1: \documentclass[apjl]{emulateapj}
2: %\documentclass[12pt,preprint]{aastex}
3:
4: % \usepackage{psfig,amsfonts,amsmath,graphicx,natbib,apjfonts}
5: \usepackage{psfig,amsfonts,amsmath,graphicx,natbib}
6: \citestyle{aa}
7:
8: \def\ra#1#2#3{#1$^{\rm h}$#2$^{\rm m}$#3$^{\rm s}$}
9: \def\dec#1#2#3{$#1^\circ#2'#3''$}
10: \def\swift{{\it Swift}}
11: \def\nod{\nodata}
12: \def\eiso{\ifmmode {E_{iso}}\else
13: ${E_{iso}}$\fi}
14: \def\pc{\ifmmode {\rm pc}\else
15: pc\fi}
16: \def\sfr{\ifmmode \dot{{\rm M}}_\star \else
17: $\dot{{\rm M}}_\star$\fi}
18: \def\mstar{\ifmmode {\rm M}_\star \else
19: ${\rm M}_\star$\fi}
20: \def\msun{M_\odot}
21:
22: %\def\jdot{ \dot{\,J}\! }
23: \def\jdot{ \dot{J} }
24: \def\erg{{\rm erg}}
25: \def\cm{{\rm cm}}
26: \def\De{{\cal D}_{est}}
27:
28: \begin{document}
29:
30: \title{Linking Short Gamma Ray Bursts and their Host Galaxies}
31:
32: \author{
33: James E. Rhoads\altaffilmark{1}
34: }
35:
36: \altaffiltext{1}{Arizona State University}
37: \begin{abstract}
38: %Short duration gamma ray bursts (SGRBs) remain the least understood
39: %class of cosmic explosion. While we can measure their durations and
40: %spectra, and sometimes also their redshifts and isotropic equivalent
41: %energies, these properties are decoupled from many important details
42: %of the progenitor objects under the fireball model of GRBs. To learn
43: %about the progenitors, we can study the environments of SGRBs, and in
44: %particular, the properties of their host galaxies. Here we show that
45: The luminosities of SGRB host galaxies are anticorrelated with both
46: the isotropic equivalent gamma ray energy and the gamma ray luminosity
47: of the explosions. Observational selection effects only strengthen
48: the significance of this correlation. The correlation may indicate
49: that there are two physically distinct groups of SGRBs. If so, it
50: requires that the more luminous class of explosions be associated with
51: the younger class of progenitors. Alternatively, it could be due to a
52: continuous distribution of burst and host properties. As one possible
53: explanation, we find that the effect of binary neutron star masses on
54: inspiral time and energy reservoir produces a correlation of the
55: appropriate sign, but does not automatically reproduce the correlation
56: slope or the full range of SGRB energy scales. Any future model of
57: SGRB progenitors needs to reproduce this correlation.
58: \end{abstract}
59:
60: \keywords{gamma-rays:bursts}
61:
62: \section{Introduction}
63: \label{sec:intro}
64: The host galaxies of short gamma ray bursts (SGRBs)
65: exhibit a wide range of physical properties.
66: SGRBs have been found not only in star-forming
67: hosts, but also in elliptical galaxies with strong
68: upper limits on their star formation rate (hereafter ``quiescent
69: hosts'') (Prochaska et al 2006; Berger 2008).
70: In contrast, long gamma ray bursts
71: (LGRBs) are found exclusively in star forming galaxies
72: (e.g., Le Floc'h et al 2003, Fruchter et al 2008, Savaglio et al 2009).
73: Several lines of evidence now link LGRBs
74: with the deaths of massive stars (Hjorth et al 2003,
75: Stanek et al 2003, Fruchter et al 2006, Raskin et al 2008).
76: The origin of the SGRBs
77: is not firmly established, but suspicion centers on merging neutron
78: star binaries, or other phenomena primarily associated
79: with neutron stars (Belczynski et al 2006; Chapman, Priddey,
80: \& Tanvir 2008). Hosts offer one of the best tools for unveiling
81: the nature of the GRB progenitors, because the properties of the gamma
82: ray emission and afterglow are decoupled from many important details
83: of the progenitor objects under the fireball model of GRBs (e.g.
84: Paczy\'{n}ski \& Rhoads 1993, Katz 1994, Piran 2005).
85:
86: The situation is similar to supernovae, which divide into two
87: groups, the type Ia and the core collapse events (which
88: include types II, Ib, and Ic).
89: Core collapse events are found only in star forming galaxies,
90: and are associated with the deaths of stars having initial
91: masses $\ga 8 \msun$. In contrast, type Ias are found in both quiescent
92: and star forming hosts, and are thought to be powered by the nuclear
93: explosion of a white dwarf whose mass is pushed above the Chandrasekhar
94: limit by accretion. However, despite having a characteristic progenitor
95: mass set by fundamental physics, the Ia supernovae are not entirely
96: homogeneous in their properties. Those found in star forming
97: hosts are both more frequent (per unit stellar mass) and more luminous
98: than their cousins in quiescent galaxies (Hamuy et al 1995;
99: Scannapieco \& Bildsten 2005; Sullivan et al 2006).
100:
101: In this {\it Letter}, we look for similar trends linking the
102: propertie of SGRBs and their hosts. We find an anticorrelation
103: between SGRB isotropic energy and host galaxy luminosity. We find
104: no obvious selection effects that
105: could produce such an effect, nor do we find evidence for any similar
106: effect in a larger control sample of LGRBs.
107:
108: % The remainder of this letter is organized as follows: in
109: % section~2, we introduce the samples we use. In section~3,
110: % we present the relations between isotropic burst energy \eiso\ and
111: % host galaxy absolute magnitude, and between GRB gamma ray luminosity
112: % and the specific star formation rate of the host galaxy. In section~4,
113: % we discuss possible interpretations of the resulting correlations.
114:
115:
116: \section{Observations}
117: % \subsection{The Samples}
118: We take our SGRB sample from the compilation by Berger (2008;
119: hereafter B08) of all SGRBs localized by the X-ray telescope (XRT)
120: on the Swift satellite (Gehrels et al 2004).
121: This sample contains 23 SGRBs, of which 12
122: have reported redshifts. These 12 form the core sample for our
123: study. They are further divided into 6 with optical
124: afterglows, and hence the most accurate positions and most secure host
125: galaxy identifications (``sample 1'' of B08), and 6 with somewhat less
126: accurate positions based X-ray data alone (``sample 2'' of B08).
127: The two subsamples are similar in most properties (B08).
128:
129: We also form a comparison sample of 34 long bursts, by
130: combining host galaxy information from Savaglio et al (2008) with
131: burst durations from the GCN circulars and other GRB properties
132: from Amati et al (2008).
133:
134: % \subsection{Analysis}
135: We calculate host galaxy absolute magnitudes as
136: $B_{abs} = R - 5 \log(d_L / 10 \pc) + 2.5 \log(1+z)$. Here
137: R is the observed R-band magnitude from B08, and
138: $d_L$ is the luminosity distance,
139: calculated assuming $H_0 = 71 \hbox{km} \hbox{s}^{-1}
140: \hbox{Mpc}^{-1}$, $\Omega_{0,m} = 0.27$, and $\Omega_{0,\Lambda} =
141: 0.73$. The resulting absolute magnitude corresponds to
142: rest wavelength $6500 \hbox{\AA} / (1+z)$. For
143: our SGRB sample, this always between U and R band, and typically
144: near B band (hence our notation).
145: %
146: We calculate the isotropic equivalent gamma ray energy
147: of the bursts directly from their fluence $f$ as
148: $\eiso = 4\pi d_L^2 f / (1+z)$.
149:
150: \section{Results}
151: The isotropic-equivalent energy of the SGRBs is correlated
152: with the absolute magnitudes of their host galaxies, with a
153: correlation coefficient of 0.60. That is,
154: the brightest bursts tend to occupy the least luminous hosts, and vice
155: versa (fig.~\ref{fig:eiso_absmag}a). This is {\it not} expected from
156: simple observational selection effects. If the samples are largely
157: limited by flux and/or fluence, with most objects near the
158: detection threshold, the range of distances in the sample should then
159: lead to a {\it positive} correlation of burst energy with host
160: luminosity. Such a positive correlation {\it is} seen for our
161: control sample of long GRBs (figure~\ref{fig:eiso_absmag}b).
162:
163: The least luminous SGRBs are the closest, while the
164: least luminous hosts span a wide range of
165: redshifts. Gamma ray sensitivity limits inclusion in our sample
166: more critically than does sensitivity to the host galaxy's starlight.
167: If the host is not detected in the first relevant observation,
168: more data can be taken, but if the GRB is not detected in the
169: first observation, it can never enter the sample.
170:
171: If a low-luminosity event such as GRB 050509B occurred in
172: a subluminous host at $z \la 0.5$, we should still see
173: it. We see perhaps one such event, GRB 070724, which has
174: low isotropic energy and luminosity, but lives in a star
175: forming host galaxy.
176: %
177: We would also expect that if a highly luminous SGRB occurred
178: in a large elliptical host galaxy, we should be able
179: to observe such an event (and its luminous host) out to $z\ga 0.8$.
180: We see no such events. This suggests that
181: the more luminous SGRBs are associated with young progenitors
182: of some type that is absent in old elliptical galaxies.
183:
184:
185: We have tested the significance of the correlation in
186: figure~\ref{fig:eiso_absmag}a using a Spearman rank correlation
187: test. The test yields a rank correlation coefficient of $0.50$
188: and a significance level of $t=1.84$ with 10 degrees of
189: freedom. Formally, this corresponds to a 95\% significance.
190:
191: However, the true significance is higher, since observational
192: selection effects work to produce the opposite sign of
193: correlation. We have performed simple
194: simulations of the correlation produced by
195: observational selection in the absence of any true, physical
196: relation between the burst and host properties.
197: We first constructed independent probability distributions for
198: $\eiso$, $B_{abs}$, and $z$. To these,
199: we added selection probabilities based on gamma-ray
200: fluence and host apparent magnitude. Each of
201: these probability distributions was specified by a simple,
202: plausible functional form. The functions were adjusted until
203: the simulations were able to reproduce separately the 1D marginal
204: distributions of $f$, $\eiso$, $R$, $B_{abs}$, and $z$
205: seen in the data (all within the Poisson uncertainties).
206:
207: In each simulation, we constructed a ``parent
208: sample'' of bursts, and applied the
209: selection probabilities to simulate a ``selected sample,'' whose
210: size is constrained to equal that of the observed sample.
211: We then measure the correlation coefficient between \eiso\
212: and $B_{abs}$ in the simulation. We repeat this procedure
213: many times, and compare the distribution of simulated correlation
214: coefficients to that measured in the observed
215: sample.
216:
217: Among $10^4$ simulations, only 121 have coefficients above the observed
218: value. Thus, by considering selection effects, we estimate the
219: significance of the SGRB energy - host luminosity anticorrelation
220: at about 98.8\%.
221:
222: While the simulations include several {\it ad hoc} functional
223: forms, they capture the essential points of any realisitic model
224: that does {\it not} have correlations between burst and host
225: properties. First, the bursts and hosts span a wide range
226: of intrinsic brightness. Second, the rate as a function of
227: redshift could be matched by the volume element modified by some
228: modest amount of rate evolution. Third, the observational selection
229: functions are monotonically decreasing over the range from the
230: brightest to the faintest observed SGRBs and hosts.
231: Fourth, the selection on the SGRB fluence excludes many more simulated
232: events than does the selection on host luminosity. This induces
233: a relatively weak correlation between host and burst brightness
234: in the simulated samples. More strongly peaked intrinsic properties,
235: or a more even balance between rejection by either GRB or host
236: properties, would tend to induce stronger correlations in the
237: simulated sample. Thus, plausible changes
238: in our simulations should not strongly affect our
239: conclusions.
240:
241: Our results do depend substantially on GRB 050509b: If we
242: exclude it from the sample, the remaining 11 observed
243: points yield a correlation coefficient of 0.28. This corresponds to
244: about a 14\% chance in the simulations.
245: For comparison, a hypothetical sample of about 40 SGRBs with a
246: true correlation coefficient 0.28 would be significant at
247: the same the 99\% level as our full sample.
248: %
249: Provided that the identification of the GRB 050509b host is
250: regarded as secure, the correlation does not depend
251: critically on other bursts without optical transients. If we use
252: the subset of SGRBs with optical transients (B08's ``sample 2'')
253: plus GRB 050509b (which has no OT but still a rather secure
254: host detection), the correlation coefficient becomes 0.63,
255: and the simulations show a significance level of 95.6\%.
256:
257: For reference, our control sample of LGRBs shows a correlation
258: coefficient of -0.54 between isotropic energy and host absolute
259: magnitude. We adapted our simulations to reproduce the marginal
260: distributions of $z$, $R$, $B_{abs}$, and so forth for the LGRB
261: sample, and again compared the distribution of correlation
262: coefficients to that observed for the LGRBs. Formally, this too
263: yields a difference at the 99\% level between the simulations and the
264: data, here in the sense that the data show a stronger {\it positive}
265: correlation between GRB and host brightness than do the simulations.
266: Because the sign of the correlation matches the sign of the selection
267: effects, and because the correlation is driven substantially by the
268: X-ray flashes (020903 and 060218) at the bottom of the LGRB \eiso\
269: distribution, we do not attach great physical significance to this
270: result yet. However, it is intriguing and may merit further
271: investigation in future.
272:
273: \begin{figure}
274: %Journal format: \plottwo{f1a.eps}{f1b.eps}
275: \centering
276: \includegraphics[angle=0,scale=0.49]{f1a.eps}
277: \includegraphics[angle=0,scale=0.49]{f1b.eps}
278: \caption{
279: The relation between isotropic equivalent gamma-ray energy and host
280: galaxy absolute magnitude for short GRBs (first panel) and long GRBs
281: (second panel). The
282: observed correlation in the SGRB sample is significant at about
283: the 99\% level (see text).
284: The control sample of LGRBs shows a correlation
285: consistent with that expected from observational selection. SGRB data
286: points are labelled with the GRB ID number. Redshift ranges are denoted
287: by color. For SGRBs, blue means $z<0.3$, green
288: means $0.3<z<0.7$, and red means $z>0.7$. For LGRBs, cyan means $z<0.3$,
289: blue means $0.3<z<1.0$, green means $1.0<z<2.0$, red
290: means $2.0<z<3.5$, and magenta means $z>3.5$.
291: \label{fig:eiso_absmag}}
292: \end{figure}
293:
294: We also examined the relation between SGRB gamma ray luminosity
295: and host galaxy star formation
296: rate per unit stellar mass. Because the two most luminous SGRB hosts
297: are early type galaxies, with tight upper limits on their emission-line
298: derived SFR (Prochaska et al 2006, Berger 2008,
299: Savaglio et al 2008), this plot clearly separates these two
300: SGRBs from the rest of the sample. The specific star formation
301: rate is defined as $s = \sfr / \mstar$, where $\sfr$ is the star
302: formation rate, and $\mstar$ the stellar mass of the galaxy. However,
303: lacking the data to fit an accurate stellar mass to some galaxies,
304: we use the quantity $\log({\cal S}) \equiv \log{\sfr} + 0.4
305: B_{abs} = \log{(s)} + \log{(\mstar / L_{host})}$. To
306: demonstrate the utility of ${\cal S}$, we compared it to the
307: published $s$ (from Savaglio et al 2008) for the 34 LGRB
308: hosts from our control sample.
309: We find $\log({\cal S}) \approx 0.4 \log(s \times \hbox{Gyr}) - 7.9$,
310: with a scatter of $0.3$ dex (RMS).
311:
312: The results (fig.~\ref{fig:ssfr_lgam}a) show that the
313: separation of the SGRB hosts into actively star forming galaxies
314: and old stellar populations is accompanied by a
315: separation of burst gamma-ray luminosity, with only the least luminous
316: bursts found in the old hosts. The corresponding plot for
317: LGRBs (fig.~\ref{fig:ssfr_lgam}b) shows no correlations beyond
318: the sample selection requirement that the least luminous events
319: be reasonably nearby. A comparison of the two figures also shows
320: the tendency (noted previously by, e.g., B08) for even star-forming
321: SGRB hosts to have less vigorous star formation (i.e. lower ${\cal S}$)
322: than do LGRB hosts, and also the tendency for SGRBs to have lower
323: gamma ray fluxes.
324:
325: \begin{figure}
326: %Journal format: \plottwo{f2a.eps}{f2b.eps}
327: \includegraphics[angle=0,scale=0.49]{f2a.eps}
328: \includegraphics[angle=0,scale=0.49]{f2b.eps}
329: \caption{The relation between GRB luminosity
330: and the specific star formation rate of the host galaxy, for
331: short (first panel) and long (second panel) GRBs.
332: The least luminous SGRBs are the two in massive early-type
333: host galaxies. While some bursts in star forming hosts
334: appear comparably faint (i.e., GRB 070724), we have yet to
335: observe a bright burst in a quiescent host.
336: For the long GRBs, there is no visible correlation of
337: $L_\gamma$ and specific star formation rate.
338: Point labels follow the same definitions as in figure~1.
339: \label{fig:ssfr_lgam}}
340: \end{figure}
341:
342:
343: \section{Discussion}
344: While our sample is small, the
345: anticorrelation between SGRB $\eiso$
346: and host luminosity is probably real, based on our Monte Carlo
347: simulations. The similar anticorrelation between brightness of
348: type Ia supernovae and
349: the stellar population ages of their host galaxies was first described
350: on the basis of a similarly small sample (13 supernovae; Hamuy et al
351: 1995).
352: %
353: We now examine possible interpretations for this
354: result.
355:
356: % \subsection{Two population models}
357: Suppose, first, that two distinct mechanisms produce
358: SGRBs. Such a situation would arise
359: naturally if two or more of the different proposed progenitors
360: actually do produce SGRBs. Combinations that have been explored
361: include coalescence of double neutron star binaries and
362: of NS-black hole binaries (Belczynski et al 2006; O'Shaughnessy et al 2008;
363: Troja et al 2008), and combinations of NS-NS binaries
364: with giant flares from magnetars (Chapman, Priddey, \& Tanvir 2008).
365: Additionally, a range of SGRB
366: properties might arise from NS-NS binaries alone, provided those
367: binaries evolved through a range of different binary interaction
368: histories (Belczynski et al 2006; Salvaterra et al 2008).
369:
370: If one mechanism is associated with stars younger than 1 Gyr,
371: and the other with ancient populations (age $\sim H_0^{-1}$),
372: we can explain the properties of figures~\ref{fig:eiso_absmag}a
373: and \ref{fig:ssfr_lgam}a.
374: The fraction of old-progenitor SGRBs occurring in
375: quiescent galaxies should match the fraction of stellar mass
376: in quiescent galaxies. The present,
377: limited data shows two bursts in
378: quiescent hosts, and a third similarly low-luminosity burst (070724)
379: in a star-forming host galaxy. This matches well the fraction of
380: stellar mass in elliptical galaxies at low redshift, which is 54\% to
381: 60\% (Baldry et al 2004).
382: Conversely, the fraction of more luminous SGRBs occurring
383: in quiescent hosts is $0/9$ in the present sample
384: (fig.~\ref{fig:ssfr_lgam}a). Under a two-population scenario,
385: the mechanism responsible for the more luminous SGRBs
386: should have a negligible rate for ages exceeding a few Gyr.
387:
388: % \subsection{Continuous distribution models}
389: If instead there is a single, continuous distribution of SGRB
390: and host properties, it suggests that the SGRB progenitor's
391: properties vary systematically with either age or metallicity,
392: which vary systematically along the Hubble sequence and
393: have direct effects on stellar-scale physics.
394:
395: If SGRBs are due to binary neutron star
396: inspiral events, an anticorrelation of SGRB
397: energy and the specific star formation rate of the host
398: galaxy could come from the dependence of inspiral time and
399: available energy on mass. Gravitational radiation gives
400: %% an angular momemtum loss rate
401: %% \begin{equation}
402: %% { \jdot \over J} = - {32 G^3 \over 5 c^5} {M_1 M_2 (M_1 + M_2) \over
403: %% a^4}
404: %% \end{equation}
405: %% (e.g., Landau \& Lifschitz 1958).
406: %% Here $J$ and $\jdot$ are the angular momentum of the system and
407: %% its time derivative, $M_1$ and $M_2$ are the masses of the two bodies,
408: %% $a$ is the semimajor axis of their orbit, and $G$ and $c$ are Newton's
409: %% constant and the speed of light. Coupled with the angular momentum
410: %% $J = M_1 M_2\times \sqrt{G a / (M_1+M_2)}$, this can be integrated to find
411: an inspiral time of
412: \begin{equation}
413: t_{inspiral} = {5 c^5 \over 256 G^3} {a_0^4
414: \over M_1 M_2 (M_1 + M_2)} ~
415: \end{equation}
416: (e.g., Landau \& Lifschitz 1958), where
417: $M_1$ and $M_2$ are the masses of the two bodies,
418: $a$ is the semimajor axis of their orbit, and $G$ and $c$ are Newton's
419: constant and the speed of light.
420: Thus, if a star formation event yields a population of compact object
421: binaries with a range of mass, we expect a range of inspiral times.
422: The SGRBs occurring in quiescent hosts would be those with low masses and/or
423: wide initial separations. The available energy reservoir is
424: $E \sim G M_1 M_2 / R_{final}$, where $R_{final}$
425: is the characteristic size of the system when the GRB energy
426: is released. If the SGRB event produces a black hole,
427: we expect $R_{final} \sim R_{grav} \propto (M_1+M_2)$,
428: so that $E\sim M_1 M_2 c^2 / (M_1 + M_2)$. Then, for $M_1 \approx M_2$,
429: $t_{inspiral} \sim a_0^4 E^{-3}$.
430: Characteristic stellar ages for SGRB hosts range
431: from $\sim 10$ Gyr for quiescent hosts, to
432: $\sim s^{-1} \sim 1$ Gyr for typical star forming SGRB hosts
433: (see, e.g., Savaglio et al 2008).
434: If this tenfold range of age corresponds to a tenfold range
435: of $t_{inspiral}$, it would imply a range of $\sim 2\times$ in
436: neutron star mass, which is comparable to the range of NS masses
437: believed to exist in our Galaxy.
438:
439: However, the corresponding $2\times$ range of energy
440: cannot explain the range of \eiso\ observed in
441: figure~\ref{fig:eiso_absmag}a. A larger range in SGRB energetics
442: might follow if $R_{final} \sim R_{NS}$, especially for a
443: neutron star equation of state that gives $d R_{ns} / dM < 0$.
444: Moreover, $a_0$ may correlate with $M_1$ and $M_2$ in some
445: complex way depending on on both binary star mass transfer and
446: detailed supernova physics. While a more thorough understanding
447: of the SGRB mechanism is clearly required before we can fully
448: model the relation between host and SGRB properties, the existence
449: of such a relation will provide valuable clues to the nature of
450: the short GRBs.
451:
452: The correlation in figure~\ref{fig:eiso_absmag}a can also
453: provide a tool for estimating SGRB redshifts.
454: $\log(\eiso)$ and $B_{abs}$
455: are related empirically by
456: $\log(\eiso / \erg) \approx 50 + 0.725 (B_{abs} + 20.2)$.
457: Defining ${\cal D} = d_L / \sqrt{1+z}$, so that
458: $\eiso = (4\pi {\cal D}^2) f$ and
459: $B_{abs} = R_{AB} - 5 \log[({\cal D} / 10pc)]$,
460: we can substitute for $\eiso$ and $B_{abs}$
461: to obtain a distance estimate:
462: \begin{equation}
463: \log(\De / \hbox{Gpc}) =
464: 0.012 + 0.129 (R_{AB}-20) - 0.177 \log(f_{-6}) ~~,
465: \label{eq:dist_est}
466: \end{equation}
467: where $f_{-6}$ is the SGRB fluence in units of $10^{-6} \erg \, \cm^{-2}$.
468: The conversion from $\De$ to the estimated distance $d_{L,est}$ and redshift
469: $z_{est}$ must of course follow the cosmology we used to derive
470: equation~\ref{eq:dist_est}.
471: %
472: For our primary sample of 12 SGRBs, the rms residual in
473: $\log(\De)$ is $0.17$ dex, and in
474: redshift $z(\De)$ is 0.147.
475: The residuals from the fit substantially exceed the uncertainties
476: expected from fluence and host magnitude measurements, implying that
477: most of this scatter is intrinsic.
478:
479: As an alternative, we can directly fit a relation of the form
480: $\log(z_{est}) = \log(z_0) + a (R_{AB} - 20) + b \log(f_{-6})$
481: by minimizing $\Delta z_{rms} =
482: \langle (z_{est} - z_{obs})^2 / (N-1)\rangle^{1/2}$.
483: This approach gives $\log(z_{est}) = -0.56 + 0.107 (R_{AB}-20) - 0.14
484: \log(f_{-6})$, with $\Delta z_{rms} = 0.142$.
485: This redshift estimator is easier to apply, and is formally independent
486: of the adopted cosmological parameters, though it is less directly linked
487: to whatever physical mechanism links $\eiso$ and $B_{abs}$. Differences
488: between the predicted redshifts from the two equations are all much
489: smaller than the scatter in either.
490:
491: % \subsection{Other possibilities}
492: Another possibility is that the external medium around
493: SGRBs plays some role in their luminosity. The interstellar medium is
494: typically less dense in elliptical galaxies than in actively star
495: forming galaxies. However, this should only affect the
496: observed flux from external shocks that GRB ejecta drive into the
497: ambient medium. External shocks are believed to power the afterglow
498: emission, but not the GRB luminosity itself (e.g., Piran 2005). Since
499: our observed anticorrelation involves the SGRB $\gamma$-ray flux itself,
500: we disfavor this explanation.
501:
502: % \subsection{Summary}
503: To conclude, we have found an anticorrelation between the energy of a
504: short gamma ray burst and the luminosity of its host galaxy. Such an
505: anticorrelation occurs with a probability of $\sim 1\%$ in simulations that
506: account for observational selection effects, and
507: so is probably real. Its physical origin
508: is unclear, though it is most likely due to a correlation between the
509: age of an SGRB progenitor and the luminosity of the explosion. If
510: this correlation is a continuous distribution, it can provide
511: approximate redshift estimates for SGRBs. Time will tell
512: whether this correlation holds for a larger sample,
513: and whether it is a property of a single
514: distribution, or if it instead reflects an underlying division of
515: SGRBs into two physically distinct sets.
516:
517: \acknowledgements
518: I thank Sangeeta Malhotra, Evan Scannapieco, Patrick Young, Sumner
519: Starrfield, and Frank Timmes for stimulating discussions;
520: Sandra Savaglio and Edo Berger for their published host galaxy
521: compilations; R. Quimby, E. McMahon, and J. Murphy
522: for the GRBlog database at
523: http://grad40.as.utexas.edu/grblog.php ; and the Max
524: Planck Institute for Astronomy for hospitality
525: during completion of this work.
526:
527:
528: %\bibliographystyle{apj}
529: %\bibliography{journals_apj,refs,refs2,refs3}
530:
531: \begin{thebibliography}{}
532:
533:
534: \bibitem[\protect\citeauthoryear{{Amati} et~al.}{{Amati} et~al.}{2008}]{amati08}
535: {Amati}, L., {Guidorzi}, C., {Frontera}, F., {Landi}, R.,
536: \& {Montanari}, E. 2008, submitted to \mnras, arXiv 0805.0377
537:
538: \bibitem[\protect\citeauthoryear{{Baldry} et~al.}{{Baldry}
539: et~al.}{2004}]{baldry04}
540: {Baldry}, I. K.., et~al. 2004, \apj, 600, 681
541:
542:
543: \bibitem[\protect\citeauthoryear{{Belczynski} et~al.}{{Belczynski}
544: et~al.}{2006}]{belc06}
545: {Belczynski}, K., Perna, R., Bulik, T., Kalogera, V., Ivanova, N.,
546: \& Lamb, D. Q. 2006, \apj, 648, 1110
547:
548: \bibitem[\protect\citeauthoryear{{Berger}}{{Berger}}{2008}]{B08}
549: {Berger}, E. 2008, submitted to \apj, arXiv 0805.0306
550:
551:
552: \bibitem[\protect\citeauthoryear{{Chapman}, {Priddey}, \& {Tanvir}}
553: {{Chapman} et~al.}{2008}]{chap08}
554: {Chapman}, R., {Priddey}, R. S., \& {Tanvir}, N. 2008, submitted
555: to \mnras, arXiv 0802.0008
556:
557: \bibitem[\protect\citeauthoryear{{Fruchter} et~al.}{{Fruchter}
558: et~al.}{2006}]{fls+06}
559: {Fruchter}, A.~S., et~al. 2006, \nat, 441, 463
560:
561:
562: \bibitem[\protect\citeauthoryear{{Gehrels} et~al.}{{Gehrels}
563: et~al.}{2004}]{gehrels04}
564: {Gehrels}, N., et~al. 2004, \apj, 611, 1005
565:
566: \bibitem[\protect\citeauthoryear{{Hamuy} et~al.}{{Hamuy}
567: et~al.}{2006}]{pbc+06}
568: {Hamuy}, M., Philips, M. M., Maza, J., Suntzeff, N. B.,
569: Schommer, R. A., \& Aviel\'{e}s, R. 1995, \aj, 109, 1
570:
571: \bibitem[\protect\citeauthoryear{{Hjorth} et~al.}{{Hjorth}
572: et~al.}{2005}]{hjorth03}
573: {Hjorth}, J., et~al. 2003, \nat, 423, 847
574:
575: \bibitem[\protect\citeauthoryear{{Katz}}{{Katz}}{1994}]{katz94}
576: {Katz}, J. 1994, \apj, 422, 248
577:
578: \bibitem[\protect\citeauthoryear{{Kouveliotou} et~al.}{{Kouveliotou}
579: et~al.}{1993}]{kou93}
580: {Kouveliotou}, C., et~al. 1993, \apjl, 413, L101
581:
582:
583: \bibitem[\protect\citeauthoryear{{Landau} \& {Lifshitz}}
584: {{Landau} \& {Lifshitz}}{1958}]{ll58}
585: Landau, L. D., \& Lifshitz, {\it The Classical Theory of Fields},
586: London: Butterworth-Heinemann, 1958 (4th edition: 1980).
587:
588:
589: \bibitem[\protect\citeauthoryear{{Le Floc'h} et~al.}{{Le Floc'h}
590: et~al.}{2003}]{ldm+03}
591: {Le Floc'h}, E., et~al. 2003, \aap, 400, 499
592:
593: \bibitem[\protect\citeauthoryear{{O'Shaughnessy}, {Belczynski}, \&
594: {Kalogera}}{{O'Shaughnessy} et~al.}{2008}]{osh08}
595: {O'Shaughnessy}, R., {Belczynski}, K., \& {Kalogera}, V. 2008,
596: \apj, 675, 566
597:
598: \bibitem[\protect\citeauthoryear{{Paczy\'{n}ski} \& {Rhoads}}
599: {{Paczy\'{n}ski} \& {Rhoads}}{2007}]{pr93}
600: {Paczy\'{n}ski}, B., \& {Rhoads}, J.~E. 1993, \apjl, 418, L5
601:
602:
603: \bibitem[\protect\citeauthoryear{{Piran}}{{Piran}}{2005}]{piran05}
604: {Piran}, T. 2005, {\it Reviews of Modern Physics}, 76, 1143
605:
606: \bibitem[\protect\citeauthoryear{{Prochaska} et~al.}{{Prochaska}
607: et~al.}{2006}]{pbc+06}
608: {Prochaska}, J.~X., et~al. 2006, \apj, 642, 989
609:
610: \bibitem[\protect\citeauthoryear{{Raskin} et~al.}{{Raskin}
611: et~al.}{2006}]{pbc+06}
612: {Raskin}, C., Scannapieco, E., Rhoads, J. E., \& Della Valle, M.
613: 2008, submitted to \apj
614:
615: \bibitem[\protect\citeauthoryear{{Salvaterra} et~al.}{{Salvaterra}
616: et~al.}{2008}]{salva08}
617: {Salvaterra},R., Cerutti, A., Chincarini, G., Colpi, M.,
618: Guidorzi, C., \& Romano, P. 2008, submitted to \mnras,
619: arXiv 0710.3099
620:
621: \bibitem[\protect\citeauthoryear{{Savaglio}, {Glazebrook}, \& {Le
622: Borgne}}{{Savaglio} et~al.}{2008}]{sgl08}
623: {Savaglio}, S., {Glazebrook}, K., \& {Le Borgne}, D. 2008, submitted
624: to \apj, arXiv 0803.2718
625:
626: \bibitem[\protect\citeauthoryear{{Scannapieco} \& {Bildsten}}
627: {{Scannapieco} \& {Bildsten}}{2007}]{pr93}
628: {Scannapieco}, E., \& {Bildsten}, L. 2005, \apjl, 629, L85
629:
630: \bibitem[\protect\citeauthoryear{{Spergel} et~al.}{{Spergel}
631: et~al.}{2004}]{gehrels04}
632: {Spergel}, D. N., et~al. 2007, \apjsupp, 170, 377
633:
634: \bibitem[\protect\citeauthoryear{{Stanek} et~al.}{{Stanek}
635: et~al.}{2005}]{hwf+05}
636: {Stanek}, K., et~al. 2003, \apjl, 591, L17
637:
638: \bibitem[\protect\citeauthoryear{{Sullivan} et~al.}{{Sullivan}
639: et~al.}{2006}]{slp+06}
640: {Sullivan}, M., et~al. 2006, \apj, 648, 868
641:
642: \bibitem[\protect\citeauthoryear{{Troja} et~al.}{{Troja}
643: et~al.}{2006}]{pbc+06}
644: {Troja}, E.., King, A. R., O'Brien, P. T., Lyons, N., \&
645: Cusamano, G. 2008, \mnras, 385, L10
646:
647:
648: \end{thebibliography}
649: \end{document}
650: