ApJ,

1426: %astro-ph/0603738

1427: %\bibitem[M\'{e}sz\'{a}ros 2002]{meza02}M\'{e}sz\'{a}ros P., 2002,

1428: %ARA\&A, 40, 137

1429: %\bibitem[M\'{e}sz\'{a}ros et al. 1998]{meza98}M\'{e}sz\'{a}ros P.,

1430: %Rees M.J. \& Wijers R.A.M.J., 1998, ApJ 499, 301

1431: %\bibitem[Metzger et al. 1997]{metz97}Metzger M.R. et al., 1997, Nature,

1432: %387, 878

1433: %\bibitem[Mirabal \& Halpern 2006]{mira06}Mirabal N. \& Halpern J.P., 2006, GCN 4792

1434: %\bibitem[Monfardini et al. 2006]{monf06}Monfardini A. et al., 2006,

1435: %ApJ 648, 1125-1131

1436: %\bibitem[Nousek et al. 2006]{nous06}Nousek, J.A. et al., 2006, ApJ, 642, 389

1437: %\bibitem[O'Brien et al. 2006]{obri06}O'Brien P.T., Willingale R., Osborne J.P.,

1438: %Goad M.R., Page K.L. et al., 2006, ApJ in press.

1439: %\bibitem[Panaitescu \& Kumar 2001]{pana01}Panaitescu A. \& Kumar P., 2001,

1440: %ApJ 554, 667

1441: %\bibitem[Panaitescu \& Kumar 2002]{pana02}Panaitescu A. \& Kumar P., 2002,

1442: %ApJ 571, 779

1443: %\bibitem[Panaitescu \& Kumar 2003]{pana03}Panaitescu A. \& Kumar P., 2003,

1444: %ApJ 592, 390

1445: %\bibitem[Panaitescu et al. 2006]{pana06}Panaitescu A., M\'{e}sz\'{a}ros P.,

1446: %Burrows D. et al., 2006, MNRAS,

1447: %\bibitem[de Ugarte Postigo 2006]{post06}de Ugarte Postigo, A. et al., 2006,

1448: %GCN 5288, 5290

1449: %\bibitem[Rhoads 1997]{rhoa97}Rhoads J.E., 1997, ApJ 487, L1

1450: %\bibitem[Rhoads 1999]{rhoa99}Rhoads J.E., 1999, ApJ 525, 737

1451: %\bibitem[Sakamoto et al. 2006]{sakm06}Sakamoto et al., 2006, ApJ. 636, 73

1452: %\bibitem[Sari 1997]{sari97}Sari R., 1998,  ApJ 489, L37

1453: %\bibitem[Sari, Piran \& Halpern 1999]{sari99}Sari R., Piran T., 

1454: %Halphern J.P., 1999,  ApJ, 524, L43

1455: %\bibitem[Sata et al. 2006]{sato06}Sato G. et al., 2006, ApJ, submitted

1456: %\bibitem[Soderberg et al. 2006]{sode06}Soderberg A.M. et al., 2006,

1457: %Nature, 442, 1014

1458: %\bibitem[Stanek et al. 2003]{stan03}Stanek K.Z. et al., 2003, ApJ, 591, L17

1459: %\bibitem[Stanek et al. 2006]{stan06}Stanek K.Z. et al., 2006, astro-ph/0602495

1460: %\bibitem[Tiengo et al. 2003]{tien03}Tiengo A., Mereghetti S., Ghisellini G.,

1461: %Rossi E., Ghirlanda G. and Schartel W., 2003, A\&A 409, 983

1462: %\bibitem[Wijers \& Galama 1999]{wije99}Wijers R.A.M.J. \& Galama T.J., 1999,

1463: %ApJ 523, 177

1464: %\bibitem[Willingale et al. 2004]{will04}Willingale R., Osborne J.P.,

1465: %O'Brien P.T., Ward M.J., Levan A. and Page K.L., 2004, MNRAS 349, 31-38

1466: %\bibitem[Zhang et al. 2006]{zhan06}Zhang B. et al., 2006, astro-ph/060177,

1467: %ApJ, in press

1468: \end{thebibliography}

1469: 

1470: \clearpage

1471: 

1472: %% Use the figure environment and \plotone or \plottwo to include

1473: %% figures and captions in your electronic submission.

1474: %% To embed the sample graphics in

1475: %% the file, uncomment the \plotone, \plottwo, and

1476: %% \includegraphics commands

1477: %%

1478: %% If you need a layout that cannot be achieved with \plotone or

1479: %% \plottwo, you can invoke the graphicx package directly with the

1480: %% \includegraphics command or use \plotfiddle. For more information,

1481: %% please see the tutorial on "Using Electronic Art with AASTeX" in the

1482: %% documentation section at the AASTeX Web site,

1483: %% http://www.journals.uchicago.edu/AAS/AASTeX.

1484: %%

1485: %% The examples below also include sample markup for submission of

1486: %% supplemental electronic materials. As always, be sure to check

1487: %% the instructions to authors for the journal you are submitting to

1488: %% for specific submissions guidelines as they vary from

1489: %% journal to journal.

1490: 

1491: %% This example uses \plotone to include an EPS file scaled to

1492: %% 80% of its natural size with \epsscale. Its caption

1493: %% has been written to indicate that additional figure parts will be

1494: %% available in the electronic journal.

1495: 

1496: \begin{figure}[!htp]

1497: \begin{center}

1498: \includegraphics[height=15cm,angle=-90]{f1.eps}

1499: \end{center}

1500: \caption{The 1 keV to 10 MeV

1501: source frame spectra of GRBs listed in Table \ref{tab1}. The

1502: observed energy band is shown as the solid line in each case.

1503: $Q_{pz}$ values are marked;

1504: solid dots for long GRBs, solid stars for short GRBs and solid triangles for

1505: XRFs.}

1506: \label{fig1}

1507: \end{figure}

1508: \begin{figure}[!htp]

1509: \begin{center}

1510: \includegraphics[height=15cm,angle=-90]{f2.eps}

1511: \end{center}

1512: \caption{The 1 keV to 10 MeV

1513: $E.F_{z}(E)$ source frame spectra of GRBs listed in Table

1514: \ref{tab1}. The

1515: observed band is shown as the solid line in each case.

1516: $E_{pz}Q_{pz}$ values are marked at the peak energy $E_{pz}$;

1517: solid dots for long GRBs, solid stars for short GRBs and solid triangles for

1518: XRFs.}

1519: \label{fig2}

1520: \end{figure}

1521: \begin{figure}[!htp]

1522: \begin{center}

1523: \includegraphics[height=15cm,angle=-90]{f3.eps}

1524: \end{center}

1525: \caption{Top panels: The distribution of $E_{iso}$ and the correlation

1526: of $E_{pz}$ vs. $E_{wz}$ for GRBs listed in Table \ref{tab1};

1527: solid dots for long GRBs, solid stars for short GRBs, solid triangles for XRFs.

1528: The solid line is $E_{pz}=0.23 E_{wz}$ rather than the best fit correlation

1529: which is a little steeper (see text).

1530: Bottom panels: Comparison of the functional fit to the bolometric integral

1531: $I_{fit}$ and the value calculated by numerical integration of the Band

1532: function over

1533: the interval 1-10000 keV for the GRBs in Table \ref{tab1}. The right-hand

1534: panel shows the distribution of the

1535: ratio of $E_{iso}$ to the value obtained using $I_{fit}$.}

1536: \label{fig3}

1537: \end{figure}

1538: \begin{figure}[!htp]

1539: \begin{center}

1540: \includegraphics[height=15cm,angle=-90]{f4.eps}

1541: \end{center}

1542: \caption{Left-hand panel: The Amati relation for the GRBs in Table \ref{tab1}.

1543: Right-hand panel: $E_{wz}$ vs. $Q_{pz}$ for the same GRBs.

1544: Solid dots for long GRBs, solid stars for short GRBs, solid triangles for

1545: XRFs.}

1546: \label{fig4}

1547: \end{figure}

1548: \begin{figure}[!htp]

1549: \begin{center}

1550: \includegraphics[height=8cm,angle=-90]{f5.eps}

1551: \end{center}

1552: \caption{The rate profile function,

1553: Equation \ref{eq12}.}

1554: \label{fig5}

1555: \end{figure}

1556: \begin{figure}[!htp]

1557: \begin{center}

1558: \includegraphics[height=15cm,angle=-90]{f6.eps}

1559: \end{center}

1560: \caption{Typical prompt emission light curves and the corresponding

1561: rate profiles.

1562: The light curves have been smoothed with a boxcar function of width

1563: $T_{45}$ for display purposes.}

1564: \label{fig6}

1565: \end{figure}

1566: \begin{figure}[!htp]

1567: \begin{center}

1568: \includegraphics[height=15cm,angle=-90]{f7.eps}

1569: \end{center}

1570: \caption{Left-hand panel: The distribution of luminosity times, $T_{L}$ s,

1571: and the $\Sigma$ statistic from the fit.

1572: Right-hand panel: The correlation between rate profile index

1573: $C_{L}$ and the ratio of $T_{45}$ derived directly from the data

1574: and $T_{L}$ derived from the profile function fit.}

1575: \label{fig7}

1576: \end{figure}

1577: \begin{figure}[!htp]

1578: \begin{center}

1579: \includegraphics[height=15cm,angle=-90]{f8.eps}

1580: \end{center}

1581: \caption{Light curves and the corresponding rate profiles for which

1582: the fit statistic $\Sigma$ is high.

1583: The light curves have been smoothed with a boxcar function of width

1584: $T_{45}$ for display purposes.}

1585: \label{fig8}

1586: \end{figure}

1587: \begin{figure}[!htp]

1588: \begin{center}

1589: \includegraphics[height=15cm,angle=-90]{f9.eps}

1590: \end{center}

1591: \caption{Top panels: The correlation between luminosity time $T_{Lz}$ and

1592: $T_{90z}$ and the distribution of $T_{Lz}$. The dashed line indicates

1593: the equality $T_{Lz}=T_{90z}$.

1594: Bottom panels: The distributions of $C_{L}$ and

1595: $\beta_{X}$ for the GRBs in Table \ref{tab1}.}

1596: \label{fig9}

1597: \end{figure}

1598: \begin{figure}[!htp]

1599: \begin{center}

1600: \includegraphics[height=15cm,angle=-90]{f10.eps}

1601: \end{center}

1602: \caption{Top panels: The distributions of $E_{wz}$ keV and $Q_{pz}$

1603: ergs keV$^{-1}$.

1604: Bottom panels: The correlation of $T_{Lz}$ s vs. $Q_{pz}$

1605: and the

1606: distribution of the peak luminosity density, $Q_{pz}/T_{Lz}$

1607: ergs keV$^{-1}$ s$^{-1}$. The distribution for

1608: short bursts is shown as the white histogram.}

1609: \label{fig10}

1610: \end{figure}

1611: \begin{figure}[!htp]

1612: \begin{center}

1613: \includegraphics[height=15cm,angle=-90]{f11.eps}

1614: \end{center}

1615: \caption{Left-hand panel:

1616: The correlation between characteristic energy $E_{wz}$ keV

1617: and the peak luminosity density $Q_{pz}/T_{Lz}$ ergs keV$^{-1}$ 

1618: s$^{-1}$.

1619: Right-hand panel: $E_{wz}$ keV vs. the peak isotropic luminosity,

1620: $L_{iso}=E_{wz}Q_{pz}/T_{Lz}$ ergs s$^{-1}$.

1621: In both plots

1622: solid dots for long GRBs, solid stars for short GRBs, solid triangles for

1623: XRFs. The object with the largest $E_{wz}$, 8600 keV, is GRB050904.}

1624: \label{fig11}

1625: \end{figure}

1626: \begin{figure}[!htp]

1627: \begin{center}

1628: \includegraphics[height=15cm,angle=-90]{f12.eps}

1629: \end{center}

1630: \caption{The distribution of the photon energy/peak luminosity ratio $K_{z}$

1631: corresponding to the scatter about the correlation line in the right-hand

1632: panel of Figure \ref{fig11}. The white histogram shows the distribution

1633: of the 26 pre-{\em Swift} bursts. The curve is the best fit Gaussian

1634: distribution. The right-hand

1635: panel shows $K_{z}$ vs. redshift $z$;

1636: solid dots for long GRBs, solid stars for short GRBs, solid triangles for

1637: XRFs. The horizontal dashed lines indicate the 90\% range. The objects

1638: with unusually high $K_{z}$ are GRB980425/SN1998bw at low $z$ and

1639: GRB050904 at high $z$.}

1640: \label{fig12}

1641: \end{figure}

1642: \begin{figure}[!htp]

1643: \begin{center}

1644: \includegraphics[height=15cm,angle=-90]{f13.eps}

1645: \end{center}

1646: \caption{Top panels: Observer frame parameters

1647: $E_{p}$ and minimum observed flux density, $f_{m}$,

1648: plotted against the peak flux density, $f_{p}$. The solid

1649: line represents $f_{p}=f_{m}$.

1650: Bottom panels: Source frame parameters, peak luminosity $L_{iso}$ and

1651: characteristic energy $E_{wz}$ plotted vs. redshift $z$;

1652: solid dots for long GRBs, solid stars for short GRBs, solid triangles for

1653: XRFs.}

1654: \label{fig13}

1655: \end{figure}

1656: \newpage

1657: \input{tab1}

1658: \input{tab2}

1659: \input{tab3}

1660: \end{document}

1661: