1: \documentstyle[11pt,aaspp4,flushrt]{article}

2: %\documentstyl[11pt]{article}

3: %\documentstyle[12pt,aasms4]{article}

4: \newcommand\beq{\begin{equation}}

5: \newcommand\eeq{\end{equation}}

6: 

7: \begin{document}

8: 

9: \title{Jets in GRBs: Tests and Predictions for the Structured Jet Model}

10: 

11: \author{Rosalba Perna\altaffilmark{1,2,3}, Re'em Sari\altaffilmark{3} 

12: and Dale Frail\altaffilmark{4}}

13: 

14: 

15: \altaffiltext{1}{Harvard Society of Fellows, 74 Mount Auburn Street, Cambridge, MA 02138}

16: 

17: \altaffiltext{2}{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,

18: Cambridge, MA 02138}

19: 

20: \altaffiltext{3}{130-33, California Institute of Technology, Pasadena, CA 91125}

21: 

22: \altaffiltext{4}{National Radio Astronomy Observatory, P.O. Box O, Socorro, NM, 87801}

23: 

24: \begin{abstract}

25: 

26: The two leading interpretations of achromatic breaks that are observed

27: in the light curves of GRBs afterglow are (i) the manifestation of the

28: edge of a jet, which has a roughly uniform energy profile and a sharp

29: edge and (ii) a line of sight effect in jets with a variable energy

30: profile. The first scenario requires the inner engine to produce a jet

31: with a different opening angle each explosion, while the latter

32: requires a standard engine.  The physical structure of the jet is a

33: crucial factor in understanding GRB progenitors, and therefore

34: discriminating the two jet scenarios is particularly relevant.  In the

35: structured jet case, specific predictions can be made for the

36: distribution of observed break angles $\theta_{\rm break}$, while that

37: distribution is arbitrary in the first scenario. We derive the

38: theoretical distribution for the structured jet model. Specifically,

39: we predict the most common angle to be about 0.12 rad, in rough

40: agreement with the sample. If this agreement would hold as the sample

41: size increases, it would strengthen the case for the standard jet

42: hypothesis. We show that a prediction of this model is that the

43: average viewing angle is an increasing function of the survey

44: sensitivity, and in particular that a mission like {\em Swift} will

45: find the typical viewing angle to be about 0.3 rad.  The local

46: event rate predicted by this model is $R_{\rm GRB}(z=0)\sim 0.5 $

47: Gpc$^{-3}$ yr$^{-1}$.

48: 

49: \end{abstract}

50: 

51: \keywords{gamma rays: bursts --- cosmology: theory}

52: 

53: \section{Introduction}

54: 

55: The degree to which gamma-ray bursts (GRBs) and their afterglows are

56: beamed is an important issue. A proper understanding of the geometry

57: of the relativistic outflow affects the total energetics of GRB

58: central engines and the GRB event rates, both of which are crucial

59: parameters for constraining possible progenitor models.  Evidence for

60: non-isotropic outflows is believed to come from observations of

61: achromatic breaks in afterglow light curves (e.g., Rhoads 1999, Sari

62: et al. 1999) and the detection of polarized emission (Covino et

63: al. 1999, Wijers et al. 1999).  We should however note that, whereas

64: this is a natural interpretation for the breaks, in some cases other

65: interpretations have been proposed or different conclusions derived

66: (e.g. Nicastro et al. 1999; Vrba et al. 2000).  Here we adopt the

67: point of view that the observed breaks are indeed manifestations of

68: jets.

69: 

70: In early theoretical papers (e.g., Rhoads 1997, Sari et al. 1999) it

71: was assumed for simplicity that the ejecta had to be distributed

72: approximately uniform across the entire opening angle in the gamma-ray

73: phase and that the majority of the explosive energy in the afterglow

74: phase must have a single bulk Lorentz factor. In this ``uniform''

75: model, a break occurring in the light curves at $t_{\rm break}$ can be

76: directly translated into a jet with an opening angle $\theta_{\rm

77: break}$. Using this simple framework, Frail et al.~(2001) carried out

78: an analysis of all known afterglows and found that there was a

79: distribution of jet opening angles leading to a reduction in the

80: gamma-ray energy from its isotropic value with relatively small

81: scatter. The observed distribution was shown to be heavily weighted

82: towards small opening angles.

83: 

84: Rather than positing a uniform jet, it is equally reasonable to assume

85: that GRB jets are structured in some fashion. In the collapsar

86: progenitor model (e.g. Wang, Woosley \& MacFadyen 2002) the Lorentz

87: factor and energy are high near the rotation axis, but decrease off

88: axis as the degree of entrainment increases. Salmonson (2000) has

89: argued for such a jet structure to explain the empirical correlation

90: between the GRB peak luminosity and pulse lag (Norris et al. 2000). In

91: this case, the distribution of observed break times in afterglow light

92: curves are not due to a distribution of opening angles but rather

93: originate from variations in the viewing angle of a structured jet

94: (Postnov, Prokhorov \& Lipunov 2001). In two recent papers by Rossi,

95: Lazzati \& Rees (2002a) and Zhang \& Meszaros (2002) it was shown that

96: a jet with a universal beaming configuration could reproduce the

97: near-constant energy result of Frail et al.~(2001) provided that the

98: energy per unit solid angle (and Lorentz factor) varied as the inverse

99: square of angular distance from the jet symmetry axis.

100: 

101: Discriminating between the uniform and structured jet models is

102: important as they yield different estimates for the true GRB event

103: rate and the total energy, besides leading to clues on the physical

104: mechanism producing the jet itself. Afterglow lighcurves have argued

105: to be degenerate to the structure of the jet (Rossi et al. 2002a,

106: Zhang \& Meszaros 2002) and therefore they cannot be used as

107: diagnostics.  Rossi et al. (2002b) showed that detection of

108: polarization can provide useful constraints.  Here we concentrate on

109: diagnostics based on geometrical effects.  In particular, in the

110: structured jet model, since the inferred opening angle is just a

111: geometric effect of the viewing angle, it is possible to predict the

112: distribution of angles and compare it to the {\it observed}

113: distribution of $\theta_{\rm break}$ by Frail et al. (2001) and Bloom

114: et al.~ (2003).  Unfortunately, a specific prediction is not possible

115: for the uniform model, since there is no framework for jet formation

116: which yields the distribution of opening angles.

117: 

118: In this paper, we work out the expected distribution of burst

119: opening angles, under the structured jet scenario and given various

120: assumption of the star formation rate evolution in the universe.  We

121: find two competing effects. First, even though randomly oriented

122: bursts would be rarely observed on axis, they are much brighter and

123: therefore can be seen to larger distances rendering small opening

124: angles common. Cosmological effects limit the volume at large

125: redshifts leading to an effective cutoff at the small opening angle.

126: As a result, we predict that the most common opening angle should be

127: about 0.12 rad. Furthermore, under the structured jet model, we

128: predict that more sensitive future missions like SWIFT, will find a

129: much larger typical angle, about 0.3 rad.

130: 

131: The observed data so far is still a small sample. Selection effects

132: are hard to quantify, especially for large and small opening

133: angles. In addition, the variety of instruments with different

134: sensitivity used to detect the bursts, makes a robust conclusion

135: difficult at this stage. Yet, we show that the current set of data, as

136: given in the most updated sample of Bloom et al. (2003), is at least

137: in rough agreement with the distribution we predict for the structured

138: jet model.  If this conclusion would hold with much larger and less

139: biased sample, it would be a strong support to the structured jet

140: model.

141: 

142: \section{Computation of the observed distribution of jet angles}

143: 

144: \subsection{Scalings for a Euclidian universe}

145: 

146: If all bursts were observable, then we would expect that the number

147: $dn(\theta)/d\theta$ of bursts with angle in the interval $d\theta$

148: around $\theta$ would be proportional to $\theta$. This implies that

149: most of the observed bursts should have a large angle, which is in

150: complete contradiction with observations. Zhang \& Meszaros (2002)

151: attributed this apparent discrepancy to the small sample size or

152: afterglow selection effects. However, this argument does not account

153: for the fact that bursts with small $\theta$ are brighter by a factor

154: of $\theta^{-2}$, and therefore can be seen (in an Euclidian universe)

155: up to a distance $\theta^{-1}$ farther, which contains a volume larger

156: by a factor of $\theta^{-3}$. The expected distribution in Euclidean

157: geometry is therefore expected to be $dn/d\theta \propto

158: \theta^{-2}$. Although this is closer to what observations suggest, we

159: will show that this now exaggerates the number of small-angle bursts

160: compared to a proper cosmological prediction: GRBs originate at

161: redshifts of order unity, and therefore suffer considerable

162: cosmological corrections. This is why their log $N$ - log $S$ curve

163: does not obey the Euclidian $S^{-3/2}$, but is shallower at low

164: $S$. It is for the same reason that the number of observed bursts of

165: low $\theta$ will not be as high as predicted by the Euclidian

166: $\theta^{-2}$. In the following, we work out these cosmological

167: effects in detail.

168: 

169: \subsection{Cosmological effects}

170: 

171: Let $R_{\rm GRB}(z)$ be the GRB rate per unit comoving volume per unit

172: time, then the total (i.e. over the all sky) rate of bursts with

173: inferred jet angle between $\theta$ and $\theta\; + \;d\theta$ is

174: given by

175: \begin{equation}

176: \frac{dn(\theta)}{d\theta} = \sin\theta \int_0^{z_{\rm max}(\theta)}

177: dz\; \frac{R_{\rm GRB}(z)}{(1+z)}\; \frac{dV(z)}{dz}\;,

178: \label{eq:dndt}

179: \end{equation}

180: where $z_{\rm max}(\theta)$, is the maximal redshift up to which we

181: can observe a burst with apparent angle $\theta$. This redshift is

182: found by numerically inverting the equation

183: \begin{equation}

184: F_{ph, \rm lim}=\frac{L_{ph}(\theta)}{4\pi D^2(z_{\rm max})(1+z_{\rm

185: max})^\alpha}\;,

186: \label{eq:flim}

187: \end{equation}

188: where $F_{ph, \rm lim}$ is the limiting photons flux (photons per unit

189: area per unit time) that is detectable by the GRB detector with

190: frequency range, $\nu_l<\nu<\nu_u$. $L_{ph}(\theta)$ is the photons

191: luminosity (photons per unit time) in the same frequency rage but in

192: the local frame of the burst, of a burst with an apparent angle

193: $\theta$. A factor of $(1+z)^{\alpha-1}$ is a spectral correction,

194: assuming that the GRB has a differential photon spectral index

195: $\alpha$, and another factor of $1+z$ takes care of time dilation.

196: For BATSE, $\nu_l=50$ keV and $\nu_u =300$ keV.

197: 

198: The normalization constant is determined by the condition

199: \begin{equation}

200: \frac{L_{\nu_1-\nu_2}}{4\pi}\,2\pi\theta^2 T=E

201: \end{equation} 

202: where $E \approx 10^{51}\;{\rm ergs}$ is the roughly constant 

203: energy of GRBs as inferred by (Bloom et al. 2003). 

204: That relation used the luminosity integrated over the

205: frequency range $\nu_1=20$ keV to $\nu_2=2000$ keV as calculated by

206: Bloom et al. (2001). For powerlaw photon spectrum, this is related to

207: the photon luminosity in the triggering band $\nu_l$ $\nu_h$ by

208: \begin{equation}

209: L_{ph}(\theta)={2E \over \theta^2 T h\nu_l} {\alpha-2 \over \alpha -1}

210: \left( \nu_l \over \nu_1 \right)^{-\alpha+2}

211: {1-(\nu_u/\nu_l)^{-\alpha+1} \over 1-(\nu_2/\nu_1)^{-\alpha+2}}

212: \end{equation}

213: 

214: If we take a spectrum with $\alpha \approx 1$, (which is appropriate

215: for the frequency range $50-300$ keV as reported by Mallozzi et al. 1996) then we

216: obtain,

217: \begin{equation}

218: L_P(\theta)= 1.1\times 10^{57} \;T^{-1}\theta^{-2}\;{\rm ph}\;{\rm

219: sec}^{-1}\;.

220: \label{eq:Lp0}

221: \end{equation}

222: It should be noted that $T$ is not the total duration of the burst,

223: but an ``effective'' duration of convenience here, that is the

224: duration that the burst would have if its energy output were constant

225: at the peak value rather than highly variable.  In the simplest

226: version of our model (\S 2 \& 3), we will assume a single value of $T$

227: for all the bursts; however we will explore (\S 4) how our results

228: vary when a scatter in $T$ is introduced (which is a more realistic

229: assumption).

230: 

231: The jet model with the energy profile $\propto\theta^{-2}$ also makes

232: detailed predictions for the observed GRB flux distribution.  This,

233: within our formalism, can be written as \beq

234: \frac{dn(S)}{dS}=\int_0^\infty dL_p \;f(L_p) \left[\frac{R_{\rm

235: GRB}(z)}{(1+z)}\;

236: \frac{dV(z)}{dz}\left|\frac{dz}{dS}(z,L_p)\right|\right]_{z=z(S,L_p)}\;

237: \label{eq:dnds}

238: \eeq

239: where, given Eq.~(\ref{eq:Lp0}), the luminosity function takes the form 

240: $f(L_p)\propto L_p^{-2}$.

241: 

242: In Eqs.(\ref{eq:dndt}) and (\ref{eq:dnds}), $dV(z)/dz$ is the comoving volume. 

243: In a flat cosmology with a cosmological constant it is given by

244: \beq

245: \frac{dV(z)}{dz} = 4\pi {D^2(z)}\frac{dD(z)}{dz}\;

246: \eeq

247: where $D(z)$ is the comoving distance, 

248: \beq

249: D(z) = {c \over H_0} \int_0^z {dz' \over \sqrt{(1+\Omega_mz)(1+z')^2-\Omega_\Lambda(2z'+z'^2)} } \;,

250: \eeq

251: We assume a cosmological model with $\Omega_m=0.3$,

252: $\Omega_\Lambda=0.7$ and $H_0=71$ km s$^{-1}$ Mpc$^{-1}$. 

253: 

254: We assume that GRBs trace the star formation history and we adopt, as our ``standard'' model

255: for $z\la 10$, the Rowan-Robinson star formation rate

256: which can be fitted with the expression

257: \begin{equation}

258:  R_{\rm GRB}(z)  = \left\{

259:   \begin{array}{ll}

260:      R_0\; 10^{0.75z}, & \hbox{$ z < z_{\rm peak}$} \\

261:    R_0 \;10^{0.75 z_{\rm peak}},   & \hbox{ $z \ge z_{\rm peak}$}   \\

262:   \end{array}\right.\;,

263: \end{equation}

264: where $z_{\rm peak}\sim 2$. For $z\ga 10$, in our standard model we use an interpolation

265: that follows the star formation history derived in numerical simulations by Gnedin \& Ostriker (1997).

266: While using this SFR as our working model, we also explore the effects on the predicted 

267: distribution $dn/d\theta$ of different star formation histories, and in particular 

268: we consider two opposite extremes, one in which the SFR does not rapidly decline for  

269: $z\ga 10$ as implied in the numerical simulations of Gnedin \& Ostriker, and another, 

270: the Madau curve (Madau 1996), in which the SFR rapidly declines at redshifts $z\ga 3$. 

271: 

272: For each model for the SFR, the normalization constant $R_0$ is determined

273: by the condition

274: \beq

275: \int_0^{\pi/2}d\theta \;\frac{dn(\theta)}{d\theta}\;=\;R_{\rm GRB}^{\rm obs}\;,

276: \eeq

277: where 

278: $R_{\rm GRB}^{\rm obs}=667$ yr$^{-1}$ is the observed BATSE rate, and

279: $F_{\rm lim}$ in Eq. (\ref{eq:flim}) is the BATSE threshold flux for which

280: this rate has been measured.  We adopted the 90\% efficiency

281: peak flux threshold for BATSE, that is $F_{\rm lim}=0.424$ ph/sec (e.g.

282: Mallozzi, Pendleton \& Paciesas 1996).

283: 

284: \section{Comparison with data and predictions for more sensitive surveys}

285: 

286: In order to compare the theoretical distributions derived in \S{2.2},

287: we require a sample of gamma-ray bursts whose values of $\theta_{\rm

288: break}$ have been measured. The largest published sample of

289: $\theta_{\rm break}$ values at the time of our work comes from the

290: analysis of 28 bursts with redshifts and well-studied afterglow light

291: curves by Bloom et al. (2003). This list of 28 includes {\it all}

292: bursts with measured redshifts at this time. For all but four, some

293: limit on $\theta_{\rm break}$ was derived. The exceptions are

294: GRB~970228, in which sparse data together with likely contribution

295: from a supernovae makes it difficult to interpret the lightcurve,

296: GRB~990506 where no sufficient data exists, GRBs~980425 which had no

297: optical counterpart, and GRB~021211 which is being analyzed at the

298: time of writing.  For eight of the remaining 24, upper or lower limits

299: where put on the opening angle. We have not used these limits in our

300: comparison. One could think that this would tend to have the effect of

301: narrowing our sample distribution. However, the upper and lower limits

302: do not tend to be at the edge of the distribution of the measured

303: opening angles (see Table 2 of Bloom et al. 2003). They therefore do

304: not necessarily reflect extreme cases, but cases with lower quality

305: data. However, it should be noted that in three cases (GRBs~970828,

306: 991216, 990705) the confidence that indeed a jet has been identified

307: is weak, since the break was observed in a single frequency only.

308: Finally, we should remark that the inferred values of $\theta_{\rm

309: break}$ have some uncertainties. These values are indeed computed

310: using the expression given in Frail et al. (2001): $\theta_{\rm break}

311: \propto t_{\rm break}^{3/8} (1+z)^{-3/8}E_{\rm

312: iso}^{-1/8}\eta_\gamma^{1/8}n^{1/8}$. The measured values of $t_{\rm

313: break}$ have relative errors within 30\%, while for the densities a

314: value of $n=10$ cm$^{-3}$ is assumed for the 5 (out of 16) cases for

315: which the data quality did not allow a self-consistent determination

316: of the density through broad-band afterglow modelling. Finally, Bloom

317: et al. assumed a constant value ($\eta_\gamma=0.2$) for the efficiency

318: of the bursts. Whereas the dependence on $\eta_\gamma$ is weak, a large

319: spread in this not-well constrained quantity would introduce a further

320: source of error in the determination of $\theta_{\rm break}$.  The

321: combination of these caveats prevents a solid comparison of the data

322: with our predictions, and the following comparison should be taken as

323: a general guide for this type of analysis, while showing that the

324: observed distribution so far does not seem to be in contradiction with

325: the prediction of the structured jet hypothesis.

326: 

327: We performed a Kolmogorov-Smirnov (KS) test to assess the

328: compatibility of the theoretical distributions with the data, and we

329: found (see Figure 1), for the Rowan-Robinson SFR, a probability of

330: $\sim 90\%$ that the data are drawn from the theoretical

331: distribution. In order to have a reasonable agreement with the data, a

332: value $T\sim 8$ sec is needed in the theoretical model.

333: 

334: As shown in Figure 1, there is little difference in the results (and

335: in the required value of $T$ to fit the data) between the case where

336: the SFR drops rapidly after $z\ga 10$ (model 1), and where it keeps

337: constant also at higher redshifts (model 2). The similarity between

338: the distributions in model 1 and model 2 is a consequence of the

339: combined effect of the decrease in volume at those high redshifts and

340: the increased time dilution of the observed rate.  On the other hand,

341: for the same value of $T$ (or equivalently peak luminosity), the

342: distribution that uses the Madau SFR (model 3) predicts significantly

343: more events with large angles. This is because this star formation

344: rate drops abruptly at a redshift $z\ga 3-4$, and therefore there is

345: no much gain in the number of small-$\theta$ (i.e. brighter) events

346: which can be seen at higher redshifts. A KS test showed that the model

347: 3 distribution is consistent with the data (at the 40\% level) if a

348: value $T\sim 25$ is used. A lower value for the peak luminosity is

349: needed to shift the $dn/d\theta$ distribution to lower $\theta$'s.

350: 

351: For a jet model to be self-consistent it is necessary that, if the

352: distribution $dn/d\theta$ has a good agreement with the data for

353: certain model parameters, the corresponding $dn/dS$ has to have a good

354: agreement with the corresponding data for the same set of parameters.

355: We found that with $T\sim 8$ (as required in model 1) the peak fluxes

356: are within a factor of a few for those bursts with measured

357: $\theta$. More generally, when comparing the theoretical $dn/dS$

358: distribution with the all BATSE catalogue we find a very good

359: agreement for peak fluxes in the range $1\la P\la 15$ ph/cm$^2$/s, and

360: a departure (as an overprediction) at higher fluxes. However, it

361: should be noted that, given the large number of bursts in the BATSE

362: catalogue, a comparison with the $dn/dS$ data distribution is much

363: more sensitive to the model parameters than the $dn/d\theta$

364: comparison. In a recent analysis, Lloyd et al. (2001) found that a

365: good fit can be obtained with a luminosity function $\propto L^{-2.2}$

366: (which would reduce the number of high $P$ bursts with respect to our

367: model) and a redshift evolution. Such a detailed analysis is beyond

368: the scope of this paper given the lack of comparable wealth of data

369: for the $\theta$ distribution

370: \footnote{It should also be noted that the value of the peak

371: luminosity that best fits the $dn/d\theta$ distribution depends on the

372: adopted value of the detection efficiency, which may vary from

373: burst to burst.}.  On the other hand, we should remark that, if we

374: adopt the value $T\sim 25$ required to fit the $dn/d\theta$

375: distribution with the Madau SFR, then the luminosities that we predict

376: are generally smaller than those measured for

377: the bursts with known redshift. Therefore, in the following we will

378: only use our model 1 with $T= 8$ for further calculations.  The

379: corresponding cumulative distribution is shown with the solid line in

380: the two panels in Figure 2 where it is compared to the binned,

381: cumulative data. We should note that the size of the bins

382: in the figure has been chosen so that there is an equal increment

383: in the distribution for every new data point. This gives a better

384: visual idea of the data distribution, but does not reflect the

385: actual magnitude of the errors as described above.

386:      

387: Figure 3 shows the predicted evolution of the distribution of observed

388: break angles with increasing sensitivity of the survey. In particular,

389: we considered the 100\% efficiency sensitivities corresponding to

390: BATSE (solid line), HETE-2 and {\em Swift} (dashed line), and an

391: intermediate sensitivity (dotted line). An interesting prediction of

392: the luminosity distribution in Eq.(\ref{eq:Lp0}) is that the average

393: observed jet angle is an increasing function of the survey

394: sensitivity. There are two counteracting effects that determine the

395: average observed jet angle $<\theta>$ of the sample as a function of

396: the survey sensitivity. As $F_{\rm lim}$ decreases, $z_{\rm max}$

397: increases, bringing into the sample a fraction of bursts with larger

398: redshifts and correspondingly smaller $\theta$ (which are the most

399: luminous).  On the other hand, the higher sensitivity also brings into

400: the sample a fraction of bursts with larger $\theta$ at lower

401: redshifts, and this latter effect dominates over the former, partly as

402: a result of the volume-reduced and time-dilated rate of the high-$z$

403: bursts.

404: 

405: \section{Extensions}

406: 

407: All the above results have been produced under the assumption of a

408: strict correlation between the total energy of the burst and its peak

409: flux.  However, as discussed in \S 2, this relation has a scatter;

410: therefore we have also investigated the extent to which our results

411: change when a dispersion in the distribution of the values of $T$

412: (i.e. in the relation ~\ref{eq:Lp0}) is introduced.  If we

413: parameterize the scatter in $T$ with a probability distribution,

414: $P(T)$, then the distribution of jet angles (\ref{eq:dndt}) is

415: generalized to

416: \begin{equation}

417: \frac{dn(\theta)}{d\theta} = 2\pi\sin\theta 

418: \int_0^\infty dT\; P(T)

419: \int_0^{z_{\rm max}(\theta,T)}

420: dz\; \frac{R_{\rm GRB}(z)}{(1+z)}\; \frac{dV(z)}{dz}\;.

421: \label{eq:dndt1}

422: \end{equation}

423: We took the probability distribution for the scatter, $P(T)$, to be

424: a log-gaussian distribution with mean equal to $T=8$, and studied

425: the dependence of the break-angle distribution $dn/d\theta$ on the

426: width of the distribution $\sigma_{T}$.  

427: A K-S test shows that, with a scatter $\sigma_{T}=0.3$, the theoretical 

428: distribution is still compatible with the data at the 80\%, while 

429: with $\sigma_{T}=0.5$ the agreement is at the 

430: 40\% level.

431: 

432: All the results so far have been derived under the assumption of an

433: energy distribution from the jet axis $\propto\theta^{-2}$ in the

434: interval $0\le\theta\le \pi/2$. However, close to the axis this

435: divergence must naturally have a cutoff which we represent by a core

436: of size $\theta_c$. We now explore how our results vary by allowing

437: for the presence of a core in the inner part of the jet and an outer

438: cutoff at some large angle $\theta_j$, where the luminosity drops

439: rapidly to zero rather than following the profile in

440: Eq.~(\ref{eq:Lp0}).  In this case the peak photon luminosity is given

441: by $L_P(\theta)$ as in Eq. (\ref{eq:Lp0}) for

442: $\theta_c<\theta<\theta_j$, and by $L_P(\theta=\theta_c)$ for all the

443: angles $\theta\le\theta_c$. Hence the cumulative $\theta$ distribution

444: is given by

445: \begin{equation}

446:  N(<\theta)  = \left\{

447:   \begin{array}{ll}

448:      0 & \hbox{$ \theta < \theta_c$} \\

449:    \int_0^{\theta_c}d\theta'\;\frac{dn[\theta',L(\theta_c)]}{d\theta'}\equiv N_c   

450: & \hbox{ $\theta = \theta_c$}   \\

451: N_c \,+\, \int_{\theta_c}^{\theta}d\theta'\;\frac{dn[\theta',L(\theta')]}{d\theta'}\equiv N_j

452: & \hbox{ $\theta_c< \theta < \theta_j$}  \\

453:   N_j & \hbox{$ \theta > \theta_j$} \\

454:   \end{array}\right.\;.

455: \end{equation}

456: 

457: For the case where the increase in luminosity saturates at an angle

458: $\theta_c$ from the jet axis (left panel of Fig. 2), the number of

459: observed bursts with break angle $\theta<\theta_c$ is smaller than the

460: corresponding number at the same observed $\theta$ for the

461: distribution with $\theta_c=0$ (solid line in the figure). This is

462: because $z_{\rm max}$ saturates to the value given by the solution of

463: Eqn.~(\ref{eq:flim}) with $L=L_c$ for all $\theta\le\theta_c$, whereas

464: when there is no core\footnote{Note that when we say here ``no core''

465: or $\theta_c=0$, we mean an infinitesimally small core, as there would be

466: a formal divergence in the energy if $\theta_c$ were precisely equal to

467: zero.}, $z_{\rm max}$ is larger for the jet angles $\theta<\theta_c$.

468: The situation is reversed in the case of a jet with $\theta_c=0$ but a

469: total aperture $\theta_j<\pi/2$ (right panel of Fig.2).  The number of

470: bursts with observed break angle $\theta>\theta_j$ is zero.

471: Therefore, in order for the normalization (i.e. total number of

472: observed bursts) to be the same for any $\theta_j$, the cumulative

473: number $N(<\theta)$ for the case $\theta_j<\pi/2$ must be larger than

474: the corresponding number for the case $\theta_j=\pi/2$ (solid line in

475: the figure).

476: 

477: The probability remains of the same order by varying the core angle

478: $\theta_c$ in the range $0\la\theta_c\la 0.055$, and drops as

479: $\theta_c$ is increased above 0.055 (which is the smallest angle in

480: the data set).  Similarly, no significant variation in the probability is

481: found as the outer boundary of the jet angle, $\theta_j$ is decreased

482: from $\pi/2$ to 0.55 rad, which is the largest observed break angle.

483: In short, the current data is too poor to constrain either

484: the size of the core, $\theta_c$, or the outer size of the jet,

485: $\theta_j$, beyond the trivial statement that this range must include

486: the range of observed opening angles. It should however be remarked

487: that, whereas the current data on the observed $\theta_{\rm break}$ do not allow

488: to pin down the values of the model parameters $\theta_c$ and

489: $\theta_j$, the type of analysis that we are proposing has the

490: potentiality to further constrain details of the model once a larger

491: sample of jet opening angles is gathered.

492: 

493: \section{Conclusions}

494: 

495: A natural framework for the interpretation of achromatic breaks in the

496: afterglow light curves is the presence of jets in the GRB

497: ejecta. There is however a certain degree of degeneracy in the

498: resulting light curves between the case where the jet has uniform

499: energy profile and a sharp edge, and where it has instead a variable

500: energy profile. Distinguishing between the two scenarios is especially

501: important in order to have a proper estimate of the GRB event rates,

502: and a better understanding of the physics of the GRB explosion.  A

503: particularly useful discriminant of the two jet scenarios is the

504: distribution of the observed break times, which can be theoretically

505: predicted in the structured jet model, and compared to the data.  

506: 

507: In this paper, we have derived such distribution and compared it to

508: the observed sample of data on $\theta_{\rm break}$.  We found that

509: the observed data set, altough a small sample, is consistent with the

510: predictions of the structured-jet model.  We should however remark

511: that the alternative ``uniform'' model, in which there is a direct

512: correlation between the observed opening angle and the physical jet

513: angle, cannot be ruled out by this approach. However, if, as more data

514: becomes available, the agreement with the predictions of the

515: structured jet model remains intact, it would be very contrived to

516: justify it within the framework of the uniform jet. This model, in fact,

517: does not make any prediction for the distribution of

518: opening angles.  One would then need to invoke an explanation for why

519: there is the same number of bursts for each logarithmic

520: interval of openening angle.

521: 

522: Besides performing a first attempt to test the structured-jet model,

523: we have shown that this model offers a number of

524: predictions which may be testable with future, larger datasets. The

525: predicted opening angle distribution shown in Figure 1 (which assumes

526: the BATSE detection threshold), has a distinct peak at

527: $\theta\sim0.12$ rad. We predict that future, more sensitive missions will

528: predict more bursts with large opening angles. Specifically, we

529: estimate that the opening angle distribution for the SWIFT mission

530: will peak near $\theta\sim0.3$ rad. Surprisingly, in this model the

531: average redshift is only a weak function of the sensitivity and

532: consequently we expect no increase in the {\it average} redshift

533: detected by the next generation of gamma-ray instruments.

534: 

535: Another prediction that the structured-jet model makes, regards the

536: number of GRBs in the local universe.  Our ``standard'' SFR model

537: yields $R_{\rm GRB}(z=0)\sim 0.5$ Gpc$^{-3}$ yr$^{-1}$. Combining this

538: with the local galactic density $\approx 0.0048$ Mpc$^{-3}$ (Loveday

539: et al. 1992) one obtains $\approx 0.1$ GEM (galactic events per

540: Myr). This is an interesting number for detection of local GRB

541: remnants (Loeb \& Perna 1998; Efremov et al. 1998; Perna et

542: al. 2000). For the uniform jet model, precise rates are more difficult

543: to estimate, as they depend on the assumed luminosity function, or

544: equivalently the intrinsic jet angle distribution which is not known

545: apriori for this model. If one were to assume that $dn/d\theta

546: \propto\theta$ as in the structured jet model, this would result in a

547: logarithmic correction factor, $(1+\log[(\pi/2)/\theta_c])$, to the

548: rate that we estimated here for the structured jet model\footnote{The

549: correction is expected to be less than an order of magnitude as

550: $\theta_c\ga 1/\Gamma$.}. The total energy output of GRBs would

551: however remain the same, as in the structured jet model the total

552: energy of each burst is also corrected by a logarithmic factor with

553: respect to the total energy of each burst in the uniform case.

554: 

555: 

556: \acknowledgements We thank the referee for a very careful and

557: thoughtful review of our manuscript.  RP and DAF thank the California

558: Institute of Technology for its kind hospitality during the time that

559: part of this work was carried out.  RS holds a senior Sherman

560: Fairchild fellowship. This research was partially supported by a NASA

561: grant to RS.  The NRAO is a facility of the National Science

562: Foundation operated under cooperative agreement by Associated

563: Universities, Inc.

564: 

565: \begin{references}

566: 

567: \reference{} Bloom, J. S., Frail, D. A. \& Sari, R. 2001, AJ, 121, 2879 

568: \reference{} Bloom, J. S., Kulkarni, S. \& Frail, D. A., 2003, submitted to ApJ

569: \reference{} Covino, S. {\it et al.}  1999, A\&A, 348, L1.

570: \reference{} Efremov, Y. N., Elmegreen, B. G., \& Hodge, P. W. 1998, ApJ, 501, L163

571: \reference{} Frail, D. et al. 2001, ApJ, 562, 65 

572: \reference{} Gnedin, N. Y. \& Ostriker, J. P. 1997, ApJ, 486, 581

573: \reference{} Harrison, F.~A. {\it et al.}  1999, ApJ, 523, L121.

574: \reference{} Harrison, F.~A. {\it et al.}  2001, ApJ, 559, 123.

575: \reference{} Lloyd-Ronning, N. M., Fryer, C. L.  \& Ramirez-Ruiz, E. 2001, ApJ in press

576: (astro-ph/0108200) 

577: \reference{} Loeb, A., \& Perna, R. 1998, ApJ, 533, L35

578: \reference {} Loveday, J., Peterson, B. A., Efstathiou, G., Maddox, S. J.

579: 1992, ApJ, 390, 338

580: \reference{} Madau, P., Ferguson, H. C., Dickinson, M. E., Giavalisco, M.,

581:  Steidel, C. C. \&  Fruchter, Andrew, 1996, MNRAS, 283 1388

582: \reference{} Mallozzi, R. S., Pendleton, G. N., \& Paciesas, W. S.

583: 1996, ApJ, 471, 636

584: \reference{} Nicastro, L. et al. 1999, A\&AS, 138, 437

585: \reference{} Norris, J. P., Marani, G. F. \& Bonnell, J. T. 2000, ApJ,

586: 534, 284

587: \reference{} Panaitescu, A.~P. \& Kumar, P. 2002, 571, 779

588: \reference{} Perna, R., Raymond, J., \& Loeb, A. 2000, ApJ, 533, 658 

589: \reference{}Postnov, K. A., Prokhorov, M. E., Lipunov, V. M. 2001,

590: Astronomy Report, 45, 236

591: \reference{} Rhoads, J.~E. 1997, ApJ, 487, L1

592: \reference{} Rhoads, J.~E. 1999, A\&AS, 138, 539

593: \reference{} Rowan-Robinson, M. 1999, Ap\&SS, 266, 291

594: \reference{} Sari, R., Piran, T. \& Halpern, J. 1999, ApJ, 519, L17

595: \reference{} Salmonson, J. D., 2000, ApJL, 544, 115

596: \reference{} Rossi, E., Lazzati, D. \& Rees, M. J. 2002a, MNRAS,

597: 332, 945

598: \reference {} Rossi, E., Lazzati, D., Salmonson, D. J. \& Ghisellini, G. 2002b,

599: Proceeding of the workshop 

600: "Beaming and Jets in Gamma Ray Bursts", Copenhagen, August 12-30, 2002

601: \reference{} Stanek, K.~Z., Garnavich, P.~M., Kaluzny, J., Pych, W., \& Thompson,

602:   I. 1999, ApJ, 522, L39

603: \reference {} Vrba, F. et al. 2000, ApJ, 528, 254

604: \reference{} Wang, W., Woosley, S. E. \& MacFadyen, A. J., submitted to ApJ, (astro-ph/0207436)

605: \reference{} Wijers, R. A. M.~J. {\it et al.}  1999, ApJ, 523, L33.

606: \reference{} Zhang, B. \& Meszaros, P. 2002, ApJ, 571, 876 

607: \end{references}

608: 

609: \newpage

610: 

611: \begin{figure}[t]

612: \plotone{f1.eps}

613: \caption{Distribution of observed jet angles for different star

614: formation rates: in model 1 the Rowan-Robinson SFR is assumed up to

615: $z\ge 10$, and a rapid drop is assumed at larger redshifts as in the

616: numerical simulations of Gnedin \& Ostriker. In model 2, no dropout is

617: assumed for $z\ga 10$, while model 3 uses the Madau SRF. The histogram

618: shows the observed distribution from the data available so far for a

619: sample of 16 bursts.}

620: 

621: \end{figure} 

622: 

623: \begin{figure}[t]

624: \plottwo{f2a.eps}{f2b.eps}

625: \caption{Left panel: Cumulative distribution for the SFR model 1, and

626: various values of the jet core. Here $\theta_j=\pi/2$. Right panel:

627: Cumulative distribution for the SFR model 1, and various values of the

628: jet aperture. Here $\theta_c=0$.}

629: \end{figure} 

630: 

631: \begin{figure}[t]

632: \plotone{f3.eps}

633: \caption{Probability distribution for the observed jet angle 

634: $\theta$ for different values of the survey sensitivity threshold.

635: The higher the sensivity of the survey, the larger is the mean beaming

636: angle $\theta$ that is observed. Here $\theta_j=\pi/2$ and $\theta_c=0$

637: for the SFR model 1.}  

638: 

639: \end{figure} 

640: 

641: 

642: 

643: \end{document}

644: 

645: 

646: 

647: