0710.2858/ms.tex
1: % file ms.tex
2:  
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: % edit history
5: %
6: % 2007-06-16 owigger
7: %    copied from llc to hedera, edited sections 5 and 6
8: %    added 5.4 GRBs and XRFs, and 5.5 one more observation
9: % 2007-06-17 cwigger
10: %    copied from owigger's to cwigger's account
11: %    edited minor points all over the text
12: % 2007-06-17 owigger
13: %    copied from cwigger to owigger
14: %    embellished figure and table captions and titles
15: % 2007-06-18 cwigger
16: %    major moving around things in section 'results' and 'discussion'
17: %    minor changes for the rest
18: % 2007-06-22 cwigger
19: %    added Table {tab:fl}
20: % 2007-07-03 cwigger
21: %    worked (sometimes renamed to ms_new.tex)
22: % 2007-07-03 owigger
23: %    copied, renamed back to ms.tex, overhauled new text
24: % 2007-07-05 owigger
25: %    changed title, proofread
26: % 2007-07-05 cwigger
27: %    finished tables,  new table with peak resolved analysis,
28: %    new order of GRBs: time ordered. Change of references accordingly.
29: % 2007-07-06 owigger
30: %    final proofreading
31: % 2007-07-07 owigger
32: %    added corrections by Eric and Claudia, rearranged figures,
33: %    checked citations, added ebellm's F-test, applied ApJ guidelines
34: %    aligned equations
35: % 2007-07-08 owigger
36: %    Claudia's corrections added
37: % 2007-07-09 owigger
38: %    Wojtek's corrections added, new eps figures spec* and 030519_ltc
39: % 2007-07-10 owigger
40: %    removed all amstex at the request of 
41: %    William Wentworth-Sheilds wws@press.uchicago.edu
42: % 2007-10-11 cwigger
43: %    enlarged sections fit, CBM function and spectral hardening
44: %    according to referee's request
45: % 2007-10-11 owigger
46: %    little improvements
47: % 2007-10-14 cwigger
48: %    Eric's comments
49: % 2007-10-15 cwigger
50: %    final small changes
51: %
52: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
53: 
54: %% This is a manuscript marked up using the
55: %% AASTeX v5.x LaTeX 2e macros.
56: 
57: %% The first piece of markup in an AASTeX v5.x document
58: %% is the \documentclass command. LaTeX will ignore
59: %% any data that comes before this command.
60: 
61: %% The command below calls the preprint style
62: %% which will produce a one-column, single-spaced document.
63: %% Examples of commands for other substyles follow. Use
64: %% whichever is most appropriate for your purposes.
65: 
66: \documentclass[12pt,preprint]{aastex}  % use this for submission
67: 
68: %% manuscript produces a one-column, double-spaced document:
69: 
70: % \documentclass[manuscript]{aastex}
71: 
72: %% preprint2 produces a double-column, single-spaced document:
73: 
74: %\documentclass[preprint2]{aastex}
75: 
76: %% If you want to create your own macros, you can do so
77: %% using \newcommand. Your macros should appear before
78: %% the \begin{document} command.
79: %%
80: %% If you are submitting to a journal that translates manuscripts
81: %% into SGML, you need to follow certain guidelines when preparing
82: %% your macros. See the AASTeX v5.x Author Guide
83: %% for information.
84: 
85: %%%%%%%%%%%%%%%%%%%%%%%% MACROS %%%%%%%%%%%%%%%%%%%%%%%%%%
86: % math macros
87: \newcommand{\mathEbreak}{E_{\mbox{\em\scriptsize break}}}
88: \newcommand{\mathEpeak}{E_{\mbox{\em\scriptsize peak}}}
89: \newcommand{\mathEFE}{E\!\cdot\!F_E}
90: % text macros
91: \newcommand{\CB}{CBM}
92: \newcommand{\CLerrors}{90\% CL errors}
93: \newcommand{\fluenceCB}{$F_{\mbox{\scriptsize CB}}$}
94: \newcommand{\fluenceBand}{$F_{\mbox{\scriptsize Band}}$}
95: \newcommand{\fluenceCPL}{$F_{\mbox{\scriptsize CPL}}$}
96: \newcommand{\fluencemPL}{$F_{\mbox{\scriptsize mPL}}$}
97: \newcommand{\fluenceRHESSI}{$F_{\mbox{\scriptsize RHESSI}}$}
98: \newcommand{\fluenceHETE}{$F_{\mbox{\scriptsize HETE}}$}
99: \newcommand{\fluenceUlysses}{$F_{\mbox{\scriptsize Ulysses}}$}
100: \newcommand{\Ebreak}{$\mathEbreak$}
101: \newcommand{\Epeak}{$E_{\mbox{\em\scriptsize peak}}$}
102: \newcommand{\EFE}{$E\!\!\cdot\!\!F_E$}
103: \newcommand{\Ein}{E^{\mbox{\em\scriptsize in}}}
104: \newcommand{\Edet}{E^{\mbox{\em\scriptsize det}}}
105: \newcommand{\refeq}{eq.~\ref}
106: \newcommand{\fa}{\tablenotemark{a}}
107: \newcommand{\fb}{\tablenotemark{b}}
108: \newcommand{\PM}{$\pm$}
109: % macros usable in both math and text
110: \newcommand{\keV}{\mbox{keV}}
111: \newcommand{\MeV}{\mbox{MeV}}
112: \newcommand{\phS}{\phm{-}}
113: %%%%%%%%%%%%%%%%%%%%%%% END MACROS %%%%%%%%%%%%%%%%%%%%%%%
114: 
115: %% You can insert a short comment on the title page using the command below.
116: \slugcomment{submitted to \apj, 2007-07-06}
117: %\slugcomment{accepted by ???, date?}
118: 
119: %% If you wish, you may supply running head information, although
120: %% this information may be modified by the editorial offices.
121: %% The left head contains a list of authors,
122: %% usually a maximum of three (otherwise use et al.).  The right
123: %% head is a modified title of up to roughly 44 characters.  Running heads
124: %% will not print in the manuscript style.
125: 
126: %\shorttitle{Spectra of Strong RHESSI GRBs}
127: %\shortauthors{Wigger et al.}
128: 
129: %% This is the end of the preamble.  Indicate the beginning of the
130: %% paper itself with \begin{document}.
131: 
132: \begin{document}
133: 
134: %% LaTeX will automatically break titles if they run longer than
135: %% one line. However, you may use \\ to force a line break if
136: %% you desire.
137: 
138: %\title{GRB 021206: spectral analysis of RHESSI GR}
139: %\title{GRB spectra in the few MeV region:
140: %A comparative study of strong, hard GRBs observed by RHESSI}
141: %\title{Spectral analysis of strong GRBs observed by RHESSI:
142: %Band function versus Cannonball model}
143: %\title{GRB 021206 observation by RHESSI: 
144: %   spectral hardening around 5 MeV}
145: %RHESSI observation of GRB 021206: Is the spectral hardening
146: %round 5 MeV a sign of a special case?
147: \title{
148: Observation of an unexpected hardening in the spectrum of GRB~021206
149: }
150: 
151: %% Use \author, \affil, and the \and command to format
152: %% author and affiliation information.
153: %% Note that \email has replaced the old \authoremail command
154: %% from AASTeX v4.0. You can use \email to mark an email address
155: %% anywhere in the paper, not just in the front matter.
156: %% As in the title, you can use \\ to force line breaks.
157: 
158: \author{C. Wigger\altaffilmark{1,2}, 
159:         O. Wigger\altaffilmark{2}, 
160: 	E. Bellm\altaffilmark{3},
161:         W. Hajdas\altaffilmark{1}
162: }
163: \email{claudia.wigger@psi.ch}
164: %\email{owigger@hispeed.ch}
165: %\email{ebellm@ssl.berkeley.edu}
166: %\email{wojtek.hajdas@psi.ch}
167: 
168: \altaffiltext{1}{
169:     Paul Scherrer Institut,
170:     CH-5232 Villigen PSI,
171:     Switzerland
172: }
173: \altaffiltext{2}{
174:     Tellistrasse 9, 
175:     CH-5000 Aarau,
176:     Switzerland
177: }
178: \altaffiltext{3}{
179:     UC Berkeley Space Sciences Laboratory,
180:     7 Gauss Way, Berkeley, 
181:     CA 94720-7450, USA
182: }
183: 
184: %% Notice that each of these authors has alternate affiliations, which
185: %% are identified by the \altaffilmark after each name.  Specify alternate
186: %% affiliation information with \altaffiltext, with one command per each
187: %% affiliation.
188: 
189: 
190: %% Mark off your abstract in the ``abstract'' environment. In the manuscript
191: %% style, abstract will output a Received/Accepted line after the
192: %% title and affiliation information. No date will appear since the author
193: %% does not have this information. The dates will be filled in by the
194: %% editorial office after submission.
195: 
196: \begin{abstract}
197: GRB 021206 is one of the brightest GRBs ever observed.
198: Its prompt emission, as measured by RHESSI,
199: shows an unexpected spectral feature.
200: The spectrum has a peak energy of about 700~\keV\
201: and can be described by a Band function up to 4.5~\MeV.
202: Above 4.5~\MeV, the spectrum hardens again, so that
203: the Band function fails to fit the whole RHESSI
204: energy range up to 17~\MeV.
205: Nor does the sum of a blackbody function plus a power law,
206: even though such
207: a function can describe a spectral hardening.
208: The cannonball model
209: on the other hand predicts such a hardening, 
210: and we found that it 
211: fits the spectrum of GRB 021206 perfectly. 
212: We also analysed other strong GRBs observed by RHESSI,
213: namely GRBs 020715, 021008, 030329, 030406, 030519B, 
214: 031027, 031111. 
215: We found that all their spectra can
216: be fit by the cannonball model 
217: as well as by a Band function. 
218: \end{abstract}
219: 
220: %% Keywords should appear after the \end{abstract} command. The uncommented
221: %% example has been keyed in ApJ style. See the instructions to authors
222: %% for the journal to which you are submitting your paper to determine
223: %% what keyword punctuation is appropriate.
224: 
225: \keywords{
226: gamma rays: bursts
227: ---
228: gamma rays: observations
229: ---
230: techniques: spectroscopic
231: }
232: 
233: %% From the front matter, we move on to the body of the paper.
234: %% In the first two sections, notice the use of the natbib \citep
235: %% and \citet commands to identify citations.  The citations are
236: %% tied to the reference list via symbolic KEYs. The KEY corresponds
237: %% to the KEY in the \bibitem in the reference list below. We have
238: %% chosen the first three characters of the first author's name plus
239: %% the last two numeral of the year of publication as our KEY for
240: %% each reference.
241: 
242: 
243: %\newpage
244: \section{INTRODUCTION}
245: %---------------------
246: \label{sec:intro}
247: 
248: The exact mechanism which produces  $\gamma$-ray bursts (GRBs) 
249: has not yet been definitively established.
250: Their prompt $\gamma$-ray spectra can be used to distinguish 
251: between different models.
252: Several mathematical functions have been used for parametrizing
253: the prompt $\gamma$-ray emission.
254: Most commonly used is the empirical Band function \citep{Band93}, 
255: %(see \refeq{eq:Band} below), 
256: which is not motivated by a physical model.
257: 
258: There have been attempts to distinguish between spectral
259: models analysing the low energy part of the spectrum.
260: \citet{Ghirlanda2003},  \citet{Ryde2004}, and more recently
261: \citet{Ghirlanda2007} searched for blackbody components
262: in GRB spectra with varying degrees of success.
263: \citet{Preece2002}, using BATSE GRB spectra, 
264: tested the synchrotron shock model and conclude that
265: it ''does not account for the observed spectra 
266: during the GRB phase''. 
267: %TBD: Other articles to be cited?
268: 
269: % \citet{Ghirlanda2003} have studied the low energy part
270: % of GRB spectra in order to `prune down the 
271: % forest of emission models'. They find that the early
272: % phase of GRBs sometimes can be described by a thermal
273: % spectral, i.e.\ a blackbody spectrum.
274: % Also \citet{Ryde2004} discusses the fit of a thermal
275: % component and its implication for distinguishing between
276: % GRB models.
277: % In a recent study, \citet{Ghirlanda2007} find however,
278: % that a blackbody plus power law component usually  does not fit the 
279: % prompt GRB spectrum if observed over a wide enough
280: % energy range.
281: 
282: Spectral studies above the peak energy are rare,
283: one reason being the poor data quality because of lack of statistics.
284: Combining BATSE and EGRET spectra, 
285: \citet{Nature2003} report a high energy component for GRB 941017. 
286: They find a photon index of about 1.0 at energies above 5~\MeV.
287: 
288: In this paper we report a high energy 
289: component in GRB 021206 \citep{GCN_021206_IPN,GCN_021206_final},
290: observed with the Reuven Ramaty High Energy Solar Spectroscopic 
291: Imager RHESSI \citep{RHESSI}.
292: Having a peak energy of about 700~\keV, the spectrum of this burst
293: can be described by a Band function from 70~\keV\ up
294: to 4.5~\MeV, with a high energy photon index $\beta \approx 3.2$. 
295: Above 4.5~\MeV, the spectrum hardens again, and can be
296: described with a photon index  $\beta' \approx 2.2$.
297: This significant hardening around 4.5~\MeV\ can not
298: be described with a Band function. But it seems to 
299: differ from the spectral hardening in GRB 941017 as well.
300: 
301: %Dermer et al??
302: 
303: There is one model that fits the entire RHESSI spectrum
304: of GRB 021206: the cannonball model \citep{Dar2004,Dado2002,Dado2003}.
305: %Depending on the parameters, 
306: The cannonball model
307: predicts a spectral hardening at several times the 
308: peak energy with a high energy photon index reaching $\beta \approx 2.1$. 
309: 
310: The question immediately arises whether the cannonball model
311: can improve our description of other GRB spectra.
312: The difference between the Band function and the cannonball model arises
313: only at the high energy part of the spectrum, where
314: data usually suffer from low statistics.
315: Therefore, we choose the strongest GRBs registered by RHESSI
316: in the years 2002 to 2004.
317: We find that they all  can be fit
318: by the cannonball model as well as by the Band function.
319: 
320: The outline of the paper is the following:
321: %
322: We first present shortly the instrument, the GRB selection,
323: the spectrum extraction, and the fit method (\S \ref{sec:observations}).
324: %
325: In the next section (\S \ref{sec:models}), 
326: many spectral functions are given.
327: %
328: In \S \ref{sec:results}, the fit results for 
329: GRB 020715, 
330: GRB 021008, 
331: GRB 021206, 
332: GRB 030329, 
333: GRB 030406, 
334: GRB 030519B, 
335: GRB 031027, 
336: and GRB 031111
337: are presented. The fits are discussed and, 
338: if possible, compared to other measurements.
339: %%
340: The more general discussion, including
341: an outlook, follows in \S \ref{sec:discussion}.
342: %
343: We end with a short summary in \S \ref{sec:summary}.
344: 
345: \section{INSTRUMENT AND METHOD}
346: %\section{OBSERVATIONS}
347: %------------------------------
348: \label{sec:observations}
349: 
350: \subsection{Instrument}
351: % - - - - - - - - - - - - -
352: 
353: RHESSI is a NASA Small Explorer mission
354: designed to study solar flares in hard X-rays and 
355: $\gamma$-rays \citep{RHESSI}.
356: It consists of two main parts:
357: an imaging system and the spectrometer
358: with nine germanium detectors \citep{spectrometer}.
359: The satellite always points towards the Sun
360: and rotates about its axis at $15\,$rpm. 
361: The Ge detectors are arranged in a plane perpendicular 
362: to this axis. 
363: 
364: The shape of the detectors is cylindrical with a
365: height of $\approx 8.5\,$cm and diameter of $\approx 7.1\,$cm,
366: and they
367: are segmented into a thin front ($\approx 1.5\,$cm) 
368: and a thick rear segment ($\approx 7\,$cm).
369: Since the shielding of the rear segments is minimal,
370: photons with more than about 25~\keV\ can enter
371: from the side.
372: Above about 50--80~\keV, photons from any
373: direction can be observed.
374: %Thus, for GRBs, RHESSI views 65\% of the sky, the rest
375: %being occulted by the Earth.
376: Each detected photon is time- and energy-tagged from
377: 3~\keV\ to 2.8~\MeV\ (front segments) 
378: or from 20~\keV\ to 17~\MeV\ (rear segments).
379: The energy resolution is $\approx 3$~\keV\ at 1~\MeV, and the
380: time resolution is 1~$\mu$s.
381: 
382: The effective area for GRB detection
383: depends on the incident photon energy $E$
384: and the angle between the GRB direction
385: and the RHESSI axis, the incoming angle $\theta$.
386: Over a wide range of $E$ and $\theta$, the
387: effective area is around 150$\,$cm$^2$.
388: The sensitivity drops rapidly at energies below
389: $\approx 50$~\keV.
390: 
391: 
392: \subsection{GRB selection}
393: % - - - - - - - - - - - - -
394: For this study, we need well observed GRB spectra.
395: We chose GRBs from the years 2002 to 2004, because
396: radiation damage starts to play a role in 2005.
397: % The detection significance have to be $>55$.
398: The selected GRBs have to be localized by other observations of
399: the same GRB (RHESSI can not measure the incoming angle),
400: because $\theta$ enters into the simulation of the response function.
401: A further requirement was the availability of good background data. 
402: And finally, the data storing mode
403: (`rear decimation' for onboard memory saving)
404: is not allowed to change during the entire GRB and background time interval.
405: %Three GRBs from 2002, five from 2003, and none from 2004
406: %met these criteria.
407: Of all the GRBs meeting these criteria, we chose eight with
408: the best signal-to-background ratio,
409: %GRB 020715, GRB 021008, GRB 021206, 
410: %GRB 030329, GRB 030406, GRB 030519B, 
411: %GRB 031027, and GRB 031111
412: listed in Table~\ref{tab:GRBs}
413: along with their incoming angle $\theta$ and
414: the time intervals used.
415: % for spectral analysis and for background subtraction.
416: The lightcurves of these bursts are shown in
417: Figs.\ \ref{fig:ltc_GRBsI} and \ref{fig:ltc_GRBsII}.
418: 
419: 
420: \subsection{Preparation and fit of RHESSI spectra}
421: % - - - - - - - - - - - - -
422: \label{sec:fit}
423: The method of analysing RHESSI GRB spectra
424: will be described in detail in a separate article
425: (E.\ Bellm, C.\ Wigger et al., in preparation).
426: 
427: For each GRB and detector segment, 
428: the total spectrum (GRB plus background) 
429: during the burst was extracted, 
430: as well as the background spectra during two time intervals
431: before and after the burst.
432: The background was linearly (sometimes quadratically) interpolated
433: and subtracted.
434: The exact time intervals are listed in Table \ref{tab:GRBs}.
435: Then we added all rear and all front segments,
436: except for detector \#2 which
437: is slightly damaged and has a bad energy resolution.
438: Since all GRBs in this study are strong,
439: the observational errors 
440: are dominated by the statistical error of the GRB
441: counts, not of the background.
442: 
443: We simulate RHESSI using GEANT3 \citep{Geant3}.
444: %, the simulation software for high energy physics experiments by CERN.
445: Knowing the direction of the GRB from other instruments, we simulate
446: RHESSI's response to photons coming from angle $\theta$.
447: The energy of the incoming photons is simulated as
448: a power law spectrum (i.e.\ $dN/dE \propto E^{-\gamma_{sim}}$)
449: with typically $\gamma_{sim} = 2\,$. 
450: %
451: %This power law simulation is not intended to represent 
452: %the intrinsic GRB spectrum, 
453: %but instead provides simulated data representing RHESSI's conversion of
454: %photons to counts.  We determine the GRB source spectrum via weight
455: %factors for the resulting simulated count spectrum, as described below.
456: %
457: This power law simulation is only a rough approximation
458: and is not not intended to represent the intrinsic GRB spectrum, 
459: but instead provides simulated data representing RHESSI's conversion of
460: photons to counts.
461: The true GRB source spectrum is determined via weight factors 
462: for the resulting simulated count spectrum, as described below.
463: %
464: The upper energy limit of the simulated photon spectrum 
465: is typically 30 MeV, in the case of GRB 021206 even 40 MeV or 50 MeV.
466: This is important,  because an incoming
467: photon of e.g.\ 25 MeV may well make a signal of 15 MeV.
468: Rotation angles are generated uniformly,
469: i.e.\ we compute a RHESSI-spin averaged response function.
470: Since the detector arrangement
471: shows an approximate 120 degree symmetry,
472: the averaging gives good results as long
473: as the analysed time interval is at least one third of
474: the rotation period ($T_{rot} = 4\,$s). 
475: This was also confirmed by tests.
476: 
477: The output of the simulation is
478: an event list, or rather a hit list,
479: consisting of all signals registered
480: in the Ge detectors.
481: The simulated hit list, having $N_{s}$
482: entries indexed by the letter $l$, contains the deposited energy
483: ($\Edet_l$) as well as the initial photon energy ($\Ein_l$).
484: The measured hit list contains only the observed energy. 
485: 
486: For spectral fitting, the observed energy histogram
487: is compared with a histogram accumulated from the simulated
488: hit list.
489: More precisely: 
490: The measured histogram can be represented by
491: a $k$-element vector $\vec M$ with errors $\vec\sigma_M$, and 
492: energy boundaries $E^b_0$, $E^b_1$, $E^b_2$,
493: ..., $E^b_k$. 
494: We normalize the histogram $\vec M$ to the total
495: number of counts in the fit range: 
496: %$\vec{m} = \frac{1}{C_M}\vec{M} $
497: $\vec{m} = \vec{M}/C_M $
498: %and $\vec{\sigma}_m = \frac{1}{C_M}\vec{\sigma}_M $, where 
499: and $\vec{\sigma}_m = \vec{\sigma}_M /C_M $, where 
500: $C_M = \sum_{i \in I} M_i $ and the sum goes only over the
501: bins included in the fit, i.e.\ 
502: $I = \{i \mid \mbox{bin } i \mbox{ is included in the fit} \}$.
503: %
504: The 'theoretical' histogram $\vec{S}$ is accumulated from the 
505: simulated hit list. % in the following way:
506: %
507: Each entry is weighted with a factor in order to scale 
508: %each count resulting 
509: from the simulated power law 
510: to the probability density which would be expected, 
511: had we actually simulated the GRB source spectrum dN/dE.
512: The $j$th %histogram 
513: bin contains therefore the weighted sum of all simulated hits
514: with $\Edet_l$ belonging to that bin, i.e.: 
515: \begin{equation}
516:    S_j = \sum_{l \in L} w_l
517: \end{equation}
518: where $L = \{l \mid E^b_{j-1} \leq \Edet_l < E^b_j\}$
519: and 
520: \begin{equation}
521:    w_l = \left( \frac{\Ein_l}{E_{piv}} \right)^{\gamma_{sim}} \cdot
522:        \frac{dN}{dE}(\Ein_l)   \label{eq:weight}
523: \end{equation}
524: % 
525: %The weight factor scales each count resulting from the 
526: %simulated power law to the value which would be expected, 
527: %had we actually simulated the GRB source spectrum dN/dE.
528: %
529: The first factor in eq.\ \ref{eq:weight} accounts
530: for the spectrum assumed in the simulation and 
531: the energy $E_{piv}$ is an arbitrary normalisation.
532: The second factor accounts for
533: the spectrum of the incoming GRB photons. 
534: Possible parametrisations
535: of $dN/dE$ are given below in \S \ref{sec:models}.  
536: If the GRB spectrum had 
537: the same shape as the simulated one,
538:  i.e.\ if
539: $ dN/dE = (E/E_{piv})^{-\gamma_{sim}}$, 
540: the weights would all be 1 .
541: This  method of using weight factors when filling a histogram
542: is common in particle physics, see e.g.\ \citet{Barlow93}. 
543: %
544: The statistical error of the 'theoretical'
545: histogram $\vec{S}$ is $\sigma^2_{S_j} = \sum_{l \in L} w_l^2 $
546: \citep[\S 6 of][]{Barlow93}.
547: %%%%
548: %The method of using weighted entries for a histogram
549: %is common in particle physics, see e.g.\ \citet{Barlow93}, 
550: %in particular \S 6 therein.
551: %
552: As in the case of the measured histogram,
553: the histogram $\vec{S}$
554: is normalised:
555: %$\vec{s} = \frac{1}{C_S}\vec{S} $
556: $\vec{s} = \vec{S}/C_S $
557: %and $\vec{\sigma}_s = \frac{1}{C_S}\vec{\sigma}_S $ with 
558: and $\vec{\sigma}_s =\vec{\sigma}_S /C_S  $ with 
559: $C_S = \sum_{i \in I} S_i $. 
560: 
561: The parameters of the histogram $\vec{S}$ are varied
562: until the minimum of
563: \begin{equation}
564:    \chi^2 = \sum_{i \in I} \frac{(m_i - f s_i)^2}
565:       {\sigma_{m_i}^2 + \sigma_{s_i}^2}
566: \end{equation}
567: is found. The factor $f$ accounts for the
568: normalisation between measured and simulated histogram.
569: It is expected to be almost 1, but
570: should be treated as a free fit parameter.
571: For each fit iteration, the histogram $\vec{S}$ is recalculated
572: with different weights (eq.\ \ref{eq:weight}).  
573: 
574: Since the simulated hit list contains many more
575: photons than the measured spectrum, %so that 
576: we used the approximation $\sigma_{m_i}^2 + \sigma_{s_i}^2 \approx 
577: \sigma_{m_i}^2$ while fitting.
578: But it was always checked that
579: the statistical error from the measurement is
580: dominant.
581: 
582: %It is possible to construct a response matrix from our
583: %simulated count. One of us (EB) developed a response
584: %matrix independently from the routines described in this article.
585: %With his response matrix, he successfully could do
586: %joint fits of RHESSI/Swift or RHESSI/Konus data, see e.g. ...
587: %We checked that the two response matrices agree.
588: %%   TODO -- done!
589: It is possible to create a response matrix from our 
590: simulations and perform spectral fits via forward-fitting,
591: as in  XSPEC.  
592: In any case, our weighted histogram method gives equivalent
593: fits to response matrices which are simulated directly 
594: %We made cross checks that our weighted histogram method gives 
595: %equivalent results to response matrices which are simulated directly
596: (by EB; see E.\ Bellm, C.\ Wigger et al., in preparation).
597: 
598: 
599: %If the data are noisy, $\chi^2$ will be high even for 
600: %a good fit, as is the case e.g. for GRB 030519B.
601: %If, on the other hand, a good fit has a low $\chi$,
602: %as e.g. Band function in the case of GRB 021008 (rear),
603: %then even a fit with an acceptable $\chi^2$ looks bad. 
604: %This is the case for GRB 021008 (rear), when fitted with a BPL.
605: 
606: %From Table~\ref{tab:chi2} we conclude that
607: %the model with the best $\chi^2_r$ fits the data. 
608: %And all other models
609: %that have $= \chi^2_{r,bm} + f\,\Delta \chi^2$,
610: %with $\chi^2_{r,bm}$ is the $\chi^2_r$ of the best fitting 
611: %model and  $f =0.5$ -- $1.0$ .
612:  
613: 
614: 
615: \subsection{Systematic effects}
616: \label{sec:systematics}
617: At low energies, a small deviation of our RHESSI mass model 
618: from the true amount of material 
619: can make a considerable difference in the number of 
620: observed photons.
621: For $\theta\approx 90$ degrees,
622: this should be a small problem because the 
623: lateral shielding is thin. But for $\theta$
624: from 10 to 50 degrees this is an 
625: issue, and less prominently also
626: from 130 to 160 degrees.
627: 
628: %Since the high energy response function
629: %is less affected by a not so excellent mass model,
630: %we trust the high energy part of our fits.
631: 
632: Simulation quality also gets better with higher energy.
633: This is fortunate for the current analysis which relies on
634: high-energy properties of GRB spectra. 
635: 
636: 
637: \section{SPECTRAL MODELS}
638: % - - - - - - - - - - - - - - - 
639: \label{sec:models}
640: 
641: Let $dN/dE$ be the number of GRB photons per energy bin.
642: The peak energy \Epeak\ is defined as the energy for which
643: $\mathEFE = dN/dE \, E^2$ is maximal. 
644: The spectrum in the \EFE\ representation has at least one maximum because
645: the total emitted energy must be finite: $\int_0^\infty dN/dE\,E\, dE < \infty$.
646: Many instruments can see such a maximum \Epeak\ within their energy range.
647: 
648: Different mathematical functions, sometimes called models,
649: can describe such a shape. A collection is presented 
650: in this section. 
651: %We will call them functions, since 
652: %we prefer to use the word ``model'' in the sense
653: %of a real physical explanation of the GRB phenomenon.
654: 
655:   %This is a mathematical reason only,
656:   %and it could well be that the peak energy lays outside
657:   %the observed energy band, but indeed, most GRB spectra
658:   %show such a behaviour.
659: %We call low (high) energy part of the spectrum 
660: %the one that lies below (above) \Epeak.
661: 
662: %In this study, we use several spectral models:
663: 
664: The simplest spectral function is a power law ({\bf PL}):
665:   \begin{equation} 
666:     \frac{dN}{dE} = A \left( \frac{E_{piv}}{E} \right)^\gamma
667:     \label{eq:PL}
668:   \end{equation}
669: where $E_{piv}$ is a normalization energy, e.g. $E_{piv}=100$~\keV.
670: The PL has no peak energy.
671: It rarely fits a GRB spectrum over the entire observed
672: energy range, but it is often useful for a limited energy band.
673: Indeed, every spectrum can be fit by several joined PLs.
674: 
675: One simple way to account for a spectral
676: softening and a peak energy is the cut off power law ({\bf CPL}):
677: \begin{equation}
678:   \frac{dN}{dE} = A \left( \frac{E_0}{E} \right)^\alpha e^{-E/E_0}
679:   \label{eq:CPL}
680: \end{equation}
681: If $\alpha < 2.0$ then $\mathEpeak = E_0\, (2-\alpha)$ .
682: 
683: Another way to account for a spectral break is the
684: broken power law ({\bf BPL}) consisting of two joined PLs
685: \begin{equation}
686:   \frac{dN}{dE} = 
687:   \left\{
688:     \begin{array}{ll}
689:       A \left( \frac{E_b}{E} \right)^\alpha & \mbox{ if } E\leq E_{b} \\
690:       A \left( \frac{E_b}{E} \right)^\beta  & \mbox{ if } E\geq E_{b}.
691:     \end{array}
692:   \right.
693:   \label{eq:BPL}
694: \end{equation}
695: If $\alpha < 2.0$ and $\beta > 2.0$, then $\mathEpeak = E_b$ .
696: This function is not continuously differentiable.
697: 
698: A smooth transition between the two power laws 
699: is realized by the empirical {\bf Band} function \citep{Band93}.
700: This is a smooth composition of a CPL for low energies and
701: a PL for high energies:
702: \begin{equation}
703:   \frac{dN}{dE} =
704:   \left\{
705:     \begin{array}{ll}
706:        A \left( \frac{E_{piv}}{E} \right)^\alpha e^{-E/E_0}  & \mbox{ if } E \leq \mathEbreak \\ 
707:        B \left( \frac{E_{piv}}{E} \right)^\beta 	     & \mbox{ if } E \geq \mathEbreak
708:     \end{array}
709:   \right.
710:   \label{eq:Band}
711: \end{equation}
712: where
713: \[
714:    \mathEbreak =  E_0 (\beta-\alpha)
715: \]
716: and
717: \[
718:    B = A \left(\frac{E_0}{E_{piv}} (\beta-\alpha)\right)^{\beta-\alpha}  
719:      e^{-(\beta-\alpha)}
720: \]
721: Again, $E_{piv}$ is a normalization energy, e.g. $E_{piv}=100$~\keV.
722: If $\alpha < 2.0$ and $\beta > 2.0$, then $\mathEpeak = E_0\, (2-\alpha)$.
723: If $\beta \longrightarrow \infty$ or if \Ebreak\ lies 
724: at the upper limit of the observed energy range, 
725: then the Band function turns into a CPL.
726: As already pointed out by \citet{Preece2000}, section 3.3.1.,
727: the low energy photon index, the curvature of the spectrum,
728: and its peak energy are represented with only two parameters,
729: $\alpha$ and $E_0$.\footnote{The smoothly broken power law model (SBPL, 
730: see \citet{Preece2000,BATSEcatalog})
731: would account for this problem with an additional parameter, 
732: but we do not use it here, because it can not fit
733: a spectral hardening at high energies.}
734: 
735: Sometimes a blackbody spectrum plus power law
736: is used for spectral fitting, 
737: see e.g.\ \citet{Ryde2004,Ghirlanda2007,SMcBreen2006}. 
738: We will call this the {\bf BBPL}:
739: \begin{equation}
740:   \frac{dN}{dE} =  A  \frac{(E/(kT))^2}{\exp(E/(kT))-1} + 
741:        A\,b\, \left( \frac{E_{piv}}{E} \right)^\alpha  
742:   \label{eq:BBPL}
743:   % gewichte = (se0/100.)^(betasim-alpha) + $
744:   %    (anteil / ( (3.92*kT/100.)^2/(exp(3.92)-1)) ) * $
745:   %    (se0/100.)^2/(exp(se0/kT)-1) * (se0/100.)^betasim
746: \end{equation}  
747: We choose $E_{piv} = 3.92\, kT$, the
748: peak energy of the blackbody component.
749: %For $E = 3.92 kT$ - i.e.\ at the peak energy of the blackbody
750: %component - the relative fraction of the power law component
751: %is $b$.
752: The BBPL function can fit a spectral hardening at high 
753: energies.
754: 
755: When fitting with the BBPL model, it is often found that
756: the PL component does not fit simultaneously at low
757: and at high energies. This is also mentioned by \citet{Ryde2004}.
758: We therefore invented a blackbody plus modified power law
759: ({\bf BBmPL}):
760: \begin{eqnarray}
761:   \label{eq:BBmPL}
762:   \frac{dN}{dE} =  A  \frac{(E/(kT))^2}{\exp(E/(kT))-1}\, + \\
763:                    A \, b \, (1-e^{-E/E_0}) \left( \frac{E_0}{E} \right)^\alpha \nonumber
764: \end{eqnarray}
765: We choose again $E_0=3.92\, kT$.
766: %, because 
767: %this is the peak energy of the blackbody component.
768: The modification of the power law component was borrowed from
769: the cannonball model (see next). The BBmPL function can 
770: describe a spectral hardening at high energies.
771: 
772: %     Teff=3.92*kT
773: %  gewichte = (se0/100.)^2/(exp(se0/kT)-1.) * (se0/100.)^betasim $
774: %             + anteil * (1.-exp(-se0/kT)) *(se0/100)^(betasim-alpha)
775: %        ;     + anteil * (1.-exp(-se0/Teff)) *(se0/Teff)^(betasim-alpha)
776: 
777: %\item[CB] cannonball model
778: The cannonball model {\bf \CB} 
779: \citep{Dar2004,Dado2002,Dado2003}
780: makes a prediction for the
781: spectral shape of the prompt GRB emission.
782: It consists of a CPL and a modified power law:
783: \begin{eqnarray}
784:   \label{eq:CB}
785:   \frac{dN}{dE} = A \left( \frac{T}{E} \right)^\alpha  e^{-E/T} +  \\
786: 	          A\, b\, \left( \frac{T}{E} \right)^\beta \left( 1- e^{-E/T}\right)  \nonumber 
787: \end{eqnarray}
788: according to \citet{Dar2004},  eq.~47, or \citet{Dado2004}, eq.~13.
789: The theoretically expected values are
790: $\alpha \approx 1.0$ and 
791: $\beta \approx 2.1$. 
792: %In principle, this function does not apply for the 
793: %entire time integrated GRB spectrum, but for each pulse 
794: %separately. 
795: The \CB\ function \refeq{eq:CB} applies strictly speaking 
796: only to the spectrum caused by a single cannonball,
797: i.e.\ for every single peak of a GRB.
798: 
799: It is often observed that the peak energy \Epeak\ is
800: a more stable fit parameter than the parameter $E_0$ in the Band function
801: (\refeq{eq:Band}) or in the CPL (\refeq{eq:CPL}).
802: Therefore, we use 
803: $E_p = E_0\, (2-\alpha)$ as a fit parameter. Similarly, we use
804: %$T_p = T_{eff}\, (2-\alpha)$ as fit parameter in the case of 
805: $T_p = T\, (2-\alpha)$ as a fit parameter in the case of 
806: \CB\ (\refeq{eq:CB}).
807: 
808: A word about fitting \CB:
809: For the high energy part, it has two parameters, whereas
810: the Band function has only one.
811: Already when fitting the Band function, it is
812: often observed that the high energy power law index
813: is poorly constrained, because the high energy data
814: tend to have large statistical errors.
815: This is even worse for \CB\ with two
816: high energy parameters. 
817: %It often helps to restrain
818: %the high energy power law index $\beta$ to its
819: %theoretically expected value 2.1 .
820: %
821: %The following paragraph was moved here from the former conclusion section
822: %The high energy power law index $\beta$ of the \CB\ model
823: %often has to be fixed 
824: It often helps to freeze the parameter $\beta$
825: at its theoretical value of $\beta=2.1$
826: in order to make the fit converge. 
827:   
828: 
829: 
830: 
831: \section{FIT RESULTS AND FIT DISCUSSIONS}
832: %-------------------------------------------------
833: \label{sec:results}
834: 
835: % The upper limit of the fit range is either the maximum RHESSI energy 
836: %(16 MeV slightly time dependent) or a few times the highest 
837: % GRB energy observed. 
838: % For bursts with no counts at high energies, we do not use the full 
839: % RHESSI energy range up to 16 MeV, because we want to fit the GRB spectrum, 
840: % not background.
841: %
842: % The low energy end of the fit range depends on the incoming angle. 
843: % It is determined empirically:
844: % E.g.\ if a fit starting at 100 keV also fits down to 70 keV, 
845: % we rather used 70 keV than 100 keV as lower bound.
846: % A general rule is that the more material is in between the GRB 
847: %and the detectors, the higher the low energy bound.
848: 
849: The spectral models used and the $\chi^2$ of
850: the fits are listed in Table~\ref{tab:chi2} for all eight GRBs. 
851: For \CB\ and Band function, the fitted
852: parameters are listed in Tables~\ref{tab:CB_pars}
853: and \ref{tab:Band_pars}, respectively.
854: Throughout this article, all errors are symmetric $1\sigma$ errors
855: if not stated otherwise.
856: 
857: The measured spectra together with the \CB\ and Band fits
858: are shown in Figs.~\ref{fig:020715} to \ref{fig:031111}.
859: Note that we plot energy$^2 \,\cdot\,$counts/keV versus energy.
860: %% TODO
861: %I agree with the referee's concern about what you're
862: %calling the plotted counts.  Calling it E*F[E] will make people think
863: %you're plotting the "unfolded" spectrum.  I'd simply say you're plotting
864: %energy^2. counts/keV.
865: %% 
866: %When plotting the results, %we do not act on measured data,  
867: %%but only on simulations. The figures  
868: %we show the observed \EFE\ spectrum. 
869: The difference to a deconvolved \EFE\ distribution is
870: %(where $E$ is the energy of the incoming GRB photons) is
871: %don't show the GRB's original spectrum, but its response in RHESSI. 
872: %The difference is 
873: discernible e.g.\ in the drop of counts
874: towards lower energies in our representation.
875: The statistical scatter from the limited number
876: of simulated events is sometimes visible as a little roughness of
877: the simulated spectra.\footnote{
878: In the case of e.g.\ GRB 021206, rear (see Fig.~\ref{fig:spec_rear}), 
879: the mean measured error between 4 and 12 MeV is  
880: $0.65 \cdot 10^6\,$counts$\cdot$keV,
881: whereas the mean scatter of the simulated histogram
882: is $0.25 \cdot 10^6\,$counts$\cdot$keV. For the other
883: GRBs with less observed photons and therefore larger measurement errors,  
884: the statistical error of the
885: simulation is even more negligible.}
886: 
887: From the fit parameters obtained for the
888: \CB\ and the Band function, we calculate the fluences
889: %\fluenceCB\ and \fluenceBand\ 
890: for various energy intervals.
891: They are listed in Table~\ref{tab:fl}.
892: %
893: %The fluences \fluenceCB\ and \fluenceBand\ were determined for the fit ranges. 
894: %
895: The error of the fluence is dominated by systematics,
896: e.g.\  because we do not know the exact active volume 
897: of the single detector segments. 
898: We estimate the systematic error to be of order 5\%,
899: whereas the statistical error %of our determination of the fluence 
900: is of order 1\%.
901: Note also, that the two fluences
902: obtained by fitting \CB\ and Band function
903: are nearly equal. 
904: 
905: 
906: \subsection{GRB 020715}
907: %\paragraph{GRB 020715}
908: % - - - - - - - - - - - - - -
909: %\paragraph{Fit}
910: The lightcurve is shown in Fig.~\ref{fig:ltc_GRBsI} (top) and 
911: the spectrum is shown in Fig.~\ref{fig:020715}.
912: %The lightcurve is shown in Fig.~\ref{fig:grbsI}, upper plot,
913: %and the spectrum in Fig.~\ref{fig:020715}.
914: Two bins from 290~\keV\ to 310~\keV\ (a background line) 
915: and one bin from 500~\keV\ to 525~\keV\ (the 511~\keV\ line) 
916: were omitted in the fit 
917: because they would dominate $\chi^2$.
918: The best fit %(see Table~\ref{tab:chi2}) 
919: is a Band function, but the \CB\ also fits well.
920: % with: 
921: %\begin{eqnarray}
922: %  \mathEpeak & = & 531 \pm 24 \mbox{ keV} \nonumber \\ 
923:  % \alpha   & = & 0.776 \pm 0.044 \\ 
924:  % \beta    & = & 3.14 \pm 0.25  \nonumber
925: %\end{eqnarray}
926: %The second best fit is the Cannonball model.  %with:
927: %\begin{eqnarray}
928: %  \mathEpeak & = & 539 \pm 21 \mbox{keV}  \nonumber \\ 
929: %  \alpha   & = &  0.774 \pm 0.059    \nonumber \\ 
930: %  \beta   & = & 2.19 \pm 0.24 \\ 
931: %   b   & = &  0.055 \pm 0.039   \nonumber    
932: %\end{eqnarray}
933: %The spectral parameters are listed in Table~\ref{tab:Band_pars}
934: %for the Band function fit and in Table~\ref{tab:CB_pars}
935: %for the Cannonball model.
936: 
937: %\paragraph{GRB 020715}
938: %.. not much to be said ...
939: 
940: 
941: \subsection{GRB 021008}
942: % - - - - - - - - - - - - - 
943: Coming from a direction about 50 degrees from the Sun,
944: this GRB deposited photons not only in rear
945: detectors but also in the front detectors, 
946: as can be seen from the lightcurve in Fig.~\ref{fig:ltc_GRBsI}.
947: The front spectrum is shown in Fig.~\ref{fig:F021008}
948: and the rear in Fig.~\ref{fig:R021008}.
949: 
950: \paragraph{Fit}
951: We had difficulties to fit front and rear spectra consistently
952: below 300~\keV. We therefore chose 300~\keV\ as the lower energy bound. 
953: Band function and \CB\ give the best fits.
954: 
955: We fitted front and rear segments separately, as well as jointly.
956: The results of the joint fit are shown in 
957: Figs.~\ref{fig:F021008} and \ref{fig:R021008} .
958: For the joint \CB\ fit we find the 90\% confidence level (CL) errors:
959: \begin{eqnarray}
960:    T_p    &   =    & 641   \; ^{+54}_{-53} \mbox{ keV} \nonumber	        \\
961:    \alpha &   =    & 1.523 \; ^{+0.099}_{-0.098}	    \label{eq:CB021008} \\
962:    \beta  & \equiv & 2.1  			    \nonumber	        \\
963:     b     &   =    & 0.0198\; ^{+0.0129}_{-0.0137}  \nonumber
964: \end{eqnarray}
965: %   $T_p = 641\; ^{+54}_{-53} \,$keV, 
966: %   $\alpha = 1.523\; ^{+0.099}_{-0.098}$,
967: %   $\beta  = 2.1$ (frozen), and  
968: %   $b      =  0.0198\; ^{+0.0129}_{-0.0137} $.
969: The total $\chi^2$ is $71.2$  for $n_{DoF}= 74$.
970: The Band function fits marginally better,
971: $\chi^2= 70.0$ for $n_{DoF}= 74$ 
972: %(fitting  front and rear simultaneously), 
973: and its parameters are (\CLerrors):
974: \begin{eqnarray}
975:    E_p    &=& 677  \; ^{+51}_{-66}       \mbox{ \keV} \nonumber \\
976:    \alpha &=& 1.493\; ^{+0.104}_{-0.104} \label{eq:Band021008} \\
977:    \beta  &=& 3.73 \; ^{+0.48}_{-0.38}	 \nonumber
978: \end{eqnarray}
979: %$   E_p = 677\; ^{+51}_{-66} \,$keV,
980: %$   \alpha = 1.493\; ^{+0.104}_{-0.104}$, and   
981: %$   \beta  = 3.73\; ^{+0.48}_{-0.38}$.   
982: 
983: 
984: \paragraph{Discussion}
985: %\paragraph{GRB 021008}
986: We do not well understand the  
987: spectrum below 300~\keV.
988: Both fits, the \CB\ and the Band function,
989: overestimate the counts below 300~\keV. This can be a hint that
990: the GRB spectrum hardens below 300~\keV.
991: We find functions that fit the front and
992: the rear data from 40~\keV\ to 400~\keV\ individually,
993: but they do not agree. 
994: 
995: One possible explanation is the GRB incoming angle of 
996: about 50 degrees at which the
997: GRB photons pass through a certain amount
998: of material before reaching the detectors.
999: Our GEANT simulation tries to take that into account, but it
1000: is probably not perfect, and maybe the averaging over all rotation
1001: angles is a bad assumption for this short GRB pulse.
1002: %(see e.g. Fig.\ 2? in \citet{PSI2004} for a simplified view of RHESSI). 
1003: %Therefore, the simulation at low energies could be less trustworthy 
1004: %than in the case with incoming angles of about 90 degrees,
1005: %where the detectors are almost unshielded.
1006: 
1007: %Fitting the rear data from 40 keV to 400 keV, 
1008: %we find as best fitting model a BPL with
1009: %$E_1=166 \pm 12\,$keV, $\alpha = 0.914 \pm 0.049$ and 
1010: %$\beta = 1.796 \pm 0.041$ ($\chi^2 /n_{DoF} = 72.2/73 $).
1011: %Fitting the front data from 40 keV to 400 keV, 
1012: %we find as best fitting model a BBPL with
1013: %$kT = 48 \pm 4\,$keV, $\alpha = 1.85 \pm 0.13$ and 
1014: %$b_{PL} = 0.29 \pm 0.21$ ($\chi^2 /n_{DoF} = 77.2/73 $),
1015: %and as second best fit a SPL with  $\alpha = 1.638 \pm 0.014$
1016: %($\chi^2 /n_{DoF} = 82.2/75 $).
1017: %These two sets of parameter do not really agree.
1018: 
1019: Another difficulty for this GRB is its background.
1020: For the single rear segments, the background at low
1021: energies (below $\approx 120$~\keV)
1022: strongly depends on the rotation angle of RHESSI.
1023: We did our best to take this into account, 
1024: but maybe did not succeed completely.
1025: 
1026: 
1027: 
1028: 
1029: \subsection{GRB 021206}
1030: % - - - - - - - - - - - 
1031: 
1032: \label{sec:result021206}
1033: GRB 021206 is famous for its claimed
1034: polarization \citep{CB03}, which however
1035: turned out to be an artefact 
1036: \citep[see][]{RF04,PSI2004,cw_rome}.
1037: This GRB was also studied by \citet{Boggs2004} 
1038: to probe quantum gravity.
1039: 
1040: This GRB is only 18 degrees from the Sun, exposing
1041: mainly the front segments of RHESSI's detectors.
1042: Its lightcurve is shown in Fig.~\ref{fig:ltc_GRBsI}.
1043: Fig.~\ref{fig:spec_front} shows the energy spectrum 
1044: in the front segments, Fig.~\ref{fig:spec_rear}
1045: in the rear segments.
1046: 
1047: 
1048: \paragraph{Fit}
1049: The front spectrum can be fit from 70~\keV\ up to 2800~\keV,
1050: and the rear 
1051: % spectrum %(Fig.~\ref{fig:spec_rear}) can be fit 
1052: from 300~\keV\ to 16~\MeV.
1053: The huge number of excess counts below $300$~\keV\ in the 
1054: rear detectors is understood:
1055: The geometrical constellation of the GRB, RHESSI, and Earth
1056: was such that the GRB photons came from the front
1057: direction, where the effective area is relatively small,
1058: whereas the Earth was behind RHESSI so that the
1059: backscattered photons could easily reach the rear
1060: segments.
1061: 
1062: The only function that fits the front {\em and} rear spectra 
1063: over the entire energy range from 70~\keV\ up to 16~\MeV\ is the \CB.    
1064: %One can even freeze $\beta$ at its
1065: %theoretically expected value of 2.1\,.
1066: %The fitted parameters are listed in Table~\ref{tab:CB_pars}.
1067: %
1068: %We used different spectral models for different energy ranges.
1069: %The front data can be well fit starting at 56 keV up to 
1070: %2800 keV. The rear data below 300 keV are dominated by 
1071: %GRB photons which were scattered in the atmosphere and
1072: %then reach the detectors. Therefore, we restrict our fits
1073: %of the rear data to $> 300$ keV.
1074: %We fitted front and rear segments independently.
1075: %The results of all the fitting is summarised in Table
1076: %\ref{tab:fit_021206}.
1077: %
1078: %We also note that:
1079: %
1080: %$\bullet$ 
1081: %The full evaluation of the parameter space with frozen
1082: %$\beta$ gives for the \CB\ (\CLerrors):
1083: %\begin{eqnarray}
1084: %   T_p &=& 680 ^{+13}_{-13} \mbox{ keV} \nonumber \\
1085: %   \alpha &=& 0.593 ^{+0.042}_{-0.043}   \\
1086: %   \beta  &=& 2.1                        \nonumber \\
1087: %    b     &=& 0.0994 ^{+0.0097}_{-0.0096}   \nonumber
1088: %\end{eqnarray}
1089: %These values are used for the black line histogram 
1090: %and the residuals in Figs.\
1091: %\ref{fig:spec_front} and \ref{fig:spec_rear}. $\chi^2_rs=$? 
1092: %
1093: %$\bullet$ 
1094: %Treating $\beta$ in the \CB\ as a free fit parameter yields
1095: %$\beta = 2.12 \pm 0.13$ for the rear spectrum and
1096: %$\beta = 1.88 \pm 0.25$ for the front spectrum, both consistent
1097: %with $\beta = 2.1$. 
1098: %
1099: %$\bullet$ 
1100: Fitting front and rear spectra simultaneously with \CB\ 
1101: and all parameters free, yields (\CLerrors):
1102: \begin{eqnarray}
1103:    T_p    &= 678  \; ^{+13}_{-10}       \mbox{ \keV} \nonumber \\
1104:    \alpha &= 0.60 \; ^{+0.09}_{-0.08}   \label{eq:CB021206}    \\
1105:    \beta  &= 2.12 \; ^{+0.08}_{-0.13}   \nonumber              \\
1106:     b     &= 0.108\; ^{+0.039}_{-0.045} \nonumber
1107: \end{eqnarray}
1108: with $\chi^2 = 187.5$ for $n_{DoF}= 185$.
1109: These values are used for the black line histogram 
1110: and the residuals in Figs.~\ref{fig:spec_front} and \ref{fig:spec_rear}. 
1111: 
1112: The front spectrum alone is well described
1113: by a Band function. Its $\chi^2$ is even marginally smaller
1114: than that of the \CB\ model fit, see Table~\ref{tab:chi2}.
1115: The Band function also fits the rear spectrum up to 4.5~\MeV\ with 
1116: $\chi^2 = 72.8$ (62 DoF), but not at higher energies.
1117: The front and rear parameters (up to 4.5~\MeV) agree,
1118: see Table~\ref{tab:Band_pars}. Evaluating the full parameter space
1119: simultaneously for front and rear yields (\CLerrors):
1120: \begin{eqnarray}
1121:    E_p    &= 711  \; ^{+15}_{-17}       \mbox{ \keV} \nonumber \\
1122:    \alpha &= 0.692\; ^{+0.033}_{-0.039} \label{eq:Band021206}  \\
1123:    \beta  &= 3.19 \pm 0.08                           \nonumber
1124: \end{eqnarray}
1125: with $\chi^2=176.5$ for $n_{DoF}= 174$.
1126: These values are used for the grey line histogram in 
1127: Figs.~\ref{fig:spec_front} and \ref{fig:spec_rear}.
1128: They agree with the preliminary results by \citet{cw_venice}.
1129:  % from cw_venice:
1130:  %   The weighted means of the spectral parameters are ($1 \sigma$ errors):
1131:  %   $\alpha =  -0.65 \pm 0.04$,
1132:  %   $\mathEpeak = 705 \pm 14 \,$keV, and
1133:  %   $\beta = -3.19 \pm 0.06$.
1134: %$\bullet$ 
1135: Above 4.5~\MeV, a PL with $\gamma = 2.23 \pm 0.21$ fits
1136: the data ($\chi^2 = 12.5$ for 10 DoF).
1137: 
1138: 
1139: 
1140: \paragraph{Discussion}
1141: % - - - - - - - - - - - -
1142: As can be learned from Table~\ref{tab:chi2},  
1143: the high energy part can not be fit
1144: by Band, BPL or CPL, and the 
1145: low energy part of the spectrum can not be fit
1146: by BBPL nor by BBmPL. 
1147: The only function that fits over the whole RHESSI energy range
1148: is \CB.
1149: 
1150: The \CB\ function has one parameter more than the Band function.
1151: An F-test indicates that the chance
1152: probability of producing such an improvement in $\chi^2$ with the
1153: additional parameter
1154: is only $4.0\times10^{-9}$.
1155: The spectral hardening at 4.5~\MeV\ is significant. 
1156: 
1157: Because this GRB has so many counts at high energies,
1158: we used a simulation with $\gamma_{sim} = 1.75$
1159: for the results cited above.
1160: A power law index of 1.75  results in relatively more counts
1161: at high energies than the usual power law index (=2).
1162: We also used simulations with $\gamma_{sim} = 1.5$
1163: and $\gamma_{sim} = 2.0$. The results were almost
1164: identical, especially for the high energy parameters 
1165: $\beta$ and $b$ of the \CB\ fit.
1166: 
1167: The high energy photon index $\beta$ of the \CB\ function
1168: agrees perfectly with the theoretical expected value ($\approx 2.1$).
1169: The low energy photon index $\alpha \approx 0.6$ on the other hand 
1170: is slightly smaller than expected from theory ($\approx 1.0$). 
1171: %The spectral shape of GRB 021206 at high energies 
1172: %follows the \CB\ model, with a high energy  
1173: %power law index of 2.1$\,$.
1174: 
1175: %One also should note that the $b$-value found for
1176: %the front fit, where the spectral hardening is not
1177: %visible, is the same that is found for the rear
1178: %fit, where the high energy tail is visible.
1179: 
1180: \paragraph{Peak resolved analysis}
1181: % - - - - - - - - - - - -
1182: 
1183: %The \CB\ function applies strictly only to the single
1184: %peaks, caused by single cannonballs, of a GRB.
1185: The time structure of GRB 021206 is rather intricate.
1186: Four periods of emission can be distinguished, 
1187: see Fig.~\ref{fig:ltc_GRBsI}, each of them probably 
1188: consisting of several overlaying sub-peaks.
1189: Luckily, these time periods match quite well our
1190: minimum time resolution of one third of a rotation period
1191: for fitting with a rotation averaged response function
1192: (see \S \ref{sec:fit}).
1193: 
1194: The fitted parameters are listed in 
1195: Table~\ref{tab:dt_grb021206} for the four time
1196: intervals marked in the figure,
1197: as well as the additional {\em tail} interval.
1198: The {\em tail} interval lasts one full rotation,
1199: starting at the end of the P4 interval.
1200: The fluences of the two components in the \CB\ function (\refeq{eq:CB})
1201: are listed separately (\fluenceCPL\ for the CPL component 
1202: and \fluencemPL\ for the modified PL component).
1203: The mPL index $\beta$ was kept frozen at 2.1.\ %
1204: 
1205: The energy $T_p$ increases from the first to the second time interval
1206: and then decreases. Also $F_{CPL}$ and $F_{mPL}$ increase
1207: from the first to the second interval, and then decrease.
1208: However, the modified PL component seems to decay more slowly
1209: than the CPL component. The tail is dominated by the mPL
1210: component.
1211: 
1212: 
1213: %\newpage 
1214: 
1215: 
1216: \subsection{GRB 030329}
1217: % - - - - - - - - - - - 
1218: %\label{sec:res_030329}
1219: GRB 030329 is famous for the supernova 2003dh detected
1220: in its afterglow  %\citep{GCN_sn_obs,GCN2131,GCN2169}.
1221: \citep{Hjorth2003,GCN_sn_obs,GCN2131,GCN2169}.
1222: The authors of the \CB\ model used
1223: the lightcurve of this GRB and its early
1224: afterglow to predict the later afterglow and 
1225: the appearance of a supernova
1226: \citep[see][]{Dado_030329}.
1227: A supernova and the late afterglow was also predicted by
1228: \citet{Zeh2003}, and is discussed in \citet{Ferrero2006}.
1229: 
1230: In the lightcurve of GRB 030329, two peaks are
1231: clearly separated (Fig.~\ref{fig:ltc_GRBsI}, bottom plot).
1232: We analyse them separately.
1233: 
1234: \paragraph{Fit}
1235: The spectrum of the first peak
1236: is shown in Fig.~\ref{fig:030329_p1}.
1237: Three bins around a background line from 64~\keV\ to 70~\keV\ were 
1238: not included in the fit. 
1239: A very good fit (see Table~\ref{tab:chi2}) is a CPL with: 
1240: $  \mathEpeak  =  158.7 \pm 5.0 \mbox{ keV}$   % \nonumber \\ 
1241: and $  \alpha    =  1.662 \pm 0.032 .$
1242: Not surprisingly, the \CB\ and Band function, having more parameters,
1243: but being closely related to a CPL, fit only marginally better. 
1244: 
1245: The spectrum of the second peak
1246: is shown in Fig.~\ref{fig:030329_p2}.
1247: The spectrum has some wiggles below 160~\keV,
1248: which account for the relatively high $\chi^2$.
1249: Many models give an acceptable fit, only BBPL 
1250: does not fit. 
1251: A good fit (see Table~\ref{tab:chi2}) is a CPL with 
1252: $  \mathEpeak  =  78 \pm 13 \mbox{ keV} $   %\nonumber \\ 
1253: and $  \alpha    =  1.83 \pm 0.05 $.
1254: Also a good fit is a broken power law (BPL) with %:
1255: $  E_b  =  175 \pm 15 \mbox{ keV} $,  %\nonumber \\ 
1256:  $ \alpha    =  1.985 \pm 0.032$, and   %\\ 
1257:  $ \beta     =  2.68 \pm 0.08 $. %\nonumber
1258: Also Band function and \CB\ model fit the data well,
1259: see Fig.~\ref{fig:030329_p2}.
1260: 
1261: \paragraph{Discussion}
1262: % - - - - - - - - - - -
1263: The prompt emission was detected by HETE.
1264: Its spectrum is published by \citet{HETE030329}
1265: and by \citet{HETE2005}.
1266: \citet{HETE030329} do a time resolved analysis.
1267: For the entire burst they find (\CLerrors): 
1268: $\mathEpeak= 70.2 \pm 2.3$~\keV,
1269: $\alpha    = 1.32 \pm 0.02$,
1270: $\beta     = 2.44 \pm 0.08$.
1271: \citet{HETE2005} find for the entire burst (\CLerrors): 
1272: $\mathEpeak= 68 \pm 2$~\keV,
1273: $\alpha    = 1.26 \pm 0.02$,
1274: $\beta     = 2.28 \pm 0.06$.
1275: 
1276: The RHESSI parameters, for both peaks, are all significantly higher
1277: (see Table~\ref{tab:Band_pars}).
1278: % This is also the case, when fitting the entire burst. We 
1279: % find (for RHESSI)
1280: % $\mathEpeak= 117 \pm 8\,$keV,
1281: % $\alpha= 1.76 \pm 0.04$, and
1282: % $\beta = 3.7 \pm 1.2$.
1283: However, the high energy photon indices can not be compared directly,
1284: because the break energy (above which $\beta$ is determined, 
1285: see \refeq{eq:Band})
1286: for HETE is 116~\keV, whereas
1287: for RHESSI it is $>400$~\keV, 
1288: i.e.\ above the HETE energy range.
1289: % (see Table~\ref{tab:Band_pars}).
1290: % for RHESSI is $E_{b,\mbox{\em HSI}} =  940\,$keV, whereas for
1291: % HETE $E_{b,HET} = 116\,$keV.
1292: Fitting the RHESSI data (for the entire duration of the burst)
1293: from 135~\keV\ to 500~\keV\ only, where
1294: the RHESSI response is good,
1295: we find $\beta = 2.441 \pm 0.032$, in excellent agreement 
1296: with HETE.
1297: Fitting the RHESSI data from 400~\keV\ to 2000~\keV, i.e.\ above
1298: the HETE range, we find $\beta = 3.11 \pm 0.25$.
1299: The spectrum seems to soften above $\approx 350$~\keV. 
1300: 
1301: The high RHESSI value for $\alpha$ (almost 2.0) for the 
1302: second lightcurve peak indicates
1303: that the spectral peak (in the \EFE\ representation) is broad.
1304: Since RHESSI's sensitivity drops below $\approx 80$~\keV, and this
1305: is a GRB with \Epeak\ in the order of 100~\keV, 
1306: it is likely that RHESSI's $\alpha$ describes rather the broadness
1307: of the peak than the low energy photon index.
1308: This opinion is supported by the fit result of the BPL fit.
1309: The low energy photon index $\alpha \approx 2.0$ shows
1310: that the \EFE\ spectrum is flat from 34~\keV\ to 175~\keV.
1311: 
1312: %For the second peak, as reported in \S \ref{sec:results_otherGRBs}, 
1313: %a BPL with a low energy photon index $\alpha \approx 2.0$ and 
1314: %$E_b=175\,$keV fits as well.
1315: %This means that $\mathEpeak \leq  175\,$keV 
1316: %for the second peak.
1317: 
1318: It should also be mentioned that the $\chi^2$ of the 
1319: HETE fit, as cited by \citet{HETE2005}, is very bad.
1320: Also the RHESSI $\chi^2$ of the spectral fits for
1321: the second peak are rather high.
1322: A joint fit of RHESSI and HETE data might
1323: reveal interesting features.
1324: 
1325: 
1326: \subsection{GRB 030406}
1327: % - - - - - - - - - - - -
1328: 
1329: \paragraph{Fit}
1330: %The spectrum is shown in Fig.~\ref{fig:030406}.
1331: %The lightcurve of GRB 030406 is shown in Fig.~\ref{fig:grbsII},
1332: %top plot, and its count spectrum in Fig.~\ref{fig:030406}.
1333: %Counts are observed up to a few MeV.
1334: %The fit range is 24 - 15000 keV.
1335: The lightcurve is shown in Fig.~\ref{fig:ltc_GRBsII} (top) and 
1336: the spectrum is shown in Fig.~\ref{fig:030406}.
1337: The best fit is the \CB\ function with
1338: the parameters $\alpha$ and $\beta$ frozen at 
1339: the theoretical values.  
1340: %, see tables \ref{tab:chi2} and \ref{tab:CB_pars}.
1341: %\begin{eqnarray}
1342: %  \mathEpeak & = & 656 \pm 57 \mbox{keV}  \nonumber \\ 
1343: %  \alpha   & = & 1.0 \mbox{\hspace{1em} (fix)}  \nonumber \\ 
1344: %  \beta   & = & 2.1 \mbox{\hspace{1em} (fix)} \\ 
1345: %   b   & = &  0.149 \pm 0.091  \nonumber
1346: %\end{eqnarray}
1347: The next best fit is the Band function. 
1348:   %The parameters
1349:   %are listed in Table~\ref{tab:Band_pars}. 
1350: %with:
1351: %\begin{eqnarray}
1352: %  \mathEpeak & = & 674 \pm 70 \mbox{ keV} \nonumber \\ 
1353: %  \alpha   & = & 0.979 \pm 0.064 \\ 
1354: %  \beta    & = & 2.61 \pm 0.27  \nonumber
1355: %\end{eqnarray}
1356: 
1357: \paragraph{discussion}
1358: The spectrum of this burst was also studied by \citet{Radek2006} 
1359: using data from
1360: the INTEGRAL satellite. For the spectral analysis, they
1361: used combined ISGRI and IBIS Compton mode data. 
1362: The time interval used by \citet{Radek2006} differs from ours.
1363: Using a similar time interval as their `peak' time interval, we find:
1364: 
1365: $\bullet$
1366: Fit of a BPL from 24--2400~\keV\ (comparable to the energy range
1367: in the analysis by \citet{Radek2006}): 
1368: fits well
1369: ($\chi^2=70.3$ for $n_{DoF}=60$),
1370: and the parameters are:
1371: $E_b= 479 \pm 80$~\keV,
1372: $\alpha    = 1.08 \pm 0.05$,
1373: $\beta     = 1.96 \pm 0.02$. 
1374: 
1375: $\bullet$
1376: Fit of a BPL from 24~\keV\ to 16~\MeV: does not fit well
1377:  ($\chi^2 = 86.0$ for $n_{DoF}=71$).
1378: 
1379: $\bullet$
1380: Fit of \CB\ from 24~\keV\ to 16~\MeV\ with $\alpha=1.0$ and 
1381: $\beta=2.1$ frozen: fits well
1382: ($\chi^2 = 77.9$ for $n_{DoF}=72$), and the parameters are:
1383: $T= 1220 \pm 110$~\keV\ and $b = 0.055 \pm 0.079$,
1384: 
1385: $\bullet$
1386: Fit of Band function: fits well ($\chi^2 = 77.33$ for $n_{DoF}=71$), 
1387: and the parameters are: 
1388: $\mathEpeak= 1180 \pm 120$~\keV,
1389: $\alpha    = 0.96 \pm 0.06$,
1390: $\beta     = 3.02 \pm 0.55$. 
1391: 
1392: For the high energy part, 
1393: the parameters found by \citet{Radek2006}  
1394: and by us agree. 
1395: But we can not confirm their hard low energy photon index $\alpha < 0.0$.
1396: We even dare to say that we trust our low energy photon index better,
1397: because for this GRB incoming direction, the RHESSI response function
1398: is well understood, whereas the INTEGRAL response function of this burst
1399: might suffer from same systematic effects that we described for RHESSI in \S \ref{sec:systematics}.
1400: 
1401: 
1402: 
1403: 
1404: \subsection{GRB 030519B}
1405: % - - - - - - - - - - - -
1406: \paragraph{Fit}
1407: The lightcurve is shown in 
1408: Fig.~\ref{fig:ltc_GRBsII}
1409: and the spectrum in Fig.~\ref{fig:030519B}.
1410: %
1411: %\paragraph{Fit}
1412: The best fit is the Band function,
1413: % with:
1414: %\begin{eqnarray}
1415: %  \mathEpeak & = & 417 \pm 13 \mbox{ keV} \nonumber \\ 
1416: %  \alpha   & = & 1.048 \pm 0.042 \\ 
1417: %  \beta    & = & 3.11 \pm 0.16  \nonumber
1418: %\end{eqnarray}
1419: followed by \CB.
1420:   %,
1421:   %see tables \ref{tab:chi2}, \ref{tab:Band_pars} and \ref{tab:CB_pars}.
1422: % with:
1423: %\begin{eqnarray}
1424: %  \mathEpeak & = & 407 \pm 12 \mbox{keV}  \nonumber \\ 
1425: %  \alpha   & = &  1.006 \pm 0.056    \nonumber \\ 
1426: %  \beta   & = & 2.35 \pm 0.12 \\ 
1427: %   b   & = &  0.113 \pm 0.042   \nonumber    
1428: %\end{eqnarray}
1429: 
1430: \paragraph{Discussion}
1431: In the HETE GRB catalog by \citet{HETE2005} one finds
1432: (\CLerrors):
1433: $E_{p}= 138\; ^{+18}_{-15} $~\keV,
1434: $\alpha= 0.8 \pm 0.1$,
1435: $\beta = 1.7 \pm 0.2$.
1436: Since $\beta < 2.0$, the energy $E_p$ is not the 
1437: peak energy, but only a variable related to the parameter
1438: $E_0=E_p/(2-\alpha)$.
1439: Indeed, the RHESSI peak energy for this GRB is $>400$~\keV.
1440: But the HETE parameters do not fit the RHESSI
1441: spectrum from 70~\keV\ to 350~\keV\ ($\chi^2 = 141$ for 
1442: 53 energy bins). 
1443: 
1444: 
1445: %\newpage
1446: 
1447: \subsection{GRB 031027}
1448: % - - - - - - - - - - - -
1449: %\paragraph{Fit}
1450: %The lightcurve of GRB 031027 is shown in Fig.~\ref{fig:ltc_GRBsII}
1451: %and its count spectrum in Fig.~\ref{fig:031027}.
1452: %We observe significant counts up to about one MeV.
1453: %Therefore, we do not fit up to 15 MeV, but
1454: %from 60 keV to 6000 keV.
1455: The lightcurve  is shown in 
1456: Fig.~\ref{fig:ltc_GRBsII} and 
1457: the spectrum in Fig.~\ref{fig:031027}.
1458: The best fit is a cut off power law with % (\CLerrors)
1459: $ \mathEpeak = 336 \pm 9$~\keV\ and $\alpha = 0.940\pm 0.05$.
1460: %\begin{eqnarray}
1461: %  \mathEpeak & = & 336 ^{+15}_{-14} \mbox{ keV} \nonumber \\ 
1462: %  \alpha     & = & 0.940 ^{+0.078}_{-0.079} \;\;. 
1463: %\end{eqnarray}
1464: This function is shown in Fig.\  \ref{fig:031027}.
1465: Band function and \CB\ have difficulties to converge.
1466: Since the CPL fits so well, we expect $\beta = \infty$ 
1467: for the Band function and $b=0$ for the \CB\ function.
1468: 
1469: 
1470: \subsection{GRB 031111}
1471: % - - - - - - - - - - - -
1472: \paragraph{Fit}
1473: The lightcurve is shown in Fig.~\ref{fig:ltc_GRBsII} (bottom) and 
1474: the spectrum is shown in Fig.~\ref{fig:031111}.
1475: Band function, \CB\ and BBmPL are good fits, 
1476: the best being Band function.
1477: 
1478: \paragraph{Discussion}
1479: A preliminary CPL fit to the HETE data is
1480: published on a web page\footnote{
1481: \url{http://space.mit.edu/HETE/Bursts/GRB031111A/}
1482: }
1483: as $E_{0}= 600.5$~\keV\ and 
1484: $\alpha= 0.8366$ with a good $\chi^2$.
1485: These values describe the RHESSI spectrum
1486: well from 80~\keV\ to 350~\keV, but not at
1487: higher energies.
1488: HETE's energy range ends at 400~\keV, 
1489: thus we believe that our values for $E_0$ and $\alpha$ are better. 
1490: 
1491: 
1492: %------------------------------
1493: \section{GENERAL DISCUSSION AND CONCLUSION}
1494: %------------------------------
1495: \label{sec:discussion}
1496: 
1497: \subsection{The spectral functions}
1498: %- - - - - - - - - - - - - - - - - -
1499: 
1500: What is an acceptable $\chi^2$? 
1501: In the limit of many degrees of freedom 
1502: ($n_{DoF} > 30$), $\chi^2$ is normal distributed
1503: with an expectation value of $n_{DoF}-0.5$ and
1504: a variance of $\sigma_{\chi^2} = \sqrt{2n_{DoF}-1}$.
1505: A fit is acceptable if $\chi^2$ is close to its expectation
1506: value {\em and} if the residuals
1507: scatter around zero over the whole fit range, 
1508: i.e.\ if the fit ``looks good''.
1509: 
1510: From Table~\ref{tab:chi2} we conclude that
1511: the \CB\ gives acceptable $\chi^2$ for {\em all} GRBs studied.
1512: And they also look good, as can be seen in 
1513: Figs.~\ref{fig:020715} to \ref{fig:031111}.
1514: Except for GRB 021206, rear (Fig.~\ref{fig:spec_rear}), 
1515: the same can be said for the Band function.
1516: 
1517: In many cases, a cut off power law (CPL) 
1518: fits the spectrum up to
1519: high energies, e.g.\ GRB 021008, GRB 030329 or GRB 031027.
1520: In these cases, Band and \CB\ 
1521: improve the goodness-of-fit slightly, but all three spectral
1522: shapes fit the data acceptably.
1523: 
1524: A broken power law fits sometimes, but usually not well. 
1525: 
1526: BBPL and BBmPL do not fit in general, BBPL worse than BBmPL.
1527: However, it should be mentioned that a blackbody component
1528: is expected---if at all---only
1529: at the beginning of a GRB (see e.g.\ \citet{Ryde2006}
1530: and references therein),
1531: whereas we fitted the entire duration of the bursts.
1532: When using BBPL, we often find that the PL component
1533: fits either at high energies or at low energies.
1534: This is also discussed by \citet{Ghirlanda2007}
1535: who studied six BATSE GRBs in detail,
1536: where low energy data from the WFC instrument (on board BeppoSAX) are
1537: available. They find that the WFC data fit
1538: the Band function or CPL extrapolation, but not the
1539: BBPL extrapolation to low energies. 
1540: Arguing that the PL contribution is too simple, they
1541: try to fit a blackbody spectrum plus CPL.  %cut off power law shape.
1542: %They find the blackbody component to be insignificant,
1543: %which is not surprising, since the CPL alone already fitted
1544: %the data well. 
1545: We suggest to use our BBmPL function instead.
1546: Its modified PL component describes a spectral break from
1547: $dN/dE \propto E^{-(\beta-1)}$ at low energies to
1548: $dN/dE \propto E^{-\beta}$ at high energies.
1549: 
1550: 
1551: %\subsection{\CB\ function versus CPL and Band function}
1552: \subsection{\CB\ function }
1553: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
1554:  
1555: The present work is, to our knowledge, the first systematic 
1556: attempt to fit the \CB\ function to prompt GRB spectra.
1557: The two terms in \refeq{eq:CB}
1558: have a simple meaning.
1559: According to the cannonball theory, all GRBs are associated with a supernova.
1560: The ambient light is Compton up-scattered by the cannonball's electrons,
1561: producing the prompt GRB emission.
1562: Some electrons are simply comoving with the cannonball, 
1563: giving rise to the CPL term in \refeq{eq:CB}.
1564: Since the photon spectrum of the ambient light can be described
1565: by a thin thermal bremsstrahlung spectrum,
1566: $\alpha$ is expected to be $\approx 1$.
1567: The second term (mPL) is caused by
1568: a small fraction of electrons accelerated 
1569: to a power law distribution, resulting in a 
1570: photon index of $\beta \approx 2.1$.
1571: %The first term thus describes the
1572: %initial spectrum of the ambient light,
1573: %boosted by the high Lorentz factor
1574: %of the cannonball's comoving electrons.
1575: See e.g.\ \citet{Dado2005}, \S 3.8. 
1576: or \citet{Dado2007}, \S 2 and 4.1 for a summary. 
1577: 
1578: In our study, the observed values for $\alpha$ are all approximately 1,
1579: as predicted by the \CB.
1580: Because of the low count statistics at high energies, 
1581: we could not always fit $\beta$.
1582: We then fixed it to its theoretical value
1583: of 2.1 in order to make the fit converge and to obtain
1584: a value for the parameter $b$.
1585: In the cases where we could fit $\beta$,
1586: we found values close to 2.1 (Table~\ref{tab:CB_pars}). 
1587: 
1588: For the factor $b$ of the modified PL component in the \CB\ function
1589: we typically found values of the order $0.1\,$.
1590: An exception is GRB 031111, where $b$ is of
1591: the order 1.0, but with a large error ($0.4$).
1592: 
1593: %Our values for $\beta$ and $b$ are similar to the 
1594: %ones found by \citet{Dado2004} 
1595: %(fit  of $X$-ray flashes XRF 971019, XRF 980128,
1596: %and XRF 990520 using BeppoSAX/WFC and 
1597: %CGRO/BATSE data) and  \citet{Dado2005} (fit of GRB 941017).
1598: Our values for $\beta$ and $b$ are similar to the 
1599: ones found by 
1600: \citet{Dado2005} (fit of GRB 941017)
1601: and by
1602: \citet{Dado2004} (fit  of $X$-ray flashes XRF 971019, XRF 980128,
1603: and XRF 990520 using BeppoSAX/WFC and CGRO/BATSE data).
1604: The authors of the \CB\ hypothesize that XRFs are simply GRBs viewed
1605: further off the jet axis.
1606: 
1607: The different time development of the CPL- and the mPL-fluences,
1608: as reported in Table~\ref{tab:dt_grb021206}, 
1609: possibly point to a different time dependence of 
1610: the two underlying electron distributions within a cannonball. 
1611: 
1612:  
1613: \subsection{Fitting \CB\ function versus CPL and Band function}
1614: %- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
1615: Both the Band function and the \CB\ function are extensions
1616: of a CPL, the Band function with one additional parameter,
1617: the \CB\ function with two. For cases where a CPL fits the
1618: data well, also a Band function with $\beta = \infty$ or
1619: a \CB\ function with $b=0$ (and $\beta=2.1$ or any other value) 
1620: fits. This is the case for GRB 031027.
1621: 
1622: Whether additional parameters are necessary in a fit,
1623: can be tested with the $F$-test.
1624: For GRB 030329, the extra parameters are barely needed.
1625: For GRB 020715, 021008, 021206, 030406, 030519B, 031111
1626: additional parameters are required at a confidence level
1627: of at least 90\%.
1628: 
1629: Concerning the question of whether the high energy power law
1630: parameter $\beta$ in the \CB\ should be treated as free parameter,
1631: the answer is 'yes' from a theoretical point of view, but
1632: in practice, see Table~\ref{tab:chi2}, the improvements in $\chi^2$ 
1633: are marginal or small for all bursts we studied.
1634: Our practice to freeze $\beta$ at its theoretically predicted
1635: value in cases of bad convergence
1636: seems to be acceptable.
1637:  
1638: It is more difficult to compare the goodness of fit using
1639: the Band function compared to using the \CB\ function.
1640: The two functions are not independent, because they both
1641: are dominated by a CPL up to the peak energy and higher.
1642: In most cases of our study, the two functions fit the observed
1643: spectrum equally well with a slight preference
1644: for the Band function. 
1645: At high energies however (typically above several
1646: times the peak energy) the two functions are different, the spectral
1647: hardening being a unique feature of the \CB\ function.
1648: There is only one case, namely GRB 021206, where this
1649: hardening is observed. For the rear data going up to high energies, 
1650: a Band function fit gives $\chi^2 / \mbox{dof}
1651: = 133.3 / 74$ (see Table~\ref{tab:chi2}). This is not 
1652: acceptable at $<0.01$\% probability of being accidentally so high.
1653: The \CB\ fit on the other hand gives  $\chi^2 / \mbox{dof}
1654: = 82.7 / 73$, which is fully acceptable at a $20$\% level.
1655: 
1656: We would like to stress again that, while the \CB\ gives acceptable
1657: fits for {\em all} cases, the Band function fails in one case.
1658: This seems enough to us to give some credit to the \CB.
1659: 
1660: But it is, of course, no proof that the \CB\ is the only theory capable
1661: of describing the spectrum of GRB 021206. 
1662: For example, a Band function plus a PL with $\gamma\approx 1.5$ 
1663: would also fit.
1664: But there is no theory to predict such a shape.
1665: To our knowledge, \CB\ is to date the only existing GRB model % astronomical theory 
1666: that explains the prompt GRB spectra from first principles. 
1667: 
1668: At this place we also would like to note the the mean 
1669: $\alpha$-value found for the BATSE catalogue is 1 
1670: \citep[see][]{BATSEcatalog}). We cite from their summary:
1671: %''
1672: ``{\em We confirmed, using a much larger sample, that
1673: the most common value for the low-energy index is $\approx -1\,$}\footnote{ 
1674: this corresponds to $+1$ in our notation}
1675: {\em \citep{Preece2000, Ghirlanda02}. 
1676:  The overall distribution of this parameter 
1677: shows no clustering or distinct features at the values expected from
1678: various emission models, such as $-2/3$ for synchrotron 
1679: \citep{Katz94, Tavani96},
1680: $0$ for jitter radiation \citep{Medv00}, 
1681: or $-3/2$ for cooling synchrotron \citep{GhisCel99}.}''
1682: They do not mention the \CB\ which would explain $\alpha \approx 1$.
1683: 
1684: Note that the $\beta$ values of the \CB\ are systematically lower 
1685: than the $\beta$ values of the Band function,
1686: compare Tables \ref{tab:CB_pars} and \ref{tab:Band_pars}.
1687: From Band function fits to BATSE GRBs, it is known that
1688: $\beta$ is clustered around 2.3, with a long tail
1689: towards higher values,   see \citet{BATSEcatalog}.
1690: For \CB\ we would expect $\beta$ to cluster at
1691: slightly lower values.
1692: 
1693: For criticism of the \CB, see e.g.\ \citet{Hillas}, but see
1694: also the answer by \citet{Dar2006}.
1695: 
1696: 
1697: \subsection{The spectral hardening}
1698: %- - - - - - - - - - - - - - - - - 
1699: 
1700: The difference of a \CB\ spectrum and
1701: Band function is the hardening at high energies. 
1702: This becomes visible---for the GRBs studied here---in 
1703: the few \MeV\ region, but it depends on the peak energy
1704: and the factor $b$. For $\alpha=1.0$ and $b=0.10$ the hardening
1705: typically appears at several times the peak energy
1706: and the second term dominates at 10 times the peak energy.
1707: For the spectral fit of XRFs done by \citet{Dado2004}, 
1708: %using BeppoSAX/WFC and CGRO/BATSE data,
1709: the spectral hardening is expected in the few hundred \keV\ region,
1710: just where the number of photons detected runs low.
1711: Most of our GRBs also suffer from this lack of statistics at high energies,
1712: preventing the detection of a hardening.
1713: 
1714: A spectral coverage of two decades and 
1715: good detection efficiency at high energies
1716: is necessary to experimentally observe the 
1717: full shape of the \CB\ function. 
1718: In the case of GRB 021206 we were able to detect this hardening, 
1719: thanks to RHESSI's broad energy range (30~\keV\ to 15~\MeV),
1720: and because this is one of the brightest GRBs ever observed.
1721: 
1722: There is a GRB observed by SMM from 20 keV up to 100 MeV,
1723: namely GRB 840805.
1724: As reported by \citet{Share84},
1725: the spectrum of this burst shows emission up 100 MeV. 
1726: In order to fit the spectrum, ``a classical thermal synchrotron function
1727: plus a power law'' was used. The power law component was 
1728: required to fit the data above about 6 MeV. This is a hint
1729: of a spectral hardening around 6 MeV, and we suppose that
1730: the spectrum of this GRB can be fit by a \CB\ function.
1731:  
1732: The spectral hardening observed in GRB 941017
1733: \citep{Nature2003} seems to be different. 
1734: The photon index of GRB 941017 above a few \MeV\ is $\approx 1.0$.
1735: This case is discussed by \citet{Dado2005} as a possible 
1736: additional feature in the \CB\ spectrum.
1737: 
1738:  
1739: \subsection{Outlook}
1740: %------------------------------
1741: \label{sec:outlook}
1742: 
1743: In order to find more GRB spectra that show the hardening
1744: characteristic for the \CB\ function,
1745: strong GRBs have to be observed over a broad enough energy range.
1746: With the forthcoming GLAST mission, we expect 
1747: that more such spectra will be observed.
1748: But also joint analyses with more than one instrument
1749: %GRBs or XRFs 
1750: %from Swift/RHESSI and Swift/Konus
1751: could reveal this hardening.
1752: We therefore suggest:
1753: 
1754: %$\bullet$ 
1755: %%Apart from RHESSI, Konus has an as broad energy range,
1756: %%and a huge amount of GRB data.
1757: %To search for \CB\ spectrum candidates among the Konus GRBs.
1758: %% Are there \CB\ candidates in the Konus GRB data?
1759: 
1760: $\bullet$ 
1761: %Apart from RHESSI, Konus has an as broad energy range,
1762: %and a huge amount of GRB data.
1763: to search for \CB\ spectrum candidates among joint Swift/RHESSI
1764: GRBs and XRFs, and joint Swift/Konus GRBs.
1765: 
1766: $\bullet$ 
1767: to reanalyse some BATSE bursts. Looking
1768: at the BATSE spectra published by \citet{Ghirlanda2007}, 
1769: we suppose that the \CB\ can possibly improve the fits of 
1770: GRB 980329, GRB 990123, and GRB 990510. 
1771: The same can be said for  
1772: GRB 911031 as published by \citet{Ryde2006}. 
1773: And GRB 000429, as published in Fig.\ 19 of \citet{BATSEcatalog},
1774: looks like a promising candidate as well.
1775: 
1776: $\bullet$ 
1777: to search in KONUS data for suitable GRBs.
1778: 
1779: $\bullet$ 
1780: %We use our own IDL fitting routines.
1781: to add the \CB\ function to XSPEC
1782: in order to make it more accessible
1783: to the astronomical community.
1784: %. It would help to make this function used more often.
1785: 
1786: 
1787: \section{SUMMARY}
1788: %------------------
1789: \label{sec:summary}
1790: 
1791: We have presented the time integrated spectra of 8 bright
1792: GRBs observed by RHESSI in the years 2002 and 2003.
1793: 
1794: The spectrum of GRB 021206 shows a hardening above 4~\MeV. 
1795: From 70~\keV\ to 4.5~\MeV, the spectrum can be well
1796: fitted by a Band function -- but not above that.
1797: The cannonball model successfully describes
1798: the entire spectrum up to 16~\MeV, the upper limit
1799: of RHESSI's energy range.
1800: For the spectra of the seven other GRBs analysed, we found that they
1801: can be fitted by the \CB\ as well as by the Band function.
1802: 
1803: We therefore suggest that the cannonball model should
1804: be considered for fitting GRB spectra.
1805: 
1806: 
1807: 
1808: %% If you wish to include an acknowledgments section in your paper,
1809: %% separate it off from the body of the text using the \acknowledgments
1810: %% command.
1811: 
1812: %% Included in this acknowledgments section are examples of the
1813: %% AASTeX hypertext markup commands. Use \url without the optional [HREF]
1814: %% argument when you want to print the url directly in the text. Otherwise,
1815: %% use either \url or \anchor, with the HREF as the first argument and the
1816: %% text to be printed in the second.
1817: 
1818: 
1819: \acknowledgments
1820: %-----------------
1821: 
1822: We thank K. Hurley, A. Kann, S. McGlynn, J. \v{R}ipa and
1823: L.\ Hanlon for helpful discussion and comments.
1824: 
1825: %O.\ Wigger  
1826: %and E.\ Kirk  for many
1827: %helpful discussions and encouragement.
1828: 
1829: 
1830: %% The reference list follows the main body and any appendices.
1831: %% Use LaTeX's thebibliography environment to mark up your reference list.
1832: %% Note \begin{thebibliography} is followed by an empty set of
1833: %% curly braces.  If you forget this, LaTeX will generate the error
1834: %% "Perhaps a missing \item?".
1835: %%
1836: %% thebibliography produces citations in the text using \bibitem-\cite
1837: %% cross-referencing. Each reference is preceded by a
1838: %% \bibitem command that defines in curly braces the KEY that corresponds
1839: %% to the KEY in the \cite commands (see the first section above).
1840: %% Make sure that you provide a unique KEY for every \bibitem or else the
1841: %% paper will not LaTeX. The square brackets should contain
1842: %% the citation text that LaTeX will insert in
1843: %% place of the \cite commands.
1844: 
1845: %% We have used macros to produce journal name abbreviations.
1846: %% AASTeX provides a number of these for the more frequently-cited journals.
1847: %% See the Author Guide for a list of them.
1848: 
1849: %% Note that the style of the \bibitem labels (in []) is slightly
1850: %% different from previous examples.  The natbib system solves a host
1851: %% of citation expression problems, but it is necessary to clearly
1852: %% delimit the year from the author name used in the citation.
1853: %% See the natbib documentation for more details and options.
1854: 
1855: \begin{thebibliography}{}
1856: 
1857: \bibitem[Band et al.(1993)]{Band93} 
1858:      Band, D.\ et al. 1993,
1859:      \apj\ 413, p. 281.
1860: \bibitem[Barlow \& Beeston(1993)]{Barlow93} 
1861:      Barlow, R.\ \& Beeston, C. 1993, 
1862: %     Barlow, Roger \& Beeston, Christine 1993, 
1863: % " Fitting using finite Monte Carlo Samples. "
1864:      Comp. Phys. Comm. 77, p.\ 219 
1865: \bibitem[Boggs et al.(2004)]{Boggs2004}	
1866:      Boggs, S.E., Wunderer, C.B., Hurley, K., Coburn, W. 2004,
1867:      \apj\ Letter 611, L77-L80
1868: 
1869: \bibitem[Chornock et al.(2003)]{GCN2131} R.\ Chornock al. 2003,
1870:      GRB Circular Network 2131
1871: 
1872: \bibitem[CERN(1993)]{Geant3}
1873:      CERN program Library Office 1993,
1874:      GEANT -- Detector Description and Simulation Tool
1875:      (CERN Geneva, Switzerland)
1876: \bibitem[Coburn \& Boggs(2003)]{CB03}
1877:      Coburn, W., \& Boggs, S.E. 2003,
1878:      Nature 423, p.\ 415
1879: %\bibitem{ref:CB03} \BY{Coburn~W.\ \atque Boggs~S.E.}
1880: %  \IN{Nature}{423}{2003}{415}.
1881: \bibitem[Dado \& Dar(2005)]{Dado2005}
1882:      Dado, S. \& Dar, A. 2005,
1883:      \apj\ Letter 627, L109-L112
1884: \bibitem[Dado, Dar \& de R\'ujula(2002)]{Dado2002}
1885:      Dado, S., Dar, A.\ \& de R\'ujula, A. 2002,
1886:      A\&A 388, p.\ 1079 
1887:      %  "On the optical and X-ray afterglows of gamma ray bursts"
1888: \bibitem[Dado, Dar \& de R\'ujula(2003a)]{Dado2003}
1889:      Dado, S., Dar, A.\ \& de R\'ujula, A. 2003a,
1890:      A\&A 401, p.\ 243 
1891:      %  "On the radio afterglow of gamma ray bursts"
1892: \bibitem[Dado, Dar \& de R\'ujula(2003b)]{Dado_030329}
1893:      Dado, S., Dar, A.\ \& de R\'ujula, A. 2003b,
1894:      \apj\ Letter 594, L89-L92 
1895:      %  "The Supernova associated with GRB 030329"
1896: \bibitem[Dado, Dar \& de R\'ujula(2004)]{Dado2004}
1897:      Dado, S., Dar, A.\ \& de R\'ujula, A. 2004,
1898:      A\&A 422, p.\ 381
1899:      %  "On the origin of X-ray flashes"
1900: \bibitem[Dado, Dar \& de R\'ujula(2007)]{Dado2007}
1901:      Dado, S., Dar, A.\ \& de R\'ujula, A. 2007,
1902:      arXiv:0706.0880v1 [astro-ph]
1903:      %  "On the X-ray emission of Gamma Ray Bursts"
1904: \bibitem[Dar \& de R\'ujula(2004)]{Dar2004}
1905:      Dar, A. \& de R\'ujula, A. 2004,
1906:      Physics Reports 405, 203
1907:      %  "Towards a complete theory of Gamma Ray Bursts"
1908: \bibitem[Dar \& de R\'ujula(2006)]{Dar2006}
1909:      Dar, A.\ \& de R\'ujula, A. 2006,
1910:      arXiv:hep-ph/0611369v1
1911: %  " The vicissitudes of "cannonballs": a response to criticisms 
1912: %     by A.M. Hillas and a brief review of our claims "
1913: \bibitem[Ferrero et al.(2006)]{Ferrero2006}	
1914: %  "The GRB 060218/SN 2006aj event in the context of other 
1915: % gamma-ray burst supernovae     
1916:     P.\ Ferrero et al. 2006,
1917:     A\&A 457, pp.\ 857-864 
1918: \bibitem[Ghirlanda et al.(2007)]{Ghirlanda2007}
1919:      Ghirlanda, G.\ et al. 2007,
1920:      MNRAS 379, pp.\ 73-85
1921: %     "Blackbody components in gamma-ray bursts spectra?"
1922: \bibitem[Ghirlanda et al.(2003)]{Ghirlanda2003}
1923:      Ghirlanda, G.\ et al. 2003,
1924:      A\&A 406, pp.\ 879-892
1925: \bibitem[Ghirlanda et al.(2002)]{Ghirlanda02}
1926:      Ghirlanda, G.\ et al. 2002,
1927:      A\&A 393, p.\ 409
1928: \bibitem[Ghisellini \& Celotti(1999)]{GhisCel99}
1929:      Ghisellini, G.\ \& Celotti, A. 1999,
1930:      \apj\ Supplements 138, 149
1931: \bibitem[Gonz\'alez et al.(2003)]{Nature2003}
1932:      Gonz\'alez, M.M.\ et al. 2003,
1933:      Nature 424, 749
1934: \bibitem[Hillas(2006)]{Hillas}
1935:      Hillas, A.M. 2006,
1936:      arXiv:astro-ph/0607109v2 
1937: %   " Cosmic Rays: Recent Progress and some Current Questions"
1938: \bibitem[Hjorth et al.(2003)]{Hjorth2003} Hjorth, J.\ et al. 2003,
1939:      Nature 423, p.\ 847
1940: \bibitem[Hurley et al.(2002a)]{GCN_020715_IPN} Hurley, K.\ et al. 2002a,
1941:      GRB Circular Network 1454, 1456
1942: \bibitem[Hurley et al.(2002b)]{GCN_021008_IPN} Hurley, K.\ et al. 2002b,
1943:      GRB Circular Network 1629, 1617
1944: \bibitem[Hurley et al.(2002c)]{GCN_021206_IPN} Hurley, K.\ et al. 2002c,
1945:      GRB Circular Network 1727, 1728
1946: %\bibitem[GCN 1728]{GCN1728} Hurley, K.\ et al. 2002,
1947: %     GRB Circular Network 1728
1948: \bibitem[Hurley et al.(2003a)]{GCN_030406_IPN} Hurley, K.\ et al. 2003a,
1949:      GRB Circular Network 2127
1950: \bibitem[Hurley et al.(2003b)]{GCN_021206_final} Hurley, K.\ et al. 2003b,
1951:      GRB Circular Network 2281
1952: \bibitem[Hurley et al.(2003c)]{GCN_030519B_IPN} Hurley, K.\ et al. 2003c,
1953:      GRB Circular Network 2237
1954: \bibitem[Hurley et al.(2003d)]{GCN_031027_IPN} Hurley, K.\ et al. 2003d,
1955:      GRB Circular Network 2438
1956: \bibitem[Hurley et al.(2003e)]{GCN_031111_IPN} Hurley, K.\ et al. 2003e,
1957:      GRB Circular Network 2443
1958: 
1959: 	
1960: \bibitem[Kaneko et al.(2006)]{BATSEcatalog} 
1961:      Kaneko, Y.\ et al.\, 2006, 
1962:      \apj\ Supplements 166, pp.\ 298-340
1963: \bibitem[Katz et al.(1994)]{Katz94} 
1964:      Katz, J.I. 1994,
1965:      \apj\ Letter 432, L107
1966: \bibitem[Lin et al.(2002)]{RHESSI}
1967:      Lin, R.P.\ et al. 2002,
1968:      Solar Physics 210, p.\ 3
1969: 
1970: \bibitem[McBreen et al.(2006)]{SMcBreen2006}
1971:      McBreen, S.\ et al. 2006, 
1972:      A\&A 455, pp.\ 433-440
1973: \bibitem[Marcinkowski et al.(2006)]{Radek2006}
1974:      Marcinkowski, R.\ et al. 2006, 
1975:      A\&A 452, pp.\ 113-117
1976: 
1977: %\bibitem[Matheson et al.(2003a)]{GCN_sn_hint} T.\ Matheson\ et al. 2003,
1978: %     GRB Circular Network 2107
1979: \bibitem[Matheson et al.(2003)]{GCN_sn_obs} T.\ Matheson\ et al. 2003,
1980:      GRB Circular Network 2120
1981: \bibitem[Medvedev(2000)]{Medv00}
1982:      Medvedev, M.V. 2000,
1983:      \apj\ 540, 704
1984:      
1985: \bibitem[Lamb et al.(2003)]{GCN_030519B_HETE}
1986:      Lamb, D.\ et al. 2003,
1987:      GRB Circular Network 2235
1988: \bibitem[Preece et al.(2000)]{Preece2000}
1989:      Preece, R.D.\ et al. 2000, 
1990:      \apj\ Supplements 126, pp.\ 19-36
1991: \bibitem[Preece et al.(2002)]{Preece2002}
1992:      Preece, R.D.\ et al. 2002, 
1993:      \apj\ 581, pp.\ 1248-1255
1994: %    "On the Consistency of Gamma-Ray Burst Spectral Indices 
1995: %     with the Synchrotron Shock Model"
1996: \bibitem[Rutledge \& Fox(2004)]{RF04} 
1997:      Rutledge, R.E. and Fox, D.B. 2004,
1998:      MNRAS 350, p.\ 1288
1999: \bibitem[Ryde et al.(2006)]{Ryde2006}
2000:      Ryde, F.\ et al. 2006, 
2001:      \apj\ 652, pp.\ 1400-1415
2002: \bibitem[Ryde (2004)]{Ryde2004}
2003:      Ryde, Felix 2004, 
2004:      \apj\ 614, pp.\ 827-846.
2005: 
2006: \bibitem[Sakamoto et al.(2005)]{HETE2005}
2007:      Sakamoto, T.\ et al. 2005,
2008:      \apj\ 629, pp.\ 311-327
2009: \bibitem[Share et al.(1986)]{Share84}
2010:      Share, G.H.\ et al\ 1986,
2011:      Adv.\ Space Res.\  V.\ 6, N.\ 4, pp.\ 15-18
2012: \bibitem[Smith et al.(2002)]{spectrometer}
2013:      Smith,D.M., et al. 2002, 
2014:      Solar Physics 210, p.\ 33
2015: \bibitem[Tavani (1996)]{Tavani96}
2016:      Tavani, M. 1996, 
2017:      \apj\ 466, 768 
2018: 
2019: \bibitem[Wigger et al.(2007)]{cw_venice}
2020:      Wigger, C.\  et al. 2007, 
2021:      Il Nouvo Cimento B, 
2022:      DOI: 10.1393/ncb/i2007-10072-9  
2023: \bibitem[Wigger et al.(2005)]{cw_rome}
2024:      Wigger, C.\  et al. 2005, 
2025:      Il Nuovo Cimento C, 28, p.\ 265 
2026: \bibitem[Wigger et al.(2004)]{PSI2004}
2027:      Wigger, C.\  et al. 2004,
2028:      \apj\  613, p.\ 1088
2029: \bibitem[Vanderspek et al.(2003)]{GCN_030329_HETE}
2030:      Vanderspek, R.\ et al. 2003,
2031:      GRB Circular Network 1997
2032: \bibitem[Vanderspek et al.(2004)]{HETE030329}
2033:      Vanderspek, R.\ et al. 2004, 
2034:      \apj\ 617, pp.\ 1251-1257
2035: 
2036: \bibitem[Zaritsky et al.(2003)]{GCN2169} D.\ Zaritskyet al. 2003,
2037:      GRB Circular Network 2081
2038: \bibitem[Zeh et al.(2003)]{Zeh2003} A.\ Zeh et al. 2003,
2039:      GRB Circular Network 2081
2040: %\bibitem[Zeh,Klose \& Hartmann(2004)]{Zeh2004}	
2041: %     A.\ Zeh, S.\ Klose, D.H.\ Hartmann 2004,
2042: %%  "A Systematic Analysis of Supernova Light in Gamma-Ray Burst Afterglows"
2043: %%    The Astrophysical Journal, Volume 609, Issue 2, pp. 952-961.  
2044: %    \apj\, 609, p952-961
2045: %\bibitem[Zehnder(2002)]{Zeh02}
2046: %     Zehnder, A., et al. 2002,
2047: %     SPIE Proc.\ 4853, Waikoloa, Hawaii
2048: \end{thebibliography}
2049: 
2050: 
2051: %% Use the figure environment and \plotone or \plottwo to include 
2052: %% figures and captions in your electronic submission.
2053: 
2054: 
2055: \clearpage
2056: \begin{deluxetable}{lcccccc}
2057: %\tabletypesize{\scriptsize}
2058: %\tabletypesize{\footnotesize}
2059: %\tabletypesize{\small}
2060: \tablecolumns{7}
2061: \tablecaption{GRB analysis time intervals
2062:          \label{tab:GRBs}}
2063: \tablewidth{0pt}
2064: \tablehead{
2065: \colhead{GRB} & 
2066: \colhead{$t_0$} & 
2067: \colhead{$\Delta t_{burst}$} &
2068: \colhead{$\Delta t_{BG1}$} & 
2069: \colhead{$\Delta t_{BG2}$} & 
2070: \colhead{$\theta$} & 
2071: \colhead{ref.}
2072: \\
2073: \colhead{} &
2074: \colhead{(UT)} & 
2075: \colhead{(s)} &
2076: \colhead{(s)} &
2077: \colhead{(s)} &
2078: \colhead{(degrees)} & 
2079: \colhead{}
2080: }
2081: \startdata
2082: 020715    & 19:20:56.0 & [11.53,15.55]       & [\phn-80.46,0.0]  & [\phn48.28,168.97]    &\phn72.4 & 1 \\
2083: 021008    & 07:00:45.0 & [17.21,21.29]       & [\phn-73.37,0.0]  & [\phn36.68,\phn48.91] &\phn50.1 & 2 \\
2084: 021206    & 22:49:11.7 & [\phn2.73,\phn8.19] & [\phn-53.26,0.0]  & [\phn20.49,102.43]    &\phn18.0 & 3 \\
2085: %021206 P1& 22:49:11.7 & [\phn2.73,\phn4.10] & [\phn-53.26,0.0]  & [\phn20.49,102.43]    &\phn18.0 & 3 \\
2086: %021206 P2&    "       & [\phn4.10,\phn5.46] &       "           &         "             &    "    & " \\
2087: %021206 P3&    "       & [\phn5.46,\phn6.83] &       "           &         "             &    "    & " \\
2088: %021206 P4&    "       & [\phn6.83,\phn8.19] &       "           &         "             &    "    & " \\
2089: 030329 P1 & 11:37:10.0 & [16.56,24.84]       & [\phn-70.39,0.0]  & [\phn70.39,140.78]    &   144.1 & 4 \\  
2090: 030329 P2 &    "       & [28.98,34.50]       &       "           &            "          &     "   & " \\  
2091: 030406    & 22:41:30.0 & [85.68,89.83]       & [-140.96,0.0]     & [140.96,281.93]       &\phn96.1 & 5 \\  
2092: 030519B   & 14:04:53.0 & [\phn0.46,11.47]    & [\phn-61.94,0.0]  & [\phn28.90,\phn90.84] &   165.5 & 6 \\ 
2093: 031027    & 17:07:06.0 & [29.71,45.92]       & [-137.77,0.0]     & [\phn68.88,206.65]    &   101.5 & 7 \\
2094: 031111    & 16:45:12.0 & [\phn2.27,\phn6.35] & [-122.51,0.0]     & [\phn12.25,134.76]    &   155.6 & 8 \\  
2095: % exact 011206 interval times: 2.73133 P1 4.09700 P2 5.46267 P3 6 .82833 P4 8.19400
2096: \enddata                                                                                          
2097: \tablecomments{$t_0$: reference time; 
2098:   $\Delta t_{burst}$: time interval for spectral analysis; 
2099:   $\Delta t_{BG1}$: background time interval before GRB; 
2100:   $\Delta t_{BG2}$: background time interval after GRB;
2101:   time intervals are given relative to $t_0$.
2102:   $\theta$: angle between GRB direction and RHESSI axis;
2103:   References: 
2104:   (1) GCN 1456, 1454 \citep{GCN_020715_IPN},	 
2105:   (2) GCN 1629, 1617 \citep{GCN_021008_IPN},		
2106:   (3) GCN 1728, 1727 \citep{GCN_021206_IPN},	
2107:   (4) GCN 1997 \citep{GCN_030329_HETE},		
2108:   (5) GCN 2127 \citep{GCN_030406_IPN},
2109:   (6) GCN 2235, 2237 \citep{GCN_030519B_HETE, GCN_030519B_IPN},	
2110:   (7) GCN 2438 \citep{GCN_031027_IPN},		
2111:   (8) HETE trigger 2924, GCN 2443 \citep{GCN_031111_IPN}.}
2112: \end{deluxetable}
2113: 
2114: 
2115: \begin{deluxetable}{lcccrrrrrrr}
2116: %\tabletypesize{\scriptsize}
2117: %\tabletypesize{\footnotesize}
2118: \tabletypesize{\small}
2119: %\rotate
2120: \tablecolumns{11}
2121: \tablecaption{Chi-square of spectral fits
2122: 	 \label{tab:chi2}}
2123: \tablewidth{0pt}
2124: \tablehead{
2125: \colhead{GRB} & 
2126: \colhead{$\Delta E_{\mbox{\scriptsize front}}\,$} & 
2127: \colhead{$\Delta E_{\mbox{\scriptsize rear}}\,$} & 
2128: \colhead{$n$} & 
2129: \colhead{CPL} & 
2130: \colhead{Band} &
2131: \colhead{\CB} &
2132: \colhead{\CB} &
2133: \colhead{BPL} &
2134: \colhead{BBPL} &
2135: \colhead{BBmPL}
2136: \\ 
2137: \colhead{} &
2138: \colhead{(\keV)} &
2139: \colhead{(\keV)} &
2140: \colhead{} &
2141: \colhead{$n_{par}=3$} &
2142: \colhead{4} &
2143: \colhead{4} &
2144: \colhead{5} &
2145: \colhead{4} &
2146: \colhead{4} &
2147: \colhead{4}
2148: }
2149: %GRB        front interval  rear interval      n      CPL      BAND      CB         CB           BPL      BBPL    BBmPL
2150: \startdata
2151: 020715     & \nodata       &[\phn30,15660]   & 117 &  113.9  & 106.3   & 110.8   & 110.7   &  129.2  &  270.0  & 157.1   \\ \hline
2152: 021008     &[300,2800]     &\nodata          &  38 &   35.5  &  34.8   &  35.1   &  35.0   &   33.0  &   32.9  &  33.9   \\ 
2153: 021008     &\nodata        &[300,15660]      &  50 &   43.8  &  39.2   &  39.9   &  39.2   &   52.9  &   97.4  &  60.2   \\
2154: 021008     &[300,2800]     &[300,15660]      &  88 & \nodata &  79.5   &  80.8   & \nodata & \nodata & \nodata & \nodata \\ \hline 
2155: 021206     &[\phn70,2800]  &\nodata          & 112 &  130.6  & 103.6   & 104.1   & 103.9   &  155.5  &  315.7  & 191.8   \\  
2156: 021206     &\nodata        &[300,16000]      &  78 &  338.1  & 133.3   &  82.7   &  82.7   &  132.9  &   94.1  & 110.5   \\
2157: 021206     &[\phn70,2800]  &[300,16000]      & 190 & \nodata & \nodata & \nodata & 187.5   & \nodata & \nodata & \nodata \\
2158: 021206     &\nodata        &[300,\phn4500]   &  66 & \nodata &  72.8   &  69.8   & \nodata & \nodata & \nodata & \nodata \\
2159: 021206     &[\phn70,2800]  &[300,\phn4500]   & 178 & \nodata & 176.5   & \nodata & \nodata & \nodata & \nodata & \nodata \\ \hline
2160: %021206 P1 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?         & ?          & ?	& ?	& ?	\\
2161: %021206 P2 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?         & ?          & ?	& ?	& ?	\\
2162: %021206 P3 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?         & ?          & ?	& ?	& ?	\\
2163: %021206 P4 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?         & ?          & ?	& ?	& ?	\\ \hline 
2164: 030329 P1  &\nodata        &[\phn34,10000]   &  94 &   86.5  &  84.3   &  84.8   &  84.7    &  89.3 & 137.8 &  89.1 \\ 
2165: 030329 P2  &\nodata        &[\phn34,\phn7000]&  87 &  104.5  & 103.3   & 103.2   & 103.1    &  98.9 & 102.0 &  98.4 \\ \hline
2166: 030406     &\nodata        &[\phn24,15000]   &  75 &   79.1  &  75.6   &  75.0   &  75.0    &  88.6 & 151.2 &  88.7 \\ \hline
2167: 030519B    &\nodata        &[\phn70,15000]   &  79 &  102.2  &  86.3   &  91.1   &  89.0    &  99.8 & 189.8 & 109.8 \\ \hline 
2168: 031027     &\nodata        &[\phn60,\phn6000]&  63 &   63.6  &\phn n.c.&\phn n.c.& \phn n.c.&  93.0 & 138.9 &  99.6 \\ \hline 
2169: 031111     &\nodata        &[\phn38,15000]   & 117 &  182.4  & 128.3   & 133.2   & 130.4    & 140.5 & 266.0 & 133.0
2170: \enddata     
2171: \tablecomments{$\chi^2$ obtained by fitting different spectral models to the data.
2172:  $\Delta E_{\mbox{\scriptsize front/rear}}$: energy interval used to fit front/rear detector data;
2173:  $n$: number of energy bins;
2174:  $n_{par}$: number of free fit parameters;
2175:  CPL: cut off power law \refeq{eq:CPL},
2176:  Band: Band function \refeq{eq:Band},
2177:  \CB: cannonball model \refeq{eq:CB},
2178:  BPL: broken power law \refeq{eq:BPL},
2179:  BBPL: blackbody plus power law \refeq{eq:BBPL},
2180:  BBmPL: blackbody plus modified power law \refeq{eq:BBmPL};
2181:  n.c.: fit did not converge. In the case of \CB\ with 4 parameters, $\beta$ was fixed to its theoretically
2182:  expected value of $2.1 \,$.
2183:  For each fit, the degree of freedom is $n_{DoF}=n-n_{par}$.
2184: }
2185: % \tablenotetext{a}{$\beta\equiv2.1$ frozen ($n_{par}=4$)}
2186: % \tablenotetext{b}{$\beta\equiv2.1$ and $\alpha\equiv1.0$ frozen ($n_{par}=3$)}
2187: \end{deluxetable}
2188: % $\Delta_{\chi^2_r} = \sqrt{2 n_{DoF} - 1}/n_{DoF}$, with $n_{DoF} = n-n_{par} \approx n - 4 $. 
2189: % GRB 030329, P2, all 87 bins, vCB4 has $\beta=2.1$ frozen.
2190: 
2191: 
2192: \begin{deluxetable}{lcccccc}
2193: %\tabletypesize{\scriptsize}
2194: %\tabletypesize{\footnotesize}
2195: %\tabletypesize{\small}
2196: \tablecolumns{7}
2197: \tablecaption{Fit results for \CB
2198: \label{tab:CB_pars}}
2199: \tablewidth{0pt}
2200: \tablehead{ 
2201: \colhead{GRB} & 
2202: \colhead{$\Delta E_{\mbox{\scriptsize front}}\,$} & 
2203: \colhead{$\Delta E_{\mbox{\scriptsize rear}}\,$} & 
2204: \colhead{$T_{p}\,$} & 
2205: \colhead{$\alpha$} & 
2206: \colhead{$\beta$} & 
2207: \colhead{$b$} 
2208: \\ 
2209: \colhead{} &
2210: \colhead{(\keV)} &
2211: \colhead{(\keV)} &
2212: \colhead{(\keV)} &
2213: \colhead{} &
2214: \colhead{} &
2215: \colhead{}
2216: }
2217: \startdata
2218: 020715    &\nodata	  &[\phn30,15660]   & 532\PM20    & $0.741 \pm 0.077$     & 2.20\PM0.14    & \phS0.067\PM0.040       \\ \hline
2219: 021008    &[300,2800]	  &\nodata	    & 628\PM71    & $1.31 \pm 0.28$       & 2.1            & \phS0.052\PM0.076       \\ 
2220: 021008    &\nodata	  &[300,15660]      & 672\PM68    & $1.487 \pm 0.062$     & 2.77\PM0.55    & \phS0.085\PM0.092       \\
2221: 021008    &[300,2800]	  &[300,15660]      & 641\PM32    & $1.523 \pm 0.055$     & 2.1            & \phS0.020\PM0.008       \\ \hline
2222: 021206    &[\phn70,2800]  &\nodata          & 672\PM20    & $0.66 \pm 0.21$       & 1.92\PM0.67    & \phS0.063\PM0.142       \\ 
2223: 021206    &\nodata        &[300,16000]      & 672\PM24    & $0.67 \pm 0.19$       & 2.12\PM0.13    & \phS0.102\PM0.048       \\
2224: 021206    &[\phn70,2800]  &[300,16000]      & 678\PM\phn6 & $0.60 \pm 0.06$       & 2.10\PM0.08    & \phS0.103\PM0.028       \\
2225: 021206    &\nodata        &[300,\phn4500]   & 670\PM23    & $0.71 \pm 0.15$       & 2.1            & \phS0.091\PM0.012       \\
2226: 021206    &[\phn70,2800]  &[300,\phn4500]   & \nodata     & \nodata               & \nodata        & \nodata                 \\ \hline
2227: %021206 P1&\nodata  	  &[300,16000]      & 671\PM55    & $0.90 \pm 0.26$       & 2.1            & \phS0.050\PM0.012       \\
2228: %021206 P2&\nodata  	  &[300,16000]      & 723\PM33    & $0.57 \pm 0.22$       & 2.1            & \phS0.106\PM0.015       \\
2229: %021206 P3&\nodata  	  &[300,16000]      & 680\PM42    & $0.66 \pm 0.26$       & 2.1            & \phS0.083\PM0.014       \\
2230: %021206 P4&\nodata  	  &[300,16000]      & 538\PM79    & $0.70 \pm 0.54$       & 2.1            & \phS0.117\PM0.034       \\ \hline
2231: 030329 P1 &\nodata        &[\phn34,10000]   & 147\PM10    & $1.614 \pm 0.036$     & 2.1            & \phS0.033\PM0.029       \\
2232: 030329 P2 &\nodata        &[\phn34,\phn7000]& \phn69\PM15 & $1.841 \pm 0.049$     & 2.1            & \phS0.048\PM0.055       \\ \hline
2233: 030406    &\nodata	  &[\phn24,15000]   & 626\PM83    & $0.966 \pm 0.089$     & 2.1            & \phS0.18\phn\PM0.12\phn \\ \hline 
2234: 030519B   &\nodata	  &[\phn70,15000]   & 396\PM12    & $0.949 \pm 0.073$     & 2.388\PM0.097  & \phS0.135\PM0.048       \\ \hline
2235: 031027    &\nodata	  &[\phn60,\phn6000]& 340\PM17    & $0.950 \pm 0.055$     & 2.1            &    -0.010\PM0.025       \\ \hline
2236: 031111    &\nodata	  &[\phn38,15000]   & 690\PM45    & $0.68 \pm 0.27$       & 2.241\PM0.023  & \phS1.09\phn\PM0.36\phn
2237: %020715   & 72.4          &[30,15660]       & 540\PM26    & $0.784 \pm 0.050$     & 2.1            & \phS0.044\PM0.028       \\
2238: %021206   &[70,2800]      &\nodata          & 673\PM19    & $0.588 \pm 0.058$     & 2.1            & \phS0.108\PM0.025       \\ 
2239: %021206   &\nodata        &[300,16000]      & 671\PM23    & $0.69 \pm 0.14$       & 2.1            & \phS0.094\PM0.008       \\
2240: %021206   &[70,2800]      &?                & 680\PM\phn7 & $0.593 \pm 0.026$     & 2.1            & \phS0.099\PM0.006       \\
2241: \enddata                                                                                                                      
2242: \tablecomments{
2243:   $\Delta E_{\mbox{\scriptsize front/rear}}$: energy interval used to fit front/rear detector data;
2244:   $T_{p}, \alpha, \beta, b$: parameters as defined in the text below \refeq{eq:CB};
2245:   errors are symmetric $1 \sigma$ errors;
2246:   where no error is given, the parameter was frozen at that value.
2247:   }
2248: \end{deluxetable}
2249: 
2250: 
2251: \begin{deluxetable}{lccccc}
2252: %\tabletypesize{\scriptsize}
2253: %\tabletypesize{\footnotesize}
2254: %\tabletypesize{\small}
2255: \tablecolumns{6}
2256: \tablecaption{Fit results for Band function
2257:          \label{tab:Band_pars}}
2258: \tablewidth{0pt}
2259: \tablehead{
2260: \colhead{GRB} & 
2261: \colhead{$\Delta E_{\mbox{\scriptsize front}}$} & 
2262: \colhead{$\Delta E_{\mbox{\scriptsize rear}}$} & 
2263: \colhead{$E_{p}\,$}   & 
2264: \colhead{$\alpha$}   & 
2265: \colhead{$\beta$}
2266: \\ 
2267: \colhead{} &
2268: \colhead{(\keV)} &
2269: \colhead{(\keV)} &
2270: \colhead{(\keV)} &
2271: \colhead{} &
2272: \colhead{}
2273: }
2274: \startdata
2275: 020715     & \nodata      &[\phn30,15660]    & $531 \pm 24$    & $0.776 \pm 0.044$ & $3.14 \pm 0.25$   \\ \hline
2276: 021008     &[300,2800]    & \nodata          & $670 \pm 58$    & $1.36 \pm 0.23$   & $3.41 \pm 0.51$   \\ 
2277: 021008     & \nodata      &[300,15660]       & $678 \pm 42$    & $1.526 \pm 0.067$ & $3.86 \pm 0.25$   \\
2278: 021008     &[300,2800]    &[300,15660]       & $677 \pm 33$    & $1.493 \pm 0.056$ & $3.73 \pm 0.18$   \\ \hline
2279: 021206     &[\phn70,2800] & \nodata          & $713 \pm 17$    & $0.694 \pm 0.031$ & $3.20 \pm 0.13$   \\
2280: 021206     & \nodata      &[300,16000]       & \nodata         & \nodata           & \nodata           \\
2281: 021206     &[\phn70,2800] &[300,16000]       & \nodata         & \nodata           & \nodata           \\
2282: 021206     & \nodata      &[300,\phn4500]    & $709 \pm 18$    & $0.72 \pm 0.20$   & $3.186 \pm 0.063$ \\
2283: 021206     &[\phn70,2800] &[300,\phn4500]    & $711 \pm 7\phn$ & $0.692 \pm 0.020$ & $3.19 \pm 0.04$   \\ \hline
2284: %021206 P1 & \nodata      &[300,16000]       & \nodata         & \nodata           & \nodata           \\
2285: %021206 P2 & \nodata      &[300,16000]       & \nodata         & \nodata           & \nodata           \\
2286: %021206 P3 & \nodata      &[300,16000]       & \nodata         & \nodata           & \nodata           \\
2287: %021206 P4 & \nodata      &[300,16000]       & \nodata         & \nodata           & \nodata           \\ \hline
2288: 030329 P1  & \nodata      &[\phn34,10000]    & $157.2 \pm 5.2$ & $1.608 \pm 0.038$ & $3.48 \pm 0.53$   \\  
2289: 030329 P2  & \nodata      &[\phn34,\phn7000] & $\phn85 \pm 11$ & $ 1.781\pm 0.065$ & $3.04 \pm 0.30$   \\ \hline  
2290: 030406     & \nodata      &[\phn24,15000]    & $674 \pm 70$    & $0.979 \pm 0.064$ & $2.61 \pm 0.27$   \\ \hline
2291: 030519B    & \nodata      &[\phn70,15000]    & $417 \pm 13$    & $1.048 \pm 0.042$ & $3.11 \pm 0.18$   \\ \hline
2292: 031027     & \nodata      &[\phn60,\phn6000] & $338 \pm 15$    & $0.940 \pm 0.079$ & \nodata           \\ \hline
2293: 031111     & \nodata      &[\phn38,15000]    & $844 \pm 59$    & $1.102 \pm 0.036$ & $2.364 \pm 0.068$ 
2294: \enddata     
2295: \tablecomments{
2296:    $\Delta E_{\mbox{\scriptsize front/rear}}$: energy interval used to fit front/rear detector data;
2297:    $E_{p}, \alpha, \beta$: parameters as defined in the text below \refeq{eq:Band};
2298:    where no $\beta$ is given, a CPL (\refeq{eq:CPL}) was fitted;
2299:    errors are symmetric $1\sigma$ errors. 
2300:   } % from the IDL routine `curvefit'.
2301: \end{deluxetable}
2302: 
2303: 
2304: \begin{deluxetable}{lrrrrccc}
2305: %\tabletypesize{\scriptsize}
2306: %\tabletypesize{\footnotesize}
2307: %\tabletypesize{\small}
2308: %\rotate
2309: \tablecolumns{8}
2310: \tablecaption{Fluences in $10^{-5}\,$erg cm$^{-2}$
2311: 	 \label{tab:fl}}
2312: \tablewidth{0pt}
2313: \tablehead{
2314: \colhead{GRB} & 
2315: \colhead{\fluenceCB} & 
2316: \colhead{\fluenceBand} & 
2317: \colhead{\fluenceRHESSI}   &
2318: \colhead{\fluenceHETE}   &
2319: \colhead{HETE}  &
2320: \colhead{\fluenceUlysses}    &
2321: \colhead{Ulysses}   
2322: \\
2323: \colhead{} &
2324: \multicolumn{7}{c}{($10^{-5}\,$erg cm$^{-2}$)}
2325: }
2326: %               DE_CB   DE_Ba  0.1-10MeV  30-400keV   HETE           25-100keV  Ulysses  
2327: \startdata
2328:  020715       &  4.37 &  4.41 &  3.94 &  1.93    & \nodata           &  0.43    &  0.30   \\  \hline
2329:  021008 front & 26.85 & 26.85 & 41.24 & 22.92\fa & \nodata           &  8.67\fa & \nodata \\ 
2330:  021008 rear  & 35.54 & 35.64 & 48.85 & 27.15\fa & \nodata           & 10.27\fa &   8.5   \\  \hline 
2331:  021206 front & 52.15 & 52.44 & 55.67 & 19.97    & \nodata           &  3.81    & \nodata \\ 
2332:  021206 rear  & 58.74 & 53.45 & 70.55 & 25.29    & \nodata           &  4.82    &  16     \\  \hline
2333:  030329 P1    &  6.51 &  6.47 &  4.42 &  5.20    & \nodata           &  2.57    & \nodata \\  
2334:  030329 P2    &  3.58 &  3.58 &  2.26 &  2.95    & \nodata           &  1.69    & \nodata \\  
2335:  030329 total &\nodata&\nodata&  7.35 &  9.46    & $10.76 \pm 0.14$  &  4.93    & \nodata \\  \hline
2336:  030406       &  4.81 &  4.81 &  4.26 &  1.62    & \nodata           &  0.40    &   1.3   \\  \hline
2337:  030519B      & 10.27 & 10.36 &  9.56 &  6.07    & $6.10 \pm 0.1$    &  1.78    & \nodata \\  \hline 
2338:  031027       &  5.37 &  5.45 &  4.81 &  3.97    & \nodata           &  1.17    &   1.4   \\  \hline 
2339:  031111       &  7.40 &  7.40 &  6.59 &  2.10    & 1.714             &  0.56    &   0.21 
2340: \enddata     
2341: \tablecomments{
2342: \fluenceCB: fluence from  \CB\ fit (Table~\ref{tab:CB_pars});
2343: \fluenceBand: fluence from Band function fit (Table~\ref{tab:Band_pars});
2344: \fluenceRHESSI: fluence in [100,10000]~\keV\ (RHESSI range);
2345: \fluenceHETE: fluence in [30,400]~\keV\ (HETE range);
2346: \fluenceUlysses: fluence in [25,100]~\keV\ (Ulysses range);
2347: HETE: HETE fluences from references cited in \S \ref{sec:results};
2348: Ulysses: Ulysses fluences from references cited in Table~\ref{tab:GRBs}.
2349: }
2350: \tablenotetext{a}{Our fits overestimate the real counts. More realistic is
2351: \fluenceUlysses$=4.8\times10^{-5}\,\mbox{erg cm}^{-2}$.
2352: }
2353: \end{deluxetable}
2354: 
2355: \begin{deluxetable}{cccccccc}
2356: %\tabletypesize{\scriptsize}
2357: %\tabletypesize{\footnotesize}
2358: %\tabletypesize{\small}
2359: \tablecolumns{8}
2360: \tablecaption{Peak resolved analysis of GRB 021206
2361:          \label{tab:dt_grb021206}}
2362: \tablewidth{0pt}
2363: \tablehead{
2364: \colhead{$\Delta t$} &
2365: \colhead{duration} &
2366: \colhead{$T_p$} & 
2367: \colhead{$\alpha$} & 
2368: \colhead{$\beta$} & 
2369: \colhead{$b$} & 
2370: \colhead{$F_{CPL}$} & 
2371: \colhead{$F_{mPL}$}
2372: \\
2373: \colhead{} &
2374: \colhead{(s)} &
2375: \colhead{(\keV)} &
2376: \colhead{} &
2377: \colhead{} &
2378: \colhead{} &
2379: \multicolumn{2}{c}{($10^{-5}\,$erg cm$^{-2}$)}
2380: }
2381: \startdata
2382: P1   & 1.366 & $661 \pm 18$  & $0.77 \pm 0.06$ & $2.1$ & $0.059 \pm 0.010$        & 11.7   & \phn3.0 \\
2383: P2   & 1.366 & $732 \pm 12$  & $0.42 \pm 0.05$ & $2.1$ & $0.115 \pm 0.010$        & 14.5   & 12.9    \\
2384: P3   & 1.366 & $684 \pm 14$  & $0.63 \pm 0.05$ & $2.1$ & $0.085 \pm 0.012$        & 12.7   & \phn6.3 \\
2385: P4   & 1.366 & $530 \pm 20$  & $0.80 \pm 0.08$ & $2.1$ & $0.113 \pm 0.018$        & \phn6.9& \phn3.8 \\ 
2386: tail & 4.097 & $160 \pm 60$  &  1.0            & $2.1$ & $2.5\;^{+\infty}_{-1.5}$ & \phn0.2& \phn3.1
2387: \enddata
2388: \tablecomments{
2389:  $\Delta t$: time period, cf.\ Fig.~\ref{fig:ltc_GRBsI};
2390:  $T_p$, $\alpha$, $\beta$ and $b$: \CB\ parameters;
2391:  \fluenceCPL: fluence of the CPL-component in the \CB\ function (\refeq{eq:CB}),
2392:  \fluencemPL: fluence of the modified PL component of \refeq{eq:CB};
2393:  fluences are for the range [100,10000]~\keV.
2394: }
2395: \end{deluxetable}
2396: 
2397: 
2398: \clearpage
2399: 
2400: \begin{figure}
2401: \epsscale{0.52} %only for preprint, remove for 2-column version 
2402: \plotone{f1a.eps} %was: \plotone{020715_ltc.eps}
2403: \plotone{f1b.eps} %was: \plotone{021008_ltc.eps}
2404: \plotone{f1c.eps} %was: \plotone{021206_ltc.eps}
2405: \plotone{f1d.eps} %was: \plotone{030329_ltc.eps}
2406: \caption{lightcurves
2407:   for the energy band 20~\keV\ to 3~\MeV;
2408:   black: rear detectors,
2409:   grey: front detectors;
2410:   broken vertical lines:
2411:   time intervals used for spectral analysis.
2412:   \label{fig:ltc_GRBsI}}
2413: \end{figure}
2414: 
2415: \begin{figure}
2416: \epsscale{0.52} %only for preprint, remove for 2-column version 
2417: \plotone{f2a.eps} %was: \plotone{030406_ltc.eps}
2418: \plotone{f2b.eps} %was: \plotone{030519_ltc.eps}
2419: \plotone{f2c.eps} %was: \plotone{031027_ltc.eps}
2420: \plotone{f2d.eps} %was: \plotone{031111_ltc.eps}
2421: \caption{lightcurves
2422:   for the energy band 20~\keV\ to 3~\MeV;
2423:   black: rear detectors,
2424:   grey: front detectors;
2425:   broken vertical lines:
2426:   time intervals used for spectral analysis.
2427:   \label{fig:ltc_GRBsII}}
2428: \end{figure}
2429: 
2430: \begin{figure}
2431: \epsscale{1}
2432: \plotone{f3.eps} %was: \plotone{spec_020715.eps}
2433: \caption{rear spectrum of GRB 020715;
2434:    error bars: photon counts after background subtraction;
2435:    black histogram: \CB\ fit (\refeq{eq:CB});
2436:    grey histogram:  Band function fit (\refeq{eq:Band});
2437:    vertical broken lines: energy range used for fitting;
2438:    bottom: residuals of \CB\ (black) and Band (grey) fit.
2439:   \label{fig:020715}}  
2440: \end{figure}
2441: 
2442: 
2443: \begin{figure}
2444: \plotone{f4.eps} %was: \plotone{spec_021008_front.eps}
2445: \caption{front spectrum of GRB 021008;
2446:    explanations see caption of Fig.~\ref{fig:020715};
2447: %   Observed spectrum (crosses) and spectral fits of GRB 021008,
2448: %   front detectors.
2449: %   The Band function (black line histogram) and the Cannonball
2450: %   model (grey histogram) fit the data almost equally well.
2451: %   The residuals are plotted for the Band function.
2452:   the same set of parameters 
2453:   (eqs.~\ref{eq:CB021008} and \ref{eq:Band021008}) 
2454:   is used for this plot and Fig.~\ref{fig:R021008}.
2455:   \label{fig:F021008}}  
2456: \end{figure}
2457: 
2458: 
2459: \begin{figure}
2460: \plotone{f5.eps} %was: \plotone{spec_021008_rear.eps}
2461: \caption{rear spectrum of GRB 021008;
2462:    explanations see caption of Fig.~\ref{fig:020715}.
2463: %   Observed spectrum (crosses) and spectral fits of GRB 021008,
2464: %    rear detectors.
2465: %   The Band function (black line histogram) and the Cannonball
2466: %   model (grey histogram) fit the data equally well.
2467: %  The residuals are plotted for the Band function.
2468:   the same set of parameters
2469:   (eqs.~\ref{eq:CB021008} and \ref{eq:Band021008})
2470:   is used for this plot and Fig.~\ref{fig:F021008}.
2471:   \label{fig:R021008}}  
2472: \end{figure}
2473: 
2474: 
2475: \begin{figure}
2476: \plotone{f6.eps} %was: \plotone{spec_021206_front.eps}
2477: \caption{spectrum of GRB 021206, front detectors;
2478:   explanations see caption of Fig.~\ref{fig:020715}.
2479:   the same set of parameters 
2480:   (eqs.~\ref{eq:CB021206} and \ref{eq:Band021206}) 
2481:   is used for this plot and Fig.~\ref{fig:spec_rear}.
2482:   \label{fig:spec_front}}
2483: \end{figure}
2484: 
2485: 
2486: \begin{figure}
2487: \plotone{f7.eps} %was: \plotone{spec_021206_rear.eps}
2488: \caption{top: spectrum of GRB 021206, rear detectors;
2489:    symbols as in Fig.~\ref{fig:020715}.
2490:    bottom: residuals of \CB\ fit (black) and Band fit (grey).
2491:    The Band function was fitted only up to 4.5~\MeV.
2492:    The excess counts below 300~\keV\ are backscatters from Earth, see text.
2493:    The same set of parameters 
2494:    (eqs.~\ref{eq:CB021206} and \ref{eq:Band021206})
2495:    is used for this plot and Fig.~\ref{fig:spec_front}.
2496:    \label{fig:spec_rear}}
2497: \end{figure}
2498: 
2499: 
2500: \begin{figure}
2501: \plotone{f8.eps} %was: \plotone{spec_030329_p1.eps}
2502: \caption{rear spectrum of GRB 030329, first peak;
2503:    explanations see caption of Fig.~\ref{fig:020715}.
2504: %    Observed spectrum (crosses) and spectral fits of GRB 030329, 
2505: %   first peak. The Band function (black) and the Cannonball model (grey)
2506: %   fit the data equally well, the main contribution being - in both
2507: %   cases - a cut off power law (CPL).  
2508:    \label{fig:030329_p1}}  
2509: \end{figure}
2510: 
2511: \begin{figure}
2512: \plotone{f9.eps} %was: \plotone{spec_030329_p2.eps}
2513: \caption{rear spectrum of GRB 030329, second peak;
2514:    explanations see caption of Fig.~\ref{fig:020715}.
2515: %   Observed spectrum (crosses) and spectral fits of GRB 030329, 
2516: %   second peak. A good fit is obtained with a cut off power law (CPL, in black).
2517: %   Using a band function or Cannonball model hardly
2518: %   improves the fit. Shown in grey is the fit of a broken power law (BPL),
2519: %   whose low energy index alpha $\approx 2.0$.   
2520:   \label{fig:030329_p2}}  
2521: \end{figure}
2522: 
2523: \begin{figure}
2524: \plotone{f10.eps} %was: \plotone{spec_030406.eps}
2525: \caption{rear spectrum of GRB 030406;
2526:    explanations see caption of Fig.~\ref{fig:020715}.
2527: %   Observed spectrum (crosses) and spectral fits of GRB 030406.
2528: %   The best spectral fit is the Cannonball model with
2529: %   $\alpha$ and $\beta$ fixed at their theoretically expected values 
2530: %   (black line histogram).
2531: %   The second best fit is a Band function (grey histogram).
2532: %  The residuals are plotted for the best fit.
2533:   \label{fig:030406}}  
2534: \end{figure}
2535: 
2536: \begin{figure}
2537: \plotone{f11.eps} %was: \plotone{spec_030519b.eps}
2538: \caption{rear spectrum of GRB 030519B;
2539:    explanations see caption of Fig.~\ref{fig:020715}.
2540: %  Observed spectrum (crosses) and spectral fits of GRB 0300519B.
2541: %  The best spectral fit is a Band function (line histogram).
2542:   \label{fig:030519B}}  
2543: \end{figure}
2544: 
2545: \begin{figure}
2546: \plotone{f12.eps} %was: \plotone{spec_031027.eps}
2547: \caption{rear spectrum of GRB 031027;
2548: %   explanations see caption of Fig.~\ref{fig:020715}.
2549: %  Crosses: observed counts. 
2550:   histogram and residuals: fit of a cut off power law (CPL);
2551:   a CPL is equivalent to a Band function with
2552:   $\beta = \infty$  or \CB\ with $b=0$.
2553:   \label{fig:031027}}  
2554: \end{figure}
2555: 
2556: \begin{figure}
2557: \plotone{f13.eps} %was: \plotone{spec_031111.eps}
2558: \caption{rear spectrum of GRB 031111;
2559:    explanations see caption of Fig.~\ref{fig:020715}.
2560: %  Crosses: observed counts. 
2561: %  Black histogram and residuals: fit of \CB.
2562: %  Grey histogram: fit of Band function. 
2563:   \label{fig:031111}}  
2564: \end{figure}
2565: 
2566: 
2567: 
2568: \end{document}
2569: 
2570: %The caveat in \S \ref{sec:systematics} concerns only
2571: %the low energy part of the spectrum, thus does not
2572: %affect this conclusion. 
2573: \begin{deluxetable}{lcccllllll}
2574: %\tabletypesize{\scriptsize}
2575: %\tabletypesize{\footnotesize}
2576: %\tabletypesize{\small}
2577: %\rotate
2578: \tablecolumns{10}
2579: \tablecaption{Reduced Chi-square of spectral fits
2580: 	 \label{tab:chi2}}
2581: \tablewidth{0pt}
2582: \tablehead{
2583: \colhead{GRB} & 
2584: \colhead{$\Delta E_{\mbox{\scriptsize front}}\,$} & 
2585: \colhead{$\Delta E_{\mbox{\scriptsize rear}}\,$} & 
2586: \colhead{$n$} & 
2587: \colhead{CPL} & 
2588: \colhead{Band} &
2589: \colhead{\CB} &
2590: \colhead{BPL} &
2591: \colhead{BBPL} &
2592: \colhead{BBmPL}
2593: \\ 
2594: \colhead{} &
2595: \colhead{(\keV)} &
2596: \colhead{(\keV)} &
2597: \colhead{} &
2598: \colhead{$n_{par}=3$} &
2599: \colhead{4} &
2600: \colhead{5} &
2601: \colhead{4} &
2602: \colhead{4} &
2603: \colhead{4}
2604: }
2605: %GRB        front interval  rear interval      n      CPL      BAND    CB          BPL      BBPL    BBmPL
2606: \startdata
2607: 020715     & \nodata       &[\phn30,15660]   & 117 &  0.999  & 0.941   & 0.980\fa   & 1.143 & 2.389 & 1.390 \\ \hline
2608: 021008     &[300,2800]     &\nodata          &  38 &  1.014  & 1.024   & 1.031\fa   & 0.971 & 0.967 & 0.979 \\ 
2609: 021008     &\nodata        &[300,15660]      &  50 &  0.931  & 0.852   & 0.867\fa   & 1.150 & 2.118 & 1.308 \\
2610: 021008     &[300,2800]     &[300,15660]      &  88 & \nodata & 0.946   & 0.962\fa   &\nodata&\nodata&\nodata\\ \hline 
2611: 021206     &[\phn70,2800]  &\nodata          & 112 &  1.198  & 0.959   & 0.972      & 1.440 & 2.923 & 1.776 \\  
2612: 021206     &\nodata        &[300,16000]      &  78 &  4.511  & 1.801   & 1.133      & 1.796 & 1.271 & 1.493 \\
2613: 021206     &[\phn70,2800]  &[300,16000]      & 190 & \nodata & \nodata & 1.014      &\nodata&\nodata&\nodata\\
2614: 021206     &\nodata        &[300,\phn4500]   &  66 & \nodata & 1.174   & 1.126\fa   &\nodata&\nodata&\nodata\\
2615: 021206     &[\phn70,2800]  &[300,\phn4500]   & 178 & \nodata & 1.014   & \nodata    &\nodata&\nodata&\nodata\\ \hline
2616: %021206 P1 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?          & ?     & ?     & ?     \\
2617: %021206 P2 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?          & ?     & ?     & ?     \\
2618: %021206 P3 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?          & ?     & ?     & ?     \\
2619: %021206 P4 &\nodata        &[300,16000]      &  ?  &  ?      & ?       & ?          & ?     & ?     & ?     \\ \hline 
2620: 030329 P1  &\nodata        &[\phn34,10000]   &  94 &  0.950  & 0.937   & 0.942      & 0.992 & 1.531 & 0.990 \\ 
2621: 030329 P2  &\nodata        &[\phn34,\phn7000]&  87 &  1.244  & 1.244   & 1.244\fa   & 1.191 & 1.229 & 1.186 \\ \hline
2622: 030406     &\nodata        &[\phn24,15000]   &  75 &  1.098  & 1.066   & 1.041\fb   & 1.249 & 2.130 & 1.249 \\ \hline
2623: 030519B    &\nodata        &[\phn70,15000]   &  79 &  1.345  & 1.151   & 1.205      & 1.330 & 2.530 & 1.464 \\ \hline 
2624: 031027     &\nodata        &[\phn60,\phn6000]&  63 &  1.060  &\phn n.c.& \phn n.c.  & 1.576 & 2.354 & 1.688 \\ \hline 
2625: 031111     &\nodata        &[\phn38,15000]   & 117 &  1.600  & 1.135   & 1.158      & 1.243 & 2.354 & 1.177
2626: \enddata     
2627: \tablecomments{$\chi^2_r=\chi^2/n_{DoF}$ obtained by fitting different spectral models to the data;
2628:  $\Delta E_{\mbox{\scriptsize front/rear}}$: energy interval used to fit front/rear detector data;
2629:  $n_{par}$: number of free parameters;
2630:  $n$: number of energy bins, $n_{DoF}=n-n_{par}$;
2631:  CPL: cut off power law \refeq{eq:CPL},
2632:  Band: Band function \refeq{eq:Band},
2633:  \CB: cannonball model \refeq{eq:CB},
2634:  BPL: broken power law \refeq{eq:BPL},
2635:  BBPL: blackbody plus power law \refeq{eq:BBPL},
2636:  BBmPL: blackbody plus modified power law \refeq{eq:BBmPL};
2637:  n.c.: fit did not converge.
2638: }
2639: \tablenotetext{a}{$\beta\equiv2.1$ frozen ($n_{par}=4$)}
2640: \tablenotetext{b}{$\beta\equiv2.1$ and $\alpha\equiv1.0$ frozen ($n_{par}=3$)}
2641: \end{deluxetable}
2642: % $\Delta_{\chi^2_r} = \sqrt{2 n_{DoF} - 1}/n_{DoF}$, with $n_{DoF} = n-n_{par} \approx n - 4 $. 
2643: % GRB 030329, P2, all 87 bins, vCB4 has $\beta=2.1$ frozen.
2644: 
2645: