Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://vo.astronet.ru/~karpov/pdf/2012_Naked_Eye_Burst.pdf Äàòà èçìåíåíèÿ: Fri Feb 17 16:47:18 2012 Äàòà èíäåêñèðîâàíèÿ: Mon Apr 11 02:23:53 2016 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ñ ð ð ñ ñ ð ï ð ï ï ï ï ð ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ð ï ï ð ï ð ï ï ð ï ï ð ï ï ð ï ï ð ï ð ï ï ð ï ð ï ï ð ï ï ð ï

Bull. Astr. Soc. India (2011) 00, 1­11

Evidences of the central engine activity in the Naked-Eye Burst prompt optical emission
S. Karpov1 , G. Beskin1 , S. Bondar2 , A. Guarnieri3 , C. Bartolini3 , G. Greco4 and A. Piccioni
1 2 3 4

3

Special Astropysical Observatory, Nizhnij Arkhyz, Russia, 369167 Institute for Precise Instrumentation, Nizhnij Arkhyz, Russia, 369167 Astronomy Department of Bologna University, Bologna, Italy Astronomical Observatory of Bologna, INAF, Italy

Received 17th February 2012

Abstract. Naked-Eye Burst is the only event to date observed in optical and gammaray ranges simultaneously with sufficiently high temporal resolution. This opens unprecedented possibilities for testing various models of both emission generation and the central engine behaviour during the burst. The temporal properties of the observed light curve, along with tight optical to gamma relation, suggest the intrinsic periodicity of internal engine, which is supposedly a newborn stellar-mass compact object surrounded by an unstable precessing massive accretion disk. Keywords : gamma-ray bursts: individual (GRB 080319B)

1. Introduction
Long-duration gamma-ray bursts are supposedly produced by newborn compact relativistic objects, stellar-mass black holes or highly magnetized neutron stars, during the collapse of the core of massive stars (Piran 2005; Meszaros 2006). Multiwavelength observations of the GRBs are important tools to study these objects, whose behaviour defines the temporal structure of the bursts. Thereby, the discovery of prompt optical emission in follow-up observations (Akerlof et al. 1999; Vestrand et al. 2005; Yost et al. 2007; Page et al. 2007) has led to the refinement of different GRB models. However the insufficient temporal resolution of the observations, typically worse than 10 s, has prevented clarifying the nature of the bursts. In order to capture the initial moments of the burst optical emission with high temporal resolution, we have designed and built several wide-field monitoring cameras (Piccioni et al. 1993;


email: karpov@sao.ru


2

S. Karpov et al

Karpov et al. 2005; Molinari et al. 2006). One of these cameras, the TORTORA (Telescopio Ottimizzato per la Ricerca dei Transienti Ottici RApidi), has successfully carried out high temporal resolution observations of the Naked-Eye Burst, GRB080319B (Karpov et al. 2008; Racusin et al. 2008; Beskin et al. 2008). This was ­ and still is ­ the first case study of a GRB's optical emission from the beginning till the end, with a resolution comparable to the one of gamma-ray observations. We discovered the fast variability of the optical emission and its lag behind the gamma-ray emission, which reflects the periodic activity of the burst internal engine ­ likely, a hyperaccreting solar-mass compact object formed in the collapse of a massive stellar core. The aim of this paper is to describe the TORTORA observations and data reduction of GRB 080319B (Section 2), then to perform a detailed comparative analysis of the optical and gamma emission (Section 3), and finally to discuss the nature of the discovered optical emission and its variability (Section 4).

2. Observations
We observed the region of GRB080319B from 30 minutes before the trigger time T , until several tens of minutes after (Karpov et al. 2008; Racusin et al. 2008), with TORTORA wide field (24x32 degrees) monitoring camera (temporal resolution 0.13 s) (Molinari et al. 2006) mounted on the REM telescope at La Silla (Chile). After receiving the coordinates of GRB 080319B communicated by the Swift satellite, the position of the burst was moved from the edge of the TORTORA field of view towards its centre. Therefore from T + 24.5 s till T + 31 s the REM telescope performed an automatic repointing (see Figure 1). The acquired data were processed by the standard TORTORA pipeline including CCD dark current subtraction and flat-fielding. The reduction were performed by a customary code accomplishing circular aperture photometry and then verified by IRAF DAOPHOT code, except the data acquired during the REM repointing. Over that period of time, the target and nearby stars images are stretched up to 5 times on the time scale of a single exposure due to the motion of the field of view. Such motion significantly lowers the signal-to-noise ratio and makes it impossible to perform reliable flux measurements on every single frame. To overcome this challenge, we performed the summation of sets of 10 non-overlapping frames, spatially shifting them to compensate the motion of the telescope and to obtain profiles of the trails with a signal-to-noise ratio nearly equal to one in other intervals of the light curve (see Figure 2c). Then we processed the co-added frames through a customary elliptic aperture photometry code as well as by means of PSF-fitting, which provided consistent results. The effective temporal resolution for this interval is 1.3 s. For all other time intervals the photometry has been performed with both full (0.13 s exposure) and low temporal resolution (1.3 effective exposure for 10 co-added images). Then, we obtained the instrumental magnitudes of the object, both for the REM repointing interval and the other parts of the light curve. These were calibrated to the Johnson V band using several nearby Tycho-2 stars. The light curve of one of the comparison stars, processed and calibrated in the same way


Naked-Eye Burst prompt optical emission
T = 0 sec T = 20.59 sec T = 26.42 sec

3

T = 28.37 sec

T = 36.15 sec

T = 80.23 sec

Comparison Star

Object

Figure 1. The development of prompt optical emission from GRB080319b as seen by TORTORA camera. Sums of 10 consecutive frames with 1.3 s effective exposures are shown for the gamma-ray trigger time (T = 0 s), the maximum brightness time during the first peak (T = 20.5 s), two middle-part moments (T = 26.4 s and T = 28.4 s), at the last peak (T = 36 s) and during early afterglow (T = 80 s) stages. Image size is 2.5 x 2.5 degrees. Circles mark the transient (in the center) and the brightest of the nearest field stars used for photometric calibration, whose light curve is shown in Figure 2b.

as the transient, is shown in Figure 2b. No significant deviation from the constant flux, neither in high resolution, nor in low resolution mode is present, which argues in favour of the real nature of variations seen in the light curve of the transient. A quick-look low resolution light curve (excluding the data during the REM repointing interval) has been published (Karpov et al. 2008; Racusin et al. 2008). Our complete full resolution light curve, along with the low resolution one (after the restoring of the gap), is shown in Figure 2.


4

S. Karpov et al

(a) Swift BAT

(b) Comparison star

25 0

15

(c) GRB 080319B

Optical flux density, Jy

20

5.5 t=18.3 t=27.1 t=36.1 t=44.4
(d) Residuals

5

-5

0

0

20

40 Time since trigger, sec

60

80

Figure 2. The light curve of GRB080319B acquired by TORTORA wide-field camera (c), approximated by the sum of four peaks with parameters listed in Table 1. Residuals of such approximation are shown in (d). The gamma-emission, presented for comparison in (a), started at T -4 s and faded at T 57 s. The light curve of comparison star (b) shows no significant features.

Table 1. Best-fit parameters for the decomposition of the light curve into 4 peaks with shape described by Kocevski (Kocevski et al. 2003) profile (see Equation 1), shown in Figure 2c. Here, T 0 and F0 are the peak maximum position relative to trigger time and flux, while r and d are the power-law indices of their rising and declining parts. T is the distance between the peak and the next one.

T0, s 18.3 ± 0.3 27.0 ± 0.3 36.1 ± 0.2 44.4 ± 0.5

F0 23.2 13.4 11.4 15.1

,J ± ± ± ±

y 0.6 3.4 1.7 1.8

4.0 0.4 24.8 8.3 25.9 7.6 21.9 3.3

r ± ± ± ±

d -5.4 ± 4.1 -9.7 ± 4.9 -22.0 ± 17 -5.1 ± 0.2

T 8.7± 9.1± 8.3±

,s 0.4 0.4 0.5

9

V magnitude

15

10

7

6


Naked-Eye Burst prompt optical emission
1e+03 (a) GRB 080319B

5

6

(c) Last peak, residuals Flux Residuals, Jy 4

Power density, RMS2 Hz

1e+02

1e+01

1e+00

1e-01

0.02 1e+03

0.05

0.10

0.20 Frequency, Hz

0.50

1.00

2.00

-4

-2

0

2

40

42

44 Time since trigger, s

46

48

50

(b) Comparison star

(d) Last peak, power density Power density, RMS2 Hz 10 0 5

Power density, RMS2 Hz

1e-01

1e+00

1e+01

1e+02

0.02

0.05

0.10

0.20 Frequency, Hz

0.50

1.00

2.00

0.02

0.05

0.10

0.20 Frequency, Hz

0.50

1.00

2.00

Figure optical nearby shown

3. (a) Power density spectrum of plateau phase (since T + 14 s till T + 49 s) of the Naked-Eye Burst light curve with linear trend removed. (b) Power density spectrum of the same interval of bright star light curve. (c) Optical flux for a T + 40 s ­ T + 50 s interval (last peak) with the approximation in Figure 2 subtracted. (d) Power density spectrum of these data.

3. Analysis of optical light curve
3.1 Periodicity of optical emission The light curve of GRB080319B clearly shows four peaks with similar amplitudes, durations and shapes (see Figure 2c). We approximated the low-resolution data by the sum of four components described by a simple functions smoothly connecting two power laws (Kocevski et al. 2003) F=F t T0
r

0

d r t d + r + d + r T 0

r+1 -(r+d)/(r+1)

(1)

(where t is the time since trigger) whose parameters are listed in Table 1. The fit over the interval from T = 0 s till T = 100 s has a 2 = 33.4 for 61 degrees of freedom. The distances between peaks are the same within the errors: T 8.5 s in the observer frame, which corresponds to 4.4 s in the rest frame at z=0.937 (Racusin et al. 2008; Beskin et al. 2008). We then performed a Fourier analysis of the central part of the light curve (full-resolution one, with data points finely sampled with 0.13 s separation), defined as the interval between the first and the last peak extended by the mean distance between peaks, that is from (T 1 - 0.5T ) = 14.05 s until (T 4 + 0.5T ) = 48.65 s. The power density spectrum of this light curve interval after removal of the linear trend is shown in Figure 3a. A feature at 0 0.12 Hz (marked by vertical line) is clearly visible and corresponds to the mean distance between the light curve peaks, alongside the low-frequency components of the overall two-level step-like structure of this part of the light curve (not completely compensated by linear trend removal).


6

S. Karpov et al

Its significance, however, strongly depends on the assumed model of the shape of the underlying continuum component ­ the zero hypothesis. Reconstructing it from surrounding points, as an arithmetic mean of two nearest bins, gives a 24.5 times excess of the actual spectral density over the continuum (dotted horizontal line in Figure 3a). Assuming the standard exponential probability distribution of power density estimates over the mean value, it corresponds to p = 3 · 10-9 probability of zero hypothesis after correcting for the amount of 133 bins in the whole spectrum. A more conservative estimate can be made modeling the un-detrended light curve with the sum of red and white noise components, approximating its spectral density excluding the bin containing the periodic component in question. We generated a large set of such processes and detrended each of them in the same way as the original data. The mean values of spectral density of such artificial data are shown as a smooth solid curve in Figure 3a. The periodic feature then shows the 13.4 times excess over the mean zero-hypothesis level, which corresponds to a p = 2 · 10-4 significance level taking into account the full number of bins in the complete spectra (133 bins). The power spectrum of the comparison star, derived in the same way and for the same light curve region, does not display any features with significance better than 0.01 in a single bin (0.35 for total number of 133 bins) in respect to mean noise level (see Figure 3b), neither at the frequency of the peak seen in the transient spectrum, nor anywhere else. This rules out a possible artificial nature of the periodicity earlier discovered. Thus, for the first time, we have a clear detection of periodic variations of a GRB prompt optical emission on a few seconds time-scale. To study short time-scale optical variability, the smooth curve - formed by four fitted peaks - has been subtracted from the original full-resoluion high-resolution data. The residuals are shown in Figure 2d. The power spectral analysis of different sub-intervals of the burst revealed no signature of a variability in the 0.1-3.5 Hz (0.3-10 s) range with power exceeding 15% before and 10% after the REM repointing, except for the signature, marked by a vertical line in Figure 3d, of a periodic intensity variation during the last peak, from T + 40 s till T + 50 s, with a 13.7 times excess over mean noise level (horizontal line in Figure 3d), which implies a p = 2.4 · 10-5 significance for 39 bins. The period and amplitude of a best-fit sinusoidal approximation, shown in Figure 3c, are 1.13 s and 9%, respectively. No other interval, neither of the object nor of the comparison stars light curves, shows any variability in the 0.1-3.5 Hz (0.3-10 s) range with power exceeding 15% before and 10% after the REM repointing (see Figure 2d). The spectrum of gamma-ray light curve lacks the signature of such periodic variations on top of its stochastic variability, like the longer time scale one.

3.2 Correlation between optical and gamma-ray emission In order to reveal the similarity between optical and gamma-ray light curves, and to filter out the stochastic variability of the latter, we performed a cross-correlation analysis. Again, we


Naked-Eye Burst prompt optical emission
0.9 Low-resolution (1.3 s) Full-resolution (0.13 s)

7

CCF

0.3

0.5

0.7

0

2 Delay ( V - ) , s

4

6

25

Flux density, Jy

TORTORA, shifted 2 s back Swift BAT, all channels

5

10

15

20

15

20

25

30 Time, s

35

40

45

Figure 4. Upper panel ­ cross-correlation of the Swift-BAT gamma-ray (data from all energy channels with 64 ms temporal resolution) and TORTORA (data with 0.13 s and 1.3 s resolution) optical fluxes for the main (plateau) phase of the burst emission.Lower panel ­ TORTORA optical flux shifted back 2 seconds along with correspondingly rebinned Swift-BAT gamma-ray flux. Gamma-ray curve is arbitrarily scaled and shifted for illustrative purposes.

used solely the plateau phase of the light curves, excluding the first and last 12 seconds of the burst, both in optical and in gamma, as they are obviously highly correlated (Beskin et al. 2008). For the low-resolution data, with a 1.3 s bin size, the correlation coefficient is as high as 0.82 with a significance level of 5 · 10-7 when the optical light curve is shifted 2 seconds back with respect to gamma-ray one (see Figure 4). The correlation of unshifted data, at the same time, is as small as 0.42 with a significance of 0.03. Correspondingly rebinned gamma-ray data show the same four quasi-equidistant peaks as the optical ones (lower panel of Figure 4). Correlation of high resolution data (0.13 s bin) is systematically lower due to higher amount of noise in it. Quasi-periodic variations of cross-correlation for low resolution data are due to a sharp feature in gamma-ray data at around T + 30 s falling into either one or two bins of rebinned light curve. Nevertheless, the periodic variability components seen in the optical light curve are not manifested in the power density spectrum of the corresponding intervals of the gamma-ray one. This is most likely due to a significant amount of stochastic, shot noise-like variability of gamma emission which spans from tens to fractions of seconds (Margutti et al. 2008), possibly masking any


8

S. Karpov et al

low-amplitude regular structure in the data. Indeed, the non-stationary ejection of matter from the burst internal engine most likely modulates not the amplitudes of flares in the light curve, but their rate (see Section 4 and references therein) ­ and such modulation results in a complex transformation of underlying process power spectrum, presumably leaking the power from periodic peaks. To study such a case we performed extensive set of simulations of a "gamma-ray light curves" consisting of a short flares with varying rate (or mean distance between them) which tracks the smooth optical light curve intensity, and found that for nearly 90% of realizations the power spectra of simulated light curves do not show any features on the frequency in question, while keeping the correlation with optical light curve roughly as high as r = 0.82 observed. Obviously, there are other possible types of stochastic process modulation which may be revealed only in smoothed data, but not in the power spectrum.

4. Discussion
The t 2 s delay of the optical flash relative to the gamma-ray one inevitably suggests that they were generated in different parts of the ejecta. More precisely, the optical photons came from a distance R 2c2 t(1 + z)-1 = 1.5 · 1016 2 cm away from the central engine, where 500 is 500 the Lorentz factor in units of 500 (Piran 2005; Li and Waxman 2008). This result and other peculiarities we detected in the Naked-Eye Burst clearly contradict the proposed emission generation mechanisms based on various kinds of interactions between a single ensemble of electrons and photons they generate ­ synchrotron or inverse Compton ones (Racusin et al. 2008; Kumar and Panaitescu 2008; Fan and Piran 2008), the model of two internal shocks, forward and reverse (Yu, Wang and Dai 2009), and a relativistic turbulence model (Kumar and Narayan 2009). On the other hand, the fast rise and the similarity of durations of all four optical flashes rule out an external shock (both forward and reverse) as a source of optical emission (Zou, Piran and Sari 2009). Two internal shock models have also been proposed, in which optical and gamma-ray flashes are generated by a synchrotron mechanism in different parts of the ejecta ­ the larger the photon energy, the closer it to the central engine. These models are the residual collisions model (Li and Waxman 2008) and the model with significant input of neutron component (Fan, Zhang and Wei 2009). In these scenarios, the gamma emission is produced at a distance of 1014 -1015 cm from the center due to the electron heating caused by shock waves of colliding proton shells. In the former model, optical quanta are generated in an optically thin plasma during the collisions of "residual" shells (each shell being the result of the merging between a large number of thinner "original" shells), far ( 1016 cm) from the central engine (Li and Waxman 2008). In the latter model, the optical emission is generated by the electrons produced in -decay of neutrons, which may reach a distance of R 1016 cm without interactions with other components of the ejecta. The decay products, protons and electrons, collide with faster proton shells ejected later, producing secondary internal shocks which heat the electrons generating synchrotron optical emission. Both models easily explain the two-second delay observed in the optical light curve, as well as its general smoothness on 0.1-1 s time scale, in contrast to high level of stochastic variability in the


Naked-Eye Burst prompt optical emission

9

gamma-ray emission (Margutti et al. 2008). On the other hand, the great difference of NakedEye Burst optical and gamma-ray fluxes (Fo /F 103 ) (Racusin et al. 2008) is more naturally explained in a neutron-rich model (Fan et al. 2009). As a matter of fact, a large amount of neutrons is unavoidable in bright gamma-ray bursts like GRB080319B (Derishev, Kocharovsky and Kocharovsky 1999; Pruet, Woosley and Hoffman 2003). This model, therefore, is preferable, and our results may be a strong evidence of the existence of a significant neutron component in the ejecta. In brief, we conclude that optical and gamma emission of Naked-Eye Burst were generated at different distances from the central engine. Such conclusion is a direct consequence of the detected similarity of shifted optical and gamma-ray light curves. Such similarity depends neither on particular mechanisms of conversion of mechanical to internal electron energy, nor on the emission mechanisms. Moreover, this effect can not be caused by the density or velocity variations inside the ejecta, such as those already observed on a time scale of several tens of minutes in afterglows of other gamma-ray bursts (Jakobsson et al. 2004; Bersier et al. 2003). Obviously, it would be impossible for the relativistic ejecta itself to display similar structures and dynamics, especially periodic behaviours, in regions separated by 1016 cm. Therefore, we have to conclude that these variations have the same cause ­ namely, the cyclic variations of internal engine activity (each flash of the light curve corresponds to one of its four episodes). The detected non-stationarity of the ejection flow can be a result of non-stationary accretion due to periodically triggered gravitational instability (Masada et al. 2007) in the hot inner part of one solar mass hyperaccreting disk, around a compact object with a mass of about three solar masses, formed in the collapse of a massive star (Woosley 1993; Zhang, Woosley and Heger 2004). Such a disk must contain a large amount of neutrons (Pruet et al. 2003), and the four peaks detected in the optical light curve reflect the four episodes of accretion activity leading to the ejection of matter. The matter forming the inner part of the disk (fragmented due to instabilities) becomes the elements of the jet. The collisions of these fragments generate the internal shocks in the ejecta. Moreover, the half-second variations in the source frame, marginally seen during the last stage of the optical transient (see Figure 3c), may be interpreted as a result of the LenseThirring precession or nutation (J.Hartle, K.Thorne and R.H.Price 1986; Reynoso, Romero and Sampayo 2006) of the inner parts of this disk.

5. Conclusions
The strategy of a wide-field monitoring with high temporal resolution (Karpov et al. 2010; Beskin et al. 2010), implemented in the TORTORA camera, made it possible to perform, for the first time, a detailed investigation of the optical flash accompanying GRB080319B. We discovered the periodic variability of its optical emission with a characteristic time scale of 8.5 s (4.2 s in source frame), manifested in the four similar light curve peaks. The amplitude


10

S. Karpov et al

of stochastic variations does not exceed 10-15% on the variable gamma-ray emission. The comparison of light optical and the gamma-ray emission, clear despite of a structures. Correspondingly shifted light curves have a the T + 10 s ­ T + 47 s interval.

0.1-1 s time scale, in contrast to highly curves reveals a 2 s delay between the similarity between the overall temporal correlation coefficient of r = 0.82 over

The acquired data and their analysis led us to the following conclusions:

· the emission of GRB080319B in the optical and gamma-ray ranges is generated in different regions of the ejecta, which clearly contradicts a number of models suggested as an explanation of this burst (Racusin et al. 2008; Kumar and Panaitescu 2008; Fan and Piran 2008; Yu et al. 2009; Kumar and Narayan 2009); · peculiarities of the optical emission variations in comparison to the gamma-ray ones can naturally be explained in the framework of models with either residual collisions of shells (Li and Waxman 2008) or significant neutron component in the outflow (Fan et al. 2009). As the latter model provides a better explanation of various aspects of the observational features of the burst, we suggest the presence of vast amount of neutrons in the jet; · the four peaks characterizing the optical light curve reflect the periodic modulation of the accretion from the massive disk around the newborn compact object, formed in the corecollapse of a massive star.

Our important data on the detailed structure of GRB080319B optical emission, never acquired before, suggest that the development of methods and instruments for wide-field monitoring with high temporal resolution should continue. Especially important is the construction of instruments able not only to detect an optical transient, but also able to simultaneously acquire the spectral and polarimetric information from optical photons in real time (Beskin et al. 2010).

Acknowledgements
This work was supported by the Bologna University Progetti Pluriennali 2003, by grants of CRDF (No. RP1-2394-MO-02), RFBR (No. 04-02-17555, 06-02-08313 and 09-02-12053), INTAS (0478-7366) and by the Presidium of the Russian Academy of Sciences Program and by the grant of President of Russian Federation for the support of young Russian scientists. S.K. has also been supported by a grant of Dynasty foundation. G.B. thanks Landau Network-Cenro Volta and Cariplo Foundation for fellowship and Brera Observatory for hospitality. G.G. also gratefully acknowledges the support of Foundatione CARISBO. We thank Emilio Molinari, Stefano Covino and Cristiano Guidorzi for technical help organizing TORTORA observations and for discussions of the results.


Naked-Eye Burst prompt optical emission

11

References
Akerlof, C., et al., 1999, Nature, 398, 400 Bersier, D., et al., 2003, ApJL, 584, L43 Beskin, G., et al., 2008, in Y.-F. Huang, Z.-G. Dai, B. Zhang, eds., 2008 NANJING GAMMARAY BURST CONFERENCE, American Institute of Physics Conference Series, vol. 1065, pp. 251­254 Beskin, G., et al., 2010, Advances in Astronomy, 2010 Derishev, E. V., Kocharovsky, V. V., Kocharovsky, V. V., 1999, ApJ, 521, 640 Fan, Y.-Z., Piran, T., 2008, Frontiers of Physics in China, 3, 306 Fan, Y.-Z., Zhang, B., Wei, D.-M., 2009, Phys.Rev.D, 79, 021301 Jakobsson, P., et al., 2004, New Astronomy, 9, 435 J.Hartle, K.Thorne, R.H.Price, 1986, Black Holes: the Membrane Paradigm, ed. by K.Thorne and R.Price and D.MacDonald, New Haven: Yale University Press, chap. 5, p. 173 Karpov, S., et al., 2005, Nuovo Cimento C Geophysics Space Physics C, 28, 747 Karpov, S., et al., 2008, GRB Coordinates Network Circular, 7452, 1 Karpov, S., et al., 2010, Advances in Astronomy, 2010 Kocevski, R., Ryde, F., Liang, E., 2003, ApJ, 596, 389 Kumar, P., Narayan, R., 2009, MNRAS, 395, 472 Kumar, P., Panaitescu, A., 2008, MNRAS, 391, L19 Li, Z., Waxman, E., 2008, ApJ, 674, 65 Margutti, R., et al., 2008, in Y.-F. Huang, Z.-G. Dai, B. Zhang, eds., 2008 NANJING GAMMARAY BURST CONFERENCE, American Institute of Physics Conference Series, vol. 1065, pp. 259­262 Masada, Y., et al., 2007, ApJ, 663, 437 Meszaros, P., 2006, Reports on Progress in Physics, 69, 2259 Molinari, E., et al., 2006, Nuovo Cimento B, 121, 1525 Page, K. L., et al., 2007, ApJ, 663, 1125 Piccioni, A., et al., 1993, in M. Friedlander, N. Gehrels, D. J. Macomb, eds., American Institute of Physics Conference Series, American Institute of Physics Conference Series, vol. 280, pp. 1152­1155 Piran, T., 2005, Reviews of Modern Physics, 76, 1143 Pruet, J., Woosley, S. E., Hoffman, R. D., 2003, ApJ, 586, 1254 Racusin, J. L., et al., 2008, Nature, pp. 183­188 Reynoso, M. M., Romero, G. E., Sampayo, O. A., 2006, A&A, 454, 11 Vestrand, W. T., et al., 2005, Nature, 435, 178 Woosley, S., 1993, ApJ, 405, 273 Yost, S. A., et al., 2007, ApJ, 657, 925 Yu, Y. W., Wang, X. Y., Dai, Z. G., 2009, ApJ, 692, 1662 Zhang, W., Woosley, S. E., Heger, A., 2004, ApJ, 608, 365 Zou, Y.-C., Piran, T., Sari, R., 2009, ApJL, 692, L92