1: \documentclass[preprint2]{aastex}

2: 

3: 

4: \begin{document}

5: 

6: \title{GRBs Light Curves - Another Clue on the Inner Engine}

7: 

8: \author{E. Nakar \& T. Piran}

9: \affil{Racah Institute for Physics, The Hebrew University, Jerusalem 91904, ISRAEL}

10: 

11: 

12: \begin{abstract}

13: The nature of  the `inner engine' that accelerate and collimate

14: the relativistic flow at the cores of GRBs is the most

15: interesting current puzzle concerning GRBs. Numerical simulations

16: have shown that the internal shocks' light curve reflects the

17: activity of this inner engine. Using a simple analytic toy model

18: we clarify the relations between the observed $ \gamma $-rays

19: light curve and the inner engine's activity and the dependence of

20: the  light curves on the inner engine's parameters. This simple

21: model also explains the observed similarity between the observed

22: distributions of pulses widths and the intervals between pulses

23: and the correlation between the width of a pulse and the length

24: of the preceding interval. Our analysis suggests that the

25: variability in the  wind's Lorentz factors arises due to a

26: modulation of the mass injected into a constant energy flow.

27: \end{abstract}

28: 

29: \section{Introduction}

30: 

31: According to the current Fireball model a Gamma-Ray Burst (GRB)

32: contain four stages: (i) A catastrophic event produces an `inner

33: pine

34:  engine'. (ii) This `inner engine' accelerates a barionic wind

35: into a highly relativistic motion. (iii) Internal collisions

36: within this wind produce the prompt $ \gamma  $-ray emission (iv)

37: An external-shock between the wind and the surrounding matter

38: produces the afterglow. The most mysterious part of this process

39: is the nature of the `inner engine'. There are no direct

40: observations of the `inner engine'.

41: 

42: Numerical simulations (Kobayashi, Piran \& Sari 1997; Ramirez \&

43: Fenimore, 2000) revealed that the $ \gamma  $-rays light curve

44: replicates the temporal activity of the 'inner engine'. In this

45: letter we explain, using a simple analytic toy model, these

46: results. In our toy model the relativistic wind is described as a

47: sequence of discrete shells with various Lorentz factors, $

48: \gamma  $. The $ \gamma  $-rays light curve results from

49: collisions between shells with different values of $ \gamma  $.

50: We show that the observed time of a pulse (resulting from such a

51: collision) reflects the time that the inner faster shell was

52: ejected from the `inner engine'. We show that unless the

53: background noise prevents the detection of some pulses, the light

54: curve reflects one third to one half of the shells ejection time.

55: 

56: Nakar \& Piran (2001) discovered that the pulses width, $ \delta

57: t $, and the intervals between pulses, $ \Delta t $, have a

58: similar distributions. Moreover, the duration of an interval

59: between pulses is correlated with the width of the following

60: pulse. Our analysis shows that in an internal shocks model with

61: equal energy shells both the intervals and the pulses' widths

62: reflect the initial separation between the shells.  Therefore,

63: both observational results arise naturally in this model. If

64: instead the shells' mass is constant  then the intervals still

65: reflects the shells' separation but the pulses widths depend also

66: on the distribution of the shells' Lorentz factors. In this case

67: the variance in $ \gamma  $ wipes out both the $ \delta t $-$

68: \Delta t $ similarity and the correlation.

69: 

70: Our toy model includes some simplifying assumptions. We confirm

71: these results using  numerical simulations.  The equal energy

72: simulation fit the observations very well, while the equal mass

73: model does not fit the observations. These results suggest that

74: the `inner engine' produces a variable Lorentz factor flow by

75: modulating the mass of a constant energy flow. These results

76: provide yet another strong support to the Internal Shock model.

77: They also give one of the first clues on the nature of the `inner

78: engine'.

79: 

80: 

81: \section{The Toy Model\label{Analytical model}}

82: 

83: In our toy model the `inner engine' emits relativistic shells

84: which collides and produce the observed light curve. We make the

85: following simplifying assumptions: (i) The shells are  discrete

86: and homogeneous. Each shell has a well define boundaries and a

87: well defined $ \gamma $. (ii) The colliding shells merge  into a

88: single shell after the collision. (iii) Only efficient collisions

89: produce an observable pulse. The efficiency, $ \varepsilon $, is

90: defined as the ratio between the post shock internal energy and

91: the total energy. We consider only collisions with $ \varepsilon

92: > 0.05$.

93: 

94: Under these assumptions each shell is defined by four parameters

95: $ t_{i} $, $ m_{i} $, $ \gamma _{i} $ and $ l_{i} $, where $ i $

96: is the shell index, $ t $ is the ejection time of the shell, and

97: $ m $, $ \gamma  $, and $ l $ are the mass, Lorentz factor and

98: width of the shell respectively. For convenience we define $

99: L_{i,j} $, the interval between the rear end of the i'th shell

100: and the front of the j'th shell. Note that\footnote{Hereafter we

101: take c=1. This equality is only approximate since the shells'

102: velocity is almost (but not exactly) c. } $ L_{i,i+1}\approx

103: t_{i+1}-(t_{i}+l_{i})$.

104: 

105: 

106: \subsection{A single collision}

107: 

108: Consider a single collision between two shells with widths $ l_{1}

109: $, $ l_{2} $, a separation $ L $, and ejection times $

110: t_{2}\approx t_{1}+(l_{1}+L) $. We define $ \gamma _{1}\equiv

111: \gamma  $ and $ \gamma _{2}\equiv a\gamma  $ ($ a>1 $). The

112: collision efficiency depends strongly on $ a $ (Piran 1999). $

113: \varepsilon \, (a=2)\approx 0.05 $ , and it decreases fast with

114: decreasing $ a $. Hence, we consider only collisions with $ a>2 $.

115: 

116: The collision takes place at: $ R_{s}\approx \gamma

117: ^{2}L\frac{2a^{2}}{a^{2}-1}\approx 2\gamma ^{2}L $ ($ R_{s} $ is

118: measured in the rest frame). Note that as long as $ a>2 $, $

119: R_{s} $  depends rather weakly on $ \gamma _{2} $. The emitted

120: photons from the collision reach the observer at time (omitting

121: the photons flight time):

122: \begin{equation}

123: \label{to} t_{obs}\approx t_{1}+l_{1}+R_{s}/(2\gamma ^{2})\approx

124: t_{1}+l_{1}+L\approx t_{2}

125: \end{equation}

126: The photons from the collision are observed almost simultaneously

127: with an hypothetical photon emitted from the `inner engine'

128: together with the faster shell (at $ t_{2} $). This result is

129: accurate up to an error of $ L(a^{2}-1)^{-1} $ , which is small

130: compared to $ L $ as long as $ a>2 $.

131: 

132: 

133: \subsection{Multiple collisions}

134: 

135: The light curve obtained from multiple collisions can be

136: described in terms of three basic pairs of collisions (see fig.

137: \ref{types fig}): (I) Two collisions between four consequent

138: shells with $ \gamma _{2}=a\gamma _{1} $ and $ \gamma

139: _{4}=b\gamma _{3} $. The collision are between the first and the

140: second shells and between the third and the forth shells. (II)

141: Two collisions between three consequent shells with $ \gamma

142: _{1}=\gamma _{2}/a=\gamma _{3}/b $. The two front shells collide

143: and then the third shell collides with the merged one. (III) Same

144: as type II but here the rear shells collide first.

145: \begin{figure}

146: {\par\centering

147: \resizebox*{0.95\columnwidth}{!}{\includegraphics{f1a.eps}}

148: \par} \caption{\label{types fig}The basic types of multiple collisions. }

149: \end{figure}

150: 

151: 

152: 

153: \subsubsection{Interval between consequent pulses}

154: 

155: Type I collisions result in two observed pulses, the first at $

156: t_{2} $ and the second at $ t_{4} $. The interval between the

157: pulses, $ \Delta t $, would be: $ \Delta t\approx

158: t_{4}-t_{2}\approx l_{2}+L_{2,3}+l_{3}+L_{3,4} $. In type II

159: collisions the first pulse is observed at $ t_{2} $, and second

160: pulse is observed at $ t_{3} $. The interval is: $ \Delta

161: t\approx t_{3}-t_{2}\approx l_{2}+L_{2,3} $. In type III

162: collisions the last shell takes over the second one, and the

163: first pulse is observed at $ t_{3} $. Then the merged shell takes

164: over the first

165: one releasing another pulse observed at\footnote{%

166: Detailed calculations show that the interval between these two

167: pulses is shorter then the pulses' widths. } $ \sim t_{3} $.

168: Therefore type III collisions, results in a single wide pulse.

169: 

170: All these results are accurate up to an order of $ 1/a_{i,j}^{2}

171: $, where $ a_{i,j} $ is the ratio of the Lorentz factors between

172: the two colliding  shells. As long as the collisions are efficient,

173: these results  depend weakly on the mass distribution of the

174: shells.

175: 

176: 

177: \subsubsection{The pulses width}

178: 

179: The relevant time scales that determine the pulse width are

180: (Piran, 1999): (i) The angular time, $ t_{ang} $, which results

181: from the spherical geometry of the shells: $ t_{ang}\approx

182: R_{c}/2\gamma ^{2}_{sh} $. (ii) The hydrodynamic time, $ t_{hyd}

183: $, which arises from the shell's width and the shock crossing

184: time: $ t_{hyd}\approx l_{c_{in}} $, where $ l_{c_{in}} $ is the

185: width of the inner shell at the time of the collision.  (iii) The

186: cooling time - the time that it takes for the emitting electrons

187: to cool. For synchrotron emission with typical parameters of

188: internal shocks this time is much shorter then $ t_{ang} $ and $

189: t_{hyd} $ (Kobayashi et. al. 1997, Wu \& Fenimore 2000). Therefore

190: under the assumption of transparent shells the pulse width is $

191: \delta t\approx t_{ang}+t_{hyd} $.

192: 

193: Unlike the pulse's timing, the pulse's width depends strongly on

194: the shells' masses. The relevant Lorentz factor for the

195: calculation is the one of the shocked, and therefore radiating,

196: region - $ \gamma _{sh} $. $ \gamma _{sh} $ depends strongly on

197: the ratio of the shells' masses. We examine two possible cases:

198: equal mass shells and equal energy shells. Table \ref{dt-Dt

199: table} summarizes the intervals and the pulses' width for the two

200: different mass distributions for  the three types of collisions.

201: 

202: 

203: \begin{table*}

204: {\centering \begin{tabular}{|c|c|c|c|c|c|}

205: \hline

206: &

207: &

208: \multicolumn{2}{|c|}{{\footnotesize Equal mass}  }&

209: \multicolumn{2}{|c|}{ {\footnotesize Equal energy}  }\\

210: \hline & {\footnotesize $ \Delta t $}& {\footnotesize $ \delta

211: t_{1} $}& {\footnotesize $ \delta t_{2} $}& {\footnotesize $

212: \delta t_{1} $}&

213: {\footnotesize $ \delta t_{2} $}\\

214: \hline {\footnotesize Type I}& {\footnotesize $

215: l_{2}+L_{2,3}+l_{3}+L_{3,4} $}& {\footnotesize $

216: l_{c_{2}}+\frac{L_{1,2}}{a} $}& {\footnotesize $

217: l_{c_{4}}+\frac{L_{3,4}}{b} $}& {\footnotesize $

218: l_{c_{2}}+L_{1,2} $}&

219: {\footnotesize $ l_{c_{4}}+L_{3,4} $}\\

220: \hline {\footnotesize Type II}& {\footnotesize $ l_{2}+L_{2,3}

221: $}& {\footnotesize $ l_{c_{2}}+\frac{L_{1,2}}{a} $}&

222: {\footnotesize $

223: l_{c_{3}}+\frac{L_{1,2}}{b\sqrt{a}}+\frac{L_{2,3}\sqrt{a}}{b} $}&

224: {\footnotesize $ l_{c_{2}}+L_{1,2} $}&

225: {\footnotesize $ l_{c_{3}+}\frac{\sqrt{2}}{5}L_{1,2}+L_{2,3} $}\\

226: \hline {\footnotesize Type III}& {\footnotesize $

227: \frac{L'_{1,2}}{(ab-1)} $}& {\footnotesize $

228: l_{c_{3}}+\frac{L_{2,3}}{a} $}& {\footnotesize $

229: >a\sqrt{\frac{a}{b}}L_{2,3}+\frac{L'_{1,2}\sqrt{ab}}{(ab-1)} $}&

230: \multicolumn{2}{|c|}{ {\footnotesize No efficient collisions }}\\

231: \hline

232: \end{tabular}\footnotesize \par}

233: 

234: \caption{\label{dt-Dt table}  \protect$ \Delta t\protect $, and

235: \protect$ \delta t\protect $ for the three collisions types for

236: the equal energy/mass shells. In each case there are two pulses.

237: \protect$ \delta t_{1}\protect $ {[}\protect$ \delta

238: t_{2}\protect ${]} is the width of the first {[}second{]}

239: observed pulse. \protect$ l_{i}\protect $ and \protect$

240: l_{c_{i}}\protect $ are the width at the ejection and width  at

241: the collision of the i'th shell  ($l_i=l_{ci}$ with no

242: spreading.). \protect$ L'_{1,2}\protect $ is the separation

243: between the first and the second shells at the time that the

244: second and the third shells collide. The approximation here are

245: valid when the Lorentz factors ratio between the colliding shells

246: is larger then 2. }

247: \end{table*}

248: 

249: 

250: \begin{figure}{t}

251: {\par\centering \resizebox*{0.9\columnwidth}{0.21\textheight}

252: {\includegraphics{f2.eps}} \par} \caption{\label{dist fig} Pulses

253: width, \protect$ \delta t\protect $, intervals between pulses,

254: \protect$ \Delta t\protect $, and the separation between shells,

255: $L$. (a) : Equal mass shells. (b): Equal energy shells. }

256: \end{figure}

257: 

258: \section{Numerical Simulation}

259: 

260: The toy model demonstrates that the properties of the light curve

261: depend on the dominant type of collisions. In order to determine

262: what are the dominant collisions types we performed numerical

263: simulations of internal shocks light curves. These simulations

264: also enable us to verify  some of the approximations used in the

265: analytic toy model.

266: 

267:  Following Kobayashi et. al. (1997) each

268: shell  is defined by four parameters $ \gamma _{i},\, m_{i},\,

269: l_{i} $ and $ L_{i,i+1} $. The distribution of the separation

270: between the shells, $ L $, is taken to be  lognormal with $ \mu (ln(L))

271: =-0.5 $ and $ \sigma (ln(L)) =0.9 $ (chosen in order to fit the

272: observations). The initial shells width is taken as a constant of

273: 0.1lsec and we assume that the shells do not spread ($ l_{c}=l $).

274: The Lorentz factor distribution is uniform ($ \gamma _{min}=30 $,

275: $ \gamma _{max}=2000 $). The shells' mass is either constant

276: (equal mass model) or proportional to $ \gamma ^{-1} $ (equal

277: energy model).

278: 

279: 

280: 

281: Each simulation included 50 shells. Following the shell's motion

282: we identify the collisions. Each collision produce a pulse. The

283: duration of a pulse is taken as $ t_{ang}+t_{hyd} $. All the

284: pulses has a fast rise slow decay shape with a ratio of 3:1

285: between the decay and the rise times. The area below a pulse

286: equals to its radiated energy (no assumption is made here about

287: the efficiency). Using these pulses we prepare a binned (64ms

288: time bins) light curve. We analyze this light curve using the Li

289: \& Fenimore (1996) peak finding algorithm, obtaining the observed

290: pulses timing and width.

291: 

292: 

293: \subsection{Numerical Results}

294: 

295: In both models the number of observed pulses is between a third

296: and a half of the total shells emitted. Types I \& II collisions

297: are the dominant types (about 80\%) in the simulations. The total

298: efficiency in  both models is about 20-30\%. The equal mass model

299: prefers  type I collisions while the equal energy model prefer

300: type II collisions. Efficient type III collisions almost don't

301: exist in the equal energy model.

302: 

303: Figure \ref{dist fig} illustrates the histograms of the pulses

304: width, $ \delta t $, the interval between pulses, $ \Delta t $,

305: and the separation between shells, $ L $. In the equal mass model

306: (fig. \ref{dist fig}a) there is no $ \Delta t $-$ \delta t $

307: similarity. $ \Delta t $ reflects the $ L $ distribution while $

308: \delta t $ is much shorter and does not reflects $ L $. In the

309: equal energy model (fig. \ref{dist fig}b) both distributions of $

310: \Delta t $ and $ \delta t $ reflects the `inner engine' shells

311: separation distribution. Both distributions are consistent with a

312: log normal distribution and the best fits parameters are

313: described in table \ref{sim_res_tbl}.

314: \begin{table}

315: {\centering \begin{tabular}{|c|c|c|c|c|} \hline &

316: \multicolumn{2}{|c|}{\( \Delta t\, (sec) \)}&

317: \multicolumn{2}{|c|}{\( \delta t\, (sec) \)}\\

318: \hline & \multicolumn{1}{|c|}{\( \mu  \)}& \( 1\sigma  \)& \(

319: \mu  \)&

320: \( 1\sigma  \)\\

321: \hline Simulated& 1.4& 0.6-3.4& 1&

322: 0.5-2\\

323: \hline Observed& 1.3& 0.5-3.1& 1&

324: 0.5-2.2\\

325: \hline

326: \end{tabular}\par}

327: \caption{\label{sim_res_tbl}Best fit parameters of the pulses'

328: widths and the intervals in the equal energy model, compared to

329: the observed values ( Nakar \& Piran 2001). Note that this fit

330: was achieved by tuning only $\mu (ln(L))$ and $\sigma(ln(L))$.}

331: \end{table}

332: 

333: 

334: Both results are in a perfect agreement with the analytical

335: results obtained in section \ref{Analytical model}. The

336: similarity between the {}``observed{}'' interval distribution in

337: both models is explained by the weak dependence of the pulses'

338: timing on the mass distribution. The pulses width in the equal

339: energy model reflection of the shells initial separation, $ L $,

340: are explained by the analytical result $ \delta t\propto L $. The

341: deviation from a lognormal distribution and the short pulses in

342: the equal mass model are explained by the analytical result $

343: \delta t\propto L/a $.

344: 

345: Nakar \& Piran (2001) find a correlation between an interval

346: duration and the following pulse. In the equal energy model we

347: find a highly significant correlation between the interval

348: duration and the following pulse. There is no significant

349: correlation in the equal mass model. This result is explained

350: again by the equal mass relation $ \delta t\propto L/a $. As the

351: variations in $ a $ are larger then the variations in $ L $ it

352: wipe out the correlation.

353: 

354: 

355: \section{Discussion}

356: 

357: Former numerical simulations of internal shocks have shown that

358: the resulting light curves reflect the activity of the inner

359: engine. We have shown that this feature arise from the fact that

360: the pulse timing is approximately equal to the ejection time of

361: one of the colliding shell. Moreover, in most collisions (Types I

362: \& II) the pulses are distinguishable and each pulse reflects a

363: single collision. In all models the number of observed pulses is

364: 30\%-50\% of the number of ejected shells. Therefore the light

365: curve reflects the emission time of one third to one half of the

366: shells. The `inner engine' is slightly more variable then the

367: observed light curve.

368: 

369: The observed similarity between the $ \Delta t $ and $ \delta t $

370: distributions is explained naturally in the equal energy shells'

371: model. Both parameters reflects the the separation between the

372: shells during their ejection. In the equal mass shells' model

373: only $ \Delta t $ reflects the initial shells' separation and

374: therefore such a similarity is not expected.

375: 

376: Our numerical simulations confirmed these predictions.  Note that

377: many of the simplifying assumptions can be relaxed with no

378: significant change in the results. We present elsewhere a  more

379: detailed model and more elaborated simulations (Nakar \& Piran,

380: 2002). The equal energy model simulations fit the observations

381: very well. These results imply that the `inner engine' ejects,

382: most likely, equal energy shells. These results are yet another

383: strong support to the Internal Shock model. They also give one of

384: the first clues on the nature of the `inner engine'.

385: 

386: This research was supported by a US-Israel BSF grant.

387: 

388: \begin{thebibliography}{1}

389: \bibitem{1}Kobayashi S., Piran T. \& Sari R., 1997, ApJ, 490, 92

390: \bibitem{2}Li H. \& Fenimore E., 1996, ApJ, 469, L115

391: \bibitem{3}Nakar, E. \& Piran, T., 2001, MNRAS in press (astro-ph/0103210)

392: \bibitem{4}Nakar E. \& Piran, T., 2002 in preperation

393: \bibitem{5}Piran, T., 1999, PhR, 314, 575

394: \bibitem{6}Ramirez-Ruiz E. \& Fenimore E. E., 2000, ApJ, 539, 712

395: \bibitem{7}Wu, B., \& Fenimore, E., 2000 ApJ, 535, L29

396: \end{thebibliography}

397: 

398: \end{document}

399: