1:
2: %\documentclass[preprint,12pt]{aastex}
3: \documentclass[useAMS,onecolumn]{mn2e}
4: %\documentclass[useAMS]{mn2e}
5: \usepackage{times}
6:
7: \usepackage{graphicx}
8: \usepackage{amsmath}
9: \usepackage{amssymb}
10:
11:
12: \title{Outliers to the Isotropic Energy - Peak Energy Relation in GRBs}
13:
14:
15: \author[E. Nakar \& T. Piran]
16: {Ehud Nakar,$^1$
17: and Tsvi Piran,$^{1,2}$ \\
18: $^1$Theoretical Astrophysics,
19: Caltech, Pasadena, CA 91125, USA\\
20: $^2$Racah Institute for Physics, The Hebrew
21: University, Jerusalem 91904, Israel}
22:
23: \begin{document}
24: \maketitle
25:
26: \begin{abstract}
27:
28: The peak energy - isotropic energy (EpEi) relation is among the most
29: intriguing recent discoveries concerning GRBs. It can have numerous
30: implications on our understanding of the emission mechanism of the
31: bursts and on the application of GRBs for cosmological studies.
32: However, this relation was verified only for a small sample of
33: bursts with measured redshifts. We propose here a test whether a
34: burst with an unknown redshift can potentially satisfy the EpEi
35: relation. Applying this test to a large sample of BATSE bursts we
36: find that a significant fraction of those bursts cannot satisfy this
37: relation. Our test is sensitive only to dim and hard bursts and
38: therefore this relation might still hold as an inequality (i.e.
39: there are no intrinsically bright and soft bursts). We conclude that
40: the observed relation seen in the sample of bursts with a known
41: redshift might be influenced by observational biases and from the
42: inability to locate and well localize hard and weak bursts that have
43: only a small number of photons. In particular we point out that the
44: threshold for detection, localization and redshift measurement is
45: essentially higher than the threshold for detection alone. We
46: predict that Swift will detect some hard and weak bursts that would
47: be outliers to the EpEi relation. However, we cannot quantify this
48: prediction. We stress the importance of understanding the
49: detection-localization-redshift threshold for the coming Swift
50: detections.
51:
52: \end{abstract}
53:
54:
55:
56: \section{Introduction}
57: The detection of Gamma-ray Bursts (GRBs) afterglows enabled the
58: determination of the redshift for a few dozens bursts (out of
59: several thousands observed so far). This yielded a small sample of
60: bursts for which the {\it observed} properties can be translated
61: into {\it intrinsic} ones. This, in turn, initiated the search for
62: relations between various intrinsic properties. Such a relation can
63: have far reaching implications both on the theoretical understanding
64: of GRBs and on the application of GRBs as a tool.
65:
66:
67: Even before a large sample of bursts with redshift was available, it
68: was suggested that the intrinsic $E_p$ and $E_{iso}$ are correlated
69: (Lloyd et al. 2000, Lloyd-Ronning \& Ramirez-Ruiz 2002). Once more
70: than a dozen redshifts were measured, Amati et al. (2002) reported a
71: tight relation between the isotropic equivalent bolometric energy
72: output in $\gamma$-rays, $E_{iso}$, and the {\it intrinsic} peak
73: energy of the $\nu f_\nu$ spectrum, $E_p$ (hereafter we denote the
74: $E_p$-$E_{iso}$ relation as EpEi):
75: \begin{equation}
76: E_{iso}=A_k E_p^k \ , \label{EQ EpEi}
77: \end{equation}
78: where $k \sim 2$ and $A_k$ is a constant. This result was based on a
79: sample of 12 BeppoSAX bursts with known redshifts. Ten additional
80: bursts detected by HETE II (Lamb et al., 2004; Atteia et al., 2004)
81: supported this result and extended it down to $E_{iso} \sim
82: 10^{49}$ergs (see also Ghirlanda et al. 2004a).
83:
84:
85: Two bursts, within the current sample of bursts with a known
86: redshift, GRB 980425 and GRB 031203 are clear outliers to the EpEi
87: relation. Both are dim (low $E_{iso}$) and hard (high $E_p$). These
88: two bursts are usually ignored as a peculiar outliers to a confirmed
89: relation. Even though the EpEi relation is based on a small and
90: unique sample (bursts with confirmed redshift and a well observed
91: spectrum), and even though there are two clear outliers, this
92: relation initiated numerous attempts to explain it theoretically and
93: to use it for various applications. Therefore, testing the validity
94: of the EpEi relation with the largest available sample (of several
95: thousands BATSE bursts), is extremely important. This is the goal of
96: this letter.
97:
98: We present here (Eq. \ref{EQ d}) a simple test whether a burst can
99: potentially satisfy the EpEi relation. This test can be carried out
100: for bursts with unknown redshift as long as we have a lower limit on
101: the observed peak energy, $E_{p,obs}$ and an upper limit on the
102: observed bolometric fluence, $F$. A burst that fails this test must
103: be an outlier satisfying: $E_{iso} < A_k E_p^k$. On the other hand a
104: burst that passes this test does not necessarily satisfy the EpEi
105: relation. One of the known outliers, GRB 980425, fails the test only
106: marginally. However its low measured redshift puts it as a clear
107: outlier.
108:
109: First, we apply the test to a larger, but yet limited, sample of
110: $63$ BATSE bursts with unknown redshifts and a good spectral data
111: (taken from Band et al. 1993 and Jimenez, Band, \& Piran, 2001). We
112: find that at least $\sim 25\%$ out of these bursts significantly
113: fail the test and therefore are essentially outliers to the EpEi
114: relation. Next, we consider the full current BATSE catalog
115: \footnote{http://www.batse.msfc.nasa.gov/batse/grb/catalog/current/},
116: for which we test all the long bursts ($T_{90}>2sec$) with a
117: complete fluence data in all the four energy channels. The exact
118: spectrum for these bursts is unknown, but we can still use the BATSE
119: four energy channels data to obtain a lower limit on $E_{p,obs}$ for
120: about half of the bursts. We find that $\sim 25\%$ of the bursts in
121: the BATSE sample fail the test, and must be outliers to the EpEi
122: relation. The large numbers of outliers that we find in the
123: different samples of BATSE bursts, suggest that the EpEi relation is
124: not a generic property of GRBs. Our results do not, however, rule
125: out possible correlation between $E_p$ and $E_{iso}$. We also do not
126: test here the recently suggested relation between $E_p$ and the
127: beaming-corrected energy (Ghirlanda et al. 2004a), see however Band
128: and Preece (2005).
129:
130:
131: In \S 2 we present the basic ideas of our analysis. We apply the
132: test to the sample of BATSE bursts with a known peak energy in \S 3
133: and to the whole BATSE catalog in \S 4. We discuss the implications
134: of this result as well as possible reasons why so few outliers
135: were found in the samples of bursts with known redshifts in \S 5.
136:
137: \section{Trajectories on the $E_{iso}$-$E_p$ plane\label{Trajectory}}
138:
139: Consider a burst with known bolometric fluence, $F$, and observed
140: peak energy,$E_{p,obs}$, but an unknown redshift, $z$. Assuming a
141: $z$ value we can evaluate the intrinsics $E_{iso}$ and $E_p$. The
142: trajectory of the burst on the $(E_{iso},E_p)$ plane as we vary z is
143: given by:
144: \begin{eqnarray}
145: E_{iso} &=& 4 \pi D^2 \tilde r_c^2(z) (1+z) F \label{EQ Ep}\\
146: E_p &=& (1+z) E_{p,obs} \label{EQ Ei}\ ,
147: \end{eqnarray}
148: where $D\equiv c/H_0$ and $\tilde r_c(z)$ is the dimensionless
149: comoving distance to redshift $z$. This trajectory represent all
150: the possible values of the intrinsic $E_p$ and $E_{iso}$ for given
151: $E_{p,obs}$ and $F$. On these trajectories $E_p \propto E_{iso}^0$
152: for small $E_{iso}$ values while $E_p \propto E_{iso}$ for
153: asymptotically large values of $E_{iso}$. Several such trajectories
154: are plotted in Fig. \ref{Fig_trajectory}.
155:
156: The EpEi relation (Eq. \ref{EQ EpEi}) is represented by a curve on
157: the $(E_{iso},E_p)$ plane. For $k\ge 1$ (which is satisfied by any
158: reasonable fit to the observed data) there are values of
159: ($F,E_{p,obs}$) for which the trajectories (on the $E_{iso},E_p$
160: plane) do not intersect the EpEi curve for any value of $z$. These
161: trajectories correspond to outliers to the EpEi relations (which is
162: not satisfied for any value of $z$). Put differently, one can
163: imagine using the EpEi relation to determine the redshift of
164: observed bursts. For the bursts that the trajectories do not
165: intersect there will be no value of z for which the EpEi relation is
166: satisfied (Ghirlanda et. al. 2004b). A second group of $F,E_{p,obs}$
167: values are these for which the trajectories intersect the EpEi line.
168: These bursts can potentially satisfy the EpEi relation as there is
169: a possible $z$ value for which this relation can be satisfied. Fig.
170: \ref{Fig_trajectory} illustrates the two types of trajectories.
171:
172:
173: Substituting Eqs. \ref{EQ Ep} \& \ref{EQ Ei} in Eq. \ref{EQ EpEi} we
174: obtain a general condition for an intersection between a trajectory
175: of an observed burst and the EpEi line:
176: \begin{equation}
177: {A_k \over 4 \pi D^2} {E_{p,obs}^k \over F} = {r_c^2(z) \over
178: (1+z)^{k-1}} \ .
179: \end{equation}
180: The dimensionless function on the r.h.s. depends only on $z$. It
181: vanishes as $z$ vanish and at large values of $z$ (for $k>1$) and
182: hence it has some maximal value denoted $C_k$. All the bursts for
183: which the observables on the l.h.s. are larger than this maximal
184: value are outliers to the EpEi relation. We define a ratio
185: \begin{equation} \label{EQ d}
186: d_k \equiv {A_k \over 4 \pi D^2 C_k} {E_{p,obs}^k \over F} \ .
187: \end{equation}
188: \begin{itemize}
189: \item Bursts with $d_k<1$ can potentially satisfy the EpEi
190: relation. \item Bursts with $d_k>1$ cannot satisfy the EpEi
191: relation. For these bursts $d_k$ is a measure of the minimal
192: ``distance" of the burst from the EpEi relation. Namely, the
193: observed combination $E_p^k/F$ should decrease by this factor in
194: order that the EpEi relation would be potentially satisfied.
195: \end{itemize}
196: \begin{figure}
197: {\par\centering \resizebox*{0.7\columnwidth}{!}
198: {\includegraphics{fig1.eps}}
199: \par} \caption{\label{Fig_trajectory}. Trajectories of three bursts
200: from Band et al (1993) and Sakamoto et al (2003) on the
201: $(E_{iso},E_p)$ plane. For low redshift values the trajectory is on
202: the left side of the figure as $E_p \rightarrow E_{p,obs}$ while
203: $E_{iso} \rightarrow 0$. As $z$ increases both $E_p$ and $E_{iso}$
204: increase (asymptotically both increase linearly with $z$) and the
205: trajectory moves to the upper right. The trajectory of GRB
206: 021211(solid curve) represent a trajectory of a burst consistent
207: with the EpEi relation (for $k=2$, with $A_2 = 1^{+1}_{-.5} \cdot
208: 10^{48}$ergs/keV$^2$) as it intersects the EpEi curve (gray region).
209: The exact position of GRB 021211 (for which the redshift is known,
210: z=1) on this trajectory is marked with a full square. The trajectory
211: of GRB 910809 (dashed curve) represent a trajectory of a burst
212: inconsistent with the EpEi relation. It does not intersects the EpEi
213: curve for any value of z. The trajectory of GRB 920307 (dotted
214: curve) is marginally consistent with the EpEi relation. }
215: \end{figure}
216:
217:
218: \section{Bursts with a known Observed Peak Energy}
219:
220: Following the observations (Amati et al., 2002; Lamb et al., 2004;
221: Atteia et al., 2004) we present here (and in \S 4) the results for
222: $k=2$ with $A_2 = 1^{+1}_{-.5} \cdot 10^{48}$ergs/keV$^2$. The error
223: introduced here is our estimate of the spread in the data. All the
224: bursts in the sample of Atteia et al. (2004) are consistent within
225: $1\sigma$ with these values. Our results do not change qualitatively
226: for other values of $k$ and $A_k$ that yield a reasonable fit to the
227: data. The cosmological parameters that we consider are
228: $\Omega_m=0.3$, $\Omega_\Lambda=0.7$ and $h=0.7$, for which $C_2=
229: 0.56$. For these values we obtain:
230: \begin{equation} \label{EQ d2}
231: d_2= 8 \cdot 10^{-10} {(E_{p,obs}/1 keV)^2 \over (F/(1 erg\;
232: cm^{-2}))}
233: \end{equation}
234:
235:
236: We consider a sample of BATSE bursts (from Band et al., 1993, and
237: Jimenez et al., 2001) with unknown redshifts for which the observed
238: peak energy has been determined. We consider only bursts with a high
239: spectral index smaller than $-2$ in order to ensure that the break
240: energy in the observed spectrum is indeed the peak of $\nu F_\nu$.
241: Our sample includes $63$ ($40$ bursts from Band et al. 1993, and 23
242: bursts from Jimenez et al. 2001). Using the spectral fits for these
243: bursts we derive their bolometric fluence (0.1-10000 keV).
244:
245: Fig. \ref{Fig_C2} depicts a color map of $d_2$ for each burst on the
246: $F,E_{p,obs}$ plane. The observed values of our sample (including
247: error bars where available) are marked on this map. From Fig.
248: \ref{Fig_C2} it is evident that a significant fraction of the bursts
249: cannot satisfy the EpEi relation. Fig. \ref{Fig hist} depicts a
250: histogram of the fraction of bursts with $d_2$ larger than a given
251: value. We account for uncertainties in the measurement of
252: $E_{p,obs}$, when possible, by using an $E_{p,obs}$ value that is
253: smaller by $1\sigma$ than the measured value (unfortunately we can
254: do it only for the Band et al. 1993 sample since the uncertainties
255: in the measurement of $E_{p,obs}$ are not reported in Jimenez et
256: al., 2001). Fig. \ref{Fig hist} shows that $\approx 40\%$ of the
257: bursts have $d_2>2$ while $25\%$ of the bursts have $d_2>4$ (9/40
258: from Band et al., 1993 and 6/23 from Jimenez et al., 2001). Since
259: the scatter in the EpEi relation is a factor of 2 we consider,
260: conservatively, a burst with $d_2>4$ as an outlier. Finally, $13\%$
261: of the bursts are very far from the relation having $d_2>10$. We
262: stress that these are only lower limits. While bursts for which
263: $d_2<1$ can satisfy the EpEi relation, they do not necessarily do
264: so.
265:
266:
267:
268: \begin{figure}
269: {\par\centering \resizebox*{0.7\columnwidth}{!}
270: {\includegraphics{fig2.eps}}
271: \par} \caption{\label{Fig_C2}. A color map of $d_2$. The region marked
272: in white, where $d_2<1$ corresponds to allowed solutions of the
273: EpEi relation. Larger values are marked with darker colors and they
274: correspond to the minimal ratio between $E_{iso}$ given by the EpEi
275: relation and $E_{iso}$ given by the $(E_p,E_{iso})$ trajectory, for
276: the same value of $E_p$. Also marked on this figure are values of
277: $F$ and $E_{pobs}$ for 39 BATSE bursts from Band et al., (1993)
278: (diamonds), and 22 BATSE bursts from Jimenez et al. (2001)
279: (squares). For 29 [15] out of these 61 bursts $d_2
280: > 2 [4]$. GRB 980425 (full star) has a marginal $d_2 \approx 3$.}
281: \end{figure}
282:
283: \section{BATSE Bursts}
284:
285: Only a small fraction of BATSE bursts have a published $E_{pobs}$
286: values. Still we can obtain a lower limit of $E_{pobs} > 250$keV for
287: all BATSE bursts for which:
288: \begin{equation}
289: {F_{300,2000} \over F_{20,50} + F_{50,100}+F_{100,300}} > 1.25 \ ,
290: \label{IneqEp}
291: \end{equation}
292: where $F_{E_1,E_2}$ is the fluence between $E_1$ and $E_2$ reported
293: in the four BATSE windows. This lower limit holds for a Band spectra
294: (Band et al. 1993) over a wide range of low and high spectral
295: indices ($\alpha$ and $\beta$ respectively). As a test of the
296: validity and robustness of this criterion we apply it to the BATSE
297: bursts with known $E_p$ (Band et al. 1993 and Jimenez et al. 2001,
298: including those with $\beta>-2$ and those with known redshift). We
299: find that indeed all the bursts in the sample, apart for one, that
300: satisfy Eq. \ref{IneqEp} have $E_{p,obs} > 250$keV (23 bursts all
301: together). Using this lower limit on $E_{p,obs}$ we can obtain a
302: lower limit on $d_2$ for a large sample of BATSE bursts, where we
303: take $F$ in $20-2000$ keV energy range (the sum of all 4 channels)
304: as the bolometric fluence.
305:
306: We consider a sample of $751$ long ($T_{90}>2$sec) bursts from the
307: current BATSE catalogue. Our selection criteria were having fluence
308: in all four BATSE bands, having errors that are smaller than half of
309: the measured values in all the four bands and having a measured
310: $T_{90}$. $361$ of these bursts satisfy Eq. \ref{IneqEp} yielding a
311: lower limit on their $E_p$. Fig. \ref{Fig hist} depicts also the
312: fraction of long bursts out of the sample of $751$ bursts, that
313: satisfy Eq. \ref{IneqEp} and have $d_2>n$. We find that
314: approximately $35\%$ of these bursts have $d_2>2$, about $30\%$ have
315: $d_2>4$ and for $10\%$ this ratio is larger than 15!. While this
316: estimate is less robust than the previous ones (i.e. we cannot
317: quantify the error in the lower limit we obtain for $E_{p,obs}$) it
318: is clear that a significant fraction of long BATSE bursts cannot
319: satisfy the EpEi relation. This result has been confirmed by Band
320: and Preece (2005) that use a sample of 760 BATSE bursts where
321: $E_{p,obs}$ is known.
322:
323: Finally, we have also performed the same test for the 187 short
324: ($T_{90}<2$sec) BATSE bursts satisfying the same criteria. These
325: bursts are typically harder than long ones. As they are shorter they
326: also have a lower overall fluence. One could expect that they won't
327: satisfy the EpEi inequality. We find that more than 75\% of BATSE
328: short bursts have $d_2>10$. Short bursts cannot satisfy the EpEi
329: relation! This result is similar to the one obtained by Ghirlanda et
330: al. (2004).
331:
332:
333:
334: \begin{figure}
335: {\par\centering \resizebox*{0.7\columnwidth}{!}
336: {\includegraphics{fig3.eps}}
337: \par} \caption{\label{Fig hist}. A cumulative fraction of
338: BATSE bursts with $d_2>n$ as a function of $n$ from the samples of
339: Band et al 1993 (thick line), Jimenez et al. 2001 (dashed line) and
340: the current BATSE catalog (thin line). In the last sample (BATSE
341: catalog) $E_{pobs}$ was taken as larger than $250$keV for any burst
342: that satisfies Eq. \ref{IneqEp}.}
343: \end{figure}
344:
345:
346: \section{Discussion}
347:
348: We have presented a simple method for testing whether a burst can
349: {\it potentially} satisfy the $E_p$-$E_{iso}$ relation. This method
350: requires only two observables, the bolometric $\gamma$-rays flux and
351: the peak energy. Both can be determined for every observed burst
352: regardless of its localization and redshift determination. We have
353: carried this test for several samples of BATSE bursts. We find that
354: $\approx 25\%$ of the BATSE bursts in these samples fail the test
355: and hence they are outliers to the EpEi relation. We stress that
356: this fraction is only a lower limit as bursts that pass the test may
357: still not satisfy the EpEi relation, once their redshift is known.
358: These results imply that the EpEi relation, in its current form, may
359: not a generic property of GRBs. It is present only in the small
360: sample of bursts with confirmed redshifts and not in the whole
361: sample of observed bursts.
362:
363: None of the outliers we find has an isotropic energy larger than the
364: one predicted by the EpEi relation. Truly, our test could not find
365: such bursts. However, the two known outliers have lower isotropic
366: energy than the one predicted by the EpEi relation. Moreover,
367: already the BATSE data demonstrated the absence of soft and bright
368: bursts. The absence of such bursts is confirmed by BeppoSAX and HETE
369: II which would have easily detected and localized them. Thus, we
370: suggest that the common EpEi relation should be replaced by an EpEi
371: inequality:
372: \begin{equation}
373: E_{iso} \lesssim A_k E_p^k \ .
374: \end{equation}
375:
376: The natural question that arises is why there are so many
377: outliers in the BATSE data while there are only two outliers to
378: the EpEi relation in the current sample of bursts with confirmed
379: redshifts? One possibility is that there are systematic errors.
380: Since $d_2 \propto E_{p,obs}^2$, if for some reason $E_{p,obs}$ of
381: all the BATSE bursts is overestimated by a factor of $\gtrsim 2$
382: or if it is underestimated by the same factor for BeppoSAX and
383: HETE II bursts, then BATSE sample may be consistent with an EpEi
384: relation. The other possibility is that the difference between
385: BATSE data and the current sample of bursts with confirmed
386: redshifts results from an observational selection effect
387: (Lloyd-Ronning \& Ramirez-Ruiz 2002). This explanation is
388: supported by the fact that both outliers were not localized in the
389: usual manner by either BeppoSAX or HETE II whose localized bursts
390: compose the localized bursts sample. The first, 980425, was
391: detected and localized by BeppoSAX. However, if it was not for the
392: discovery of SN 1998bw (Galama et al., 1998) the identification of
393: its host galaxy and the measurement of its redshift would have
394: remained questionable. The second outlier, 031203 was localized by
395: INTEGRAL (Sazonov, Lutovinov \& Sunyaev 2004). Observational
396: selection affects might play a complicated roll especially since
397: the threshold for redshift measurement might be higher than the
398: threshold for detection. This is intuitively clear as the redshift
399: determination requires not only a detection of the prompt emission
400: but also a fast localization and an afterglow detection.
401:
402: Our results suggest that Swift, which is expected to reduce the
403: threshold for detection, localization and afterglow detection,
404: will detect dim and hard bursts that do not satisfy the EpEi
405: relation. It is impossible, however, to quantify this prediction
406: without a clear understanding of the threshold for redshift
407: measurement. Moreover, this second threshold would have to be
408: understood in order to use the coming sample of Swift bursts with
409: known redshifts to study the relation between $E_p$ and $E_{iso}$,
410: or other intrinsic properties of the GRB population.
411:
412: \vspace{0.25in}
413:
414: This work was supported by a BSF grant to the Hebrew University. We
415: thank Enrico Ramirez-Ruiz, Eli Waxman, Dafne Guetta, Amir Levinson
416: and Re'em Sari for helpful discussions. We thank the anonymous
417: referee for his detailed and constructive report.
418:
419: \begin{thebibliography}{99}
420: \bibitem[]{} Amati, L., Frontera, F., Tavani, M., in’t Zand, J. J. M.,
421: Antonelli, A. et al. Astr. Astrophys. 390, 81–89 (2002).
422:
423: \bibitem[]{} Atteia, J.-L., Ricker,
424: G.~R., Lamb, D.~Q., Sakamoto, T., Graziani, C., Donaghy, T.,
425: Barraud, C., \& The Hete-2 Science Team 2004, AIP Conf.~Proc.~727:
426: Gamma-Ray Bursts: 30 Years of Discovery, 727, 37
427:
428: \bibitem[]{} Band, D., et al.\ 1993,
429: ApJ, 413, 281
430:
431: \bibitem[Band \& Preece(2005)]{2005astro.ph..1559B} Band, D.~L., \& Preece,
432: R.~D.\ 2005, ArXiv Astrophysics e-prints, astro-ph/0501559
433:
434: \bibitem[]{} Galama, T.~J., et al.\
435: 1998, Nature, 395, 670
436:
437: \bibitem[Ghirlanda et al.(2004)]{2004ApJ...616..331G} Ghirlanda, G.,
438: Ghisellini, G., \& Lazzati, D.\ 2004a, ApJ, 616, 331
439:
440: \bibitem[Ghirlanda et al.(2004)]{2004A&A...422L..55G} Ghirlanda, G.,
441: Ghisellini, G., \& Celotti, A.\ 2004b, A\&A, 422, L55
442:
443:
444: \bibitem[]{} Jimenez, R.,
445: Band, D., \& Piran, T.\ 2001, ApJ, 561, 171
446:
447:
448: \bibitem[]{}Lamb, D. Q., Donaghy, T. Q. \& Graziani, C. A New Astronomy
449: Review 48, 459–464 (2004).
450:
451: \bibitem[]{}
452: Sakamoto, T., Lamb, D. Q., Graziani, C., et al., 2004, submitted to
453: Ap. J., astro-ph0409128
454:
455: \bibitem[Lloyd et al.(2000)]{2000ApJ...534..227L} Lloyd, N.~M., Petrosian,
456: V., \& Mallozzi, R.~S.\ 2000, ApJ, 534, 227
457:
458:
459: \bibitem[Lloyd-Ronning \& Ramirez-Ruiz(2002)]{2002ApJ...576..101L}
460: Lloyd-Ronning, N.~M., \& Ramirez-Ruiz, E.\ 2002, ApJ, 576, 101
461:
462:
463: \bibitem[]{} Sazonov, S.~Y., Lutovinov, A.~A., \& Sunyaev, R.~A.\ 2004,
464: Nature, 430, 646
465:
466: \bibitem[]{} Stern, B.~E.,
467: Tikhomirova, Y., Stepanov, M., Kompaneets, D., Berezhnoy, A., \&
468: Svensson, R.\ 2000, ApJ, 540, L21
469:
470:
471:
472: \end{thebibliography}
473:
474:
475: \end{document}
476: