50 Years of Gamma-Ray Bursts* * with a Biased Overemphasis on Neil & Stuff I Was Involved In

50 Years of Gamma-Ray Bursts* * With a biased overemphasis on Neil & stuff I was involved in

Josh Bloom UC Berkeley @profjsb IPN Vela Series CGRO (BATSE) Swift HETE-2 Fermi BeppoSAX INTEGRAL

High-Energy Era Afterglow Era Multimessenger Era

GRB 980425/SN 1998bw GRB 130603B GRB 090423 GRB 030329/SN 2003dh March 5 GRB 970508 GRB 080319b Events GRB 130427A Event Sw J1644+57 GRB 670702 (SGR 0525-66) GRB 970228 GRB 050509b GW/GRB 170817

1970 1980 1990 2000 2010 2020 Burrows+05 Colgate Paczyński Great Debate Frail+01 Berger 14 May 68 86 Sari, Piran, Narayan 98 Gehrels, 50 Years of Eichler+89 Mészáros & Rees 97 Ramirez-Ruiz, & Fox 09 GRBs Klebesadal, Gehrels Aggregate Insights Aggregate Band+93/Kouvelitou+93 Lamb & Reichart 00 Olson & Strong 73

& Memorial Paczyński & Rhoads 93 Meegan+92 Li & Paczyński 98 Theory MacFadyen & Woosley 99 Josh Bloom Discovery & Demographics High-energy Era

GRB 670702

Theory: Colgate 68 Discovery & Demographics High-energy Era

“short” “long” BATSE •Isotropic

SHBs LSBs • Non-Euclidean/ Inhomogeneous 1000 Meegan+92

• Two Populations

100 GRBs

XRRs

XRFs

Bloom10

Kouvelitou+93, Mazets+81 Norris+84

“No Host Problem” (cf. Larson 97)

“Great Debate” here in DC (Apr 95): Galactic or Cosmological?

Rees, Paczyński, Lamb https://apod.nasa.gov/debate/debate95.html Discovery & Demographics THE CORRECTED LOG N-LOG High-energy Era FLUENCE DISTRIBUTION OF COSMOLOGICAL v-RAY BURSTS

Joshua S. Bloom 1'2, Edward E. Fenimore 2, Jean in 't Zand 2'a

1Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 0P138 “No Host Problem” (cf. Larson 97) 2Los Alamos National Laboratory, Los Alamos, NM 875~ 3 Goddard Space Flight Center, Greenbelt, MD P0771 “Great Debate” here in DC (Apr 95): Recent analysis of relativistically expanding shells of cosmological Galactic or Cosmological? "c-ray bursts has shown that if the bursts are cosmological, then most likely total energy (E0) is standard and not peak luminosity (L0). Assuming a fiat Friedmann cosmology (qo = 1/2, A = 0) and constant rate density (p0) of bursting sources, we fit a standard candle energy to a uniformly selected log N-log S in the BATSE 3B catalog correcting 3rd Huntsville Conference (Oct 95) for fluence efficiency and averaging over 48 observed spectral shapes. We find the data consistent with E0 = 7.3+_°:0r x 1051 ergs and discuss implications of this energy for cosmological models of 7-ray bursts.

INTRODUCTION

On the basis of strong threshold effects of detectors, Klebesadel, Fenimore, and Laros (7) concluded that GRB fluence tests were largely inconclusive. As a result, nearly all subsequent number-brightness tests have used peak flux (P) rather than fluence (S). However, the standard candle peak luminosity assumption that is required by log N-log P studies is unphysical. If, for instance, bursts originateRees, at Paczycosmologicalński, distances Lamb and are produced by colliding neutron stars then one might expect that total energy would be standard and not peak luminosity. Moreover, recent analysis of relativistically expanding shellhttps://apod.nasa.gov/debate/debate95.html models has cast doubt on the standard L0 assumption (9). In this paper, we seek to eliminate the large threshold effects present in log N-log S studies by correcting the observed number of bursts at a given fluence by the trigger efficiency of the detector.

PVO CONSISTENCY CHECK

The Pioneer Venus Orbiter (PVO) had a peak flux trigger sampled on 0.25, 1.0, and 4.0 sec timescales and was sensitive to bursts down to fluxes of 5 × 10-6 erg cm -2. Despite a substantially lower fluence trigger sensitivity range, PVO saw hundreds more bright bursts than BATSE due the relatively long on-time and large sky-coverage of PVO. As the bright region of the BATSE log N-log (~) 1996 American Institute of Physics 321 at z ~ 0 especially if E0 is large. Therefore, for a given E0, Si, and ¢i(E) we first solve for the redshifts, zi, of the baseline events associated with each spectral shape. The standard// candle energy,¢ooo E0, is(.) given by, Eo = 4~rR~,= N(t,)dts E¢i dE (1) ,#30 where N(ts) is the normalization of the spectrum (ergs keV -1) at time t, at the source. The comoving distance, Ri,z, is defined in eq. [2] of ref. (2). The observed ith baseline burst fluence in the energy range 50-300 keV is,

-- abo3°° E¢i [rl+z,.E] l + zi J dE, (2)

where N(tob~) is the observed normalization of the spectrum. For a given standard candle energy, E0, we numerically determine the redshift (1 + zi) of the ith baseline burst using eqs. (1, 2) and letting zr = zi. Note that (1 + zi) f g(ts)dt, = f N(tobs)dtobs. Instead of assuming a spectral shape at the source, we use an average over baseline spectra to compute the number of expected observed bursts, ANexp[Sj to Sj+I] in some fluence range [Sj, Sj+I]:

NBAND fR(Sj+I ) P0 AN~xp[Sj to Sj+I]- 4~r ~ j~ e[Si(r)] r2dr. (3) NBAND .= R(Si) 1 + Zr

where NBAND ---- 48 is the number of baseline spectra used and po is the rate density of bursts per comoving volume. The quantity Si(r) is the predicted fluence (using eqs. [1, 2]) of the it__h baseline burst if it was at a distance r. This distance corresponds to a redshift 1 + zr. Discovery & Demographics We constructTABLE 11 fluence1. Best binsFit Distribution(in BATSE of channels E0 = 7.0 2+3× 1051 corresponding ergs to approximatelyTHE CORRECTED50-300Fluence keV) Ranges~(50-300 of roughly keV)equalLOG number AN[SjN-LOG of tobursts. S3+i ] We select High-energy Era burstsBin Number(j) with Cmin/Cmax Sj > 1 on eitherSj+I the 256 Observedor 1024 bms Predictedtimescale, 1then + zj find aFLUENCE minimized X2 between DISTRIBUTION the number of predicted burstsOF and observed 1 2.16e-07 3.82e-07 51 38.4 3.88 by varying E0. For 9 degrees of freedom we find an acceptable X2 = 14.7 COSMOLOGICAL2 3.82e-07 5.85e-07 v-RAY BURSTS42 50.5 3.24 corresponding to a standard candle E0 .....7 2 +0.7 a.0 × 1051 ergs. Table (I) gives 3 5.85e-07 7.55e-07 42 36.3 2.84 the bin ranges, number of observed bursts per bin, number of predicted bursts Joshua4 S. Bloom 7.55e-07 1'2, Edward 1.13e-06E. Fenimore 2, Jean46 in 't Zand 63.0 2'a 2.64 for the 5best fit energy, 1.13e-06 and their 1.43e-06implied redshifts. 37 36.8 2.36 1Harvard-Smithsonian6 1.43e-06 Center for 2.00e-06 Astrophysics, Cambridge,48 MA 0P138 49.7 2.22 “No Host Problem” (cf. Larson 97) 2Los Alamos National Laboratory, Los Alamos, NM 875~ 7 2.00e-06 CONCLUSIONS 2.80e-06 39 44.3 2.04 8 3 Goddard 2.80e-06 Space Flight Center, 4.05e-06 Greenbelt, MD 44P0771 41.1 1.89 9 4.05e-06 6.20e-06 37 37.2 1.74 Our fit of E0= 7 •0 +0.7-1.0x 1051 [30-2000 keV] ergs seems a plausible number “Great Debate” here in DC (Apr 95): 10 6.20e-06 1.36e-05 40 44.7 1.60 on the Recentbasis thatanalysis GRBs of relativistically last on the expandingaverage 10shells sec ofand cosmological L0 = 4.6 x 105° erg Galactic or Cosmological? 11 1.36e-05 6.60e-05 41 32.3 1.41 s -1 from"c-ray logbursts N-log has shown P studies that if (2).the burstsHowever, are cosmological, this E0 implies then most a rather large efficiencyIn ergslikely cmoftotal -2energy energy conversion (E0) is standard to 7-rays and (~not 10%)peak luminosityif the bursting (L0). mechanism bis BurstscollidingAssuming with neutron aCmin/Cmax fiat Friedmann stars >(Mtotal 1cosmology on the~ 2.8Mo).256 (qo or= 1/2,1024 Nevertheless, A =ms 0) timescale and constant this in resultBATSE would 3B. rate density (p0) of bursting sources, we fit a standard candle energy to seem a touniformly help resolve selected the log N-log"no-host" S in theproblem BATSE (cf. 3B catalogref (3)). correcting Interestingly, that 3rd Huntsville Conference (Oct 95) the dimmestfor fluence bursts efficiency (S -and 5 xaveraging 10 -s erg over cm 48 -2) observed are required spectral to shapes. be at a redshift of 1 + z We~_ 6.4find given the data this consistent E0, would with seem E0 = to7.3+_°:0 ruler xout 1051 several ergs and cosmological discuss models that implicationsrequire GRB of thisprogenitors energy for cosmologicalto be 324within models galaxies of 7-ray (although bursts. see reference (8)). This surprisingly high redshift is due to the correct blueshifting of the baseline spectra back to the source in eq. (1). If we neglect this factor, we obtain a smaller, more tenableINTRODUCTION redshift of the dimmest bursts (1 + z = 5.2). Whatever the conclusion about the models, we note two important results. First,On the the basis bend of instrong the thresholdlog N-log effects S curve of detectors,in BATSE Klebesadel, is real, not Fenimore, an artifact of andstrong Laros threshold (7) concluded effects. that This GRB implies fluence that tests we were are largely seeing eitherinconclusive. a truncated Asspatial a result, distribution nearly all of subsequent GRBs (as number-brightnessin Galactic models) tests or have an effectused peakdue to the fluxexpansion (P) rather of the than universe. fluence The(S). bendHowever, might thealso standard be caused candle by a combinationpeak lu- of minosityrate density assumption or number that density is required evolution, by log andN-log a study P studies of their is unphysical.possible effects is If, for instance, bursts originate at cosmological distances and are produced certainly warranted. Secondly, with the availability of Monte Carlo modeling by colliding neutron stars then one might expect that total energy would be standardof trigger and efficiencies, not peak luminosity. log N-log Moreover,S tests need recent no analysis longer ofbe relativistically inconclusive. expanding shellhttps://apod.nasa.gov/debate/debate95.html models has cast doubt on the standard L0 assumption (9). In this paper, we seek to eliminate the large threshold effects present in REFERENCES

1. D. Lo Band, et al., Ap.5. 413, 201 (1993). 2. E. E. Fenimore, and J. S. Bloom, Ap.J. 235, 29 (1995). 3. E. E. Fenimore, et al., Nature 366, 40 (1993). 4. E. E. Fenimore, et al., in preparation (1996). 5. 3. C. Higdon, R. E. Lingenfelter, Ap.J. 307, 197 (1986). 6. J. in t' Zand, and E. E. Fenimore, these proceedings (1996). 7. R. Klebesadel, E. Fenimore, and J. Laros, in AIP Conf. Proc. 115, 429 (1982). 8. L. Lu, W. W. Sargent, D. S. Womble, T. A. Barlow, Ap.J. 457, L1 (1996). 9. C. Madras, E. E. Fenimore, in preparation (1996). 10. C. A. Meegan, et al., Ap.J.S., in press (1996). 11. V. Petrosian, and T. Lee, these proceedings (1996).

325 at z ~ 0 especially if E0 is large. Therefore, for a given E0, Si, and ¢i(E) we first solve for the redshifts, zi, of the baseline events associated with each spectral shape. The standard// candle energy,¢ooo E0, is(.) given by, Eo = 4~rR~,= N(t,)dts E¢i dE (1) ,#30 where N(ts) is the normalization of the spectrum (ergs keV -1) at time t, at the source. The comoving distance, Ri,z, is defined in eq. [2] of ref. (2). The observed ith baseline burst fluence in the energy range 50-300 keV is,

-- abo3°° E¢i [rl+z,.E] l + zi J dE, (2)

NBAND fR(Sj+I ) P0 AN~xp[Sj to Sj+I]- 4~r ~ j~ e[Si(r)] r2dr. (3) NBAND .= R(Si) 1 + Zr

where NBAND ---- 48 is the number of baseline spectra used and po is the rate density of bursts per comoving volume. The quantity Si(r) is the predicted fluence (using eqs. [1, 2]) of the it__h baseline burst if it was at a distance r. This distance corresponds to a redshift 1 + zr. Discovery & Demographics We constructTABLE 11 fluence1. Best binsFit Distribution(in BATSE of channels E0 = 7.0 2+3× 1051 corresponding ergs to approximatelyTHE CORRECTED50-300Fluence keV) Ranges~(50-300 of roughly keV)equalLOG number AN[SjN-LOG of tobursts. S3+i ] We select High-energy Era burstsBin Number(j) with Cmin/Cmax Sj > 1 on eitherSj+I the 256 Observedor 1024 bms Predictedtimescale, 1then + zj find aFLUENCE minimized X2 between DISTRIBUTION the number of predicted burstsOF and observed 1 2.16e-07 3.82e-07 51 38.4 3.88 by varying E0. For 9 degrees of freedom we find an acceptable X2 = 14.7 COSMOLOGICAL2 3.82e-07 5.85e-07 v-RAY BURSTS42 50.5 3.24 corresponding to a standard candle E0 .....7 2 +0.7 a.0 × 1051 ergs. Table (I) gives 3 5.85e-07 7.55e-07 42 36.3 2.84 the bin ranges, number of observed bursts per bin, number of predicted bursts Joshua4 S. Bloom 7.55e-07 1'2, Edward 1.13e-06E. Fenimore 2, Jean46 in 't Zand 63.0 2'a 2.64 for the 5best fit energy, 1.13e-06 and their 1.43e-06implied redshifts. 37 36.8 2.36 1Harvard-Smithsonian6 1.43e-06 Center for 2.00e-06 Astrophysics, Cambridge,48 MA 0P138 49.7 2.22 “No Host Problem” (cf. Larson 97) 2Los Alamos National Laboratory, Los Alamos, NM 875~ 7 2.00e-06 CONCLUSIONS 2.80e-06 39 44.3 2.04 8 3 Goddard 2.80e-06 Space Flight Center, 4.05e-06 Greenbelt, MD 44P0771 41.1 1.89 9 4.05e-06 6.20e-06 37 37.2 1.74 Our fit of E0= 7 •0 +0.7-1.0x 1051 [30-2000 keV] ergs seems a plausible number “Great Debate” here in DC (Apr 95): 10 6.20e-06 1.36e-05 40 44.7 1.60 on the Recentbasis thatanalysis GRBs of relativistically last on the expandingaverage 10shells sec ofand cosmological L0 = 4.6 x 105° erg Galactic or Cosmological? 11 1.36e-05 6.60e-05 41 32.3 1.41 s -1 from"c-ray logbursts N-log has shown P studies that if (2).the burstsHowever, are cosmological, this E0 implies then most a rather large efficiencyIn ergslikely cmoftotal -2energy energy conversion (E0) is standard to 7-rays and (~not 10%)peak luminosityif the bursting (L0). mechanism bis BurstscollidingAssuming with neutron aCmin/Cmax fiat Friedmann stars >(Mtotal 1cosmology on the~ 2.8Mo).256 (qo or= 1/2,1024 Nevertheless, A =ms 0) timescale and constant this in resultBATSE would 3B. rate density (p0) of bursting sources, we fit a standard candle energy to seem a touniformly help resolve selected the log N-log"no-host" S in theproblem BATSE (cf. 3B catalogref (3)). correcting Interestingly, that 3rd Huntsville Conference (Oct 95) the dimmestfor fluence bursts efficiency (S -and 5 xaveraging 10 -s erg over cm 48 -2) observed are required spectral to shapes. be at a redshift of 1 + z We~_ 6.4find given the data this consistent E0, would with seem E0 = to7.3+_°:0 ruler xout 1051 several ergs and cosmological discuss models that implicationsrequire GRB of thisprogenitors energy for cosmologicalto be 324within models galaxies of 7-ray (although bursts. see reference (8)). This surprisingly high redshift is due to the correct blueshifting of the baseline spectra back to the source in eq. (1). If we neglect this factor, we obtain a smaller, more tenableINTRODUCTION redshift of the dimmest bursts (1 + z = 5.2). Whatever the conclusion about the models, we note two important results. First,On the the basis bend of instrong the thresholdlog N-log effects S curve of detectors,in BATSE Klebesadel, is real, not Fenimore, an artifact of andstrong Laros threshold (7) concluded effects. that This GRB implies fluence that tests we were are largely seeing eitherinconclusive. a truncated Asspatial a result, distribution nearly all of subsequent GRBs (as number-brightnessin Galactic models) tests or have an effectused peakdue to the Afterglow predictions: fluxexpansion (P) rather of the than universe. fluence The(S). bendHowever, might thealso standard be caused candle by a combinationpeak lu- of minosityrate density assumption or number that density is required evolution, by log andN-log a study P studies of their is unphysical.possible effects is Paczyński & Rhoads 93, Katz 94, If, for instance, bursts originate at cosmological distances and are produced certainly warranted. Secondly, with the availability of Monte Carlo modeling Mészáros & Rees 97 by colliding neutron stars then one might expect that total energy would be standardof trigger and efficiencies, not peak luminosity. log N-log Moreover,S tests need recent no analysis longer ofbe relativistically inconclusive. expanding shellhttps://apod.nasa.gov/debate/debate95.html models has cast doubt on the standard L0 assumption (9). In this paper, we seek to eliminate the large threshold effects present in REFERENCES

325 Afterglows

GRB 970228

X-ray afterglow discovery Optical afterglow discovery

Costa+97, van Paradijs+97

GRB 970508

Radio afterglow discovery Absorption redshift Host galaxy

Frail+97, Metzger+97, Taylor+97…

Burrrows: this afternoon –4–

a fast cooling −1/2 t

J) ν µ 2 D 10 Flux ( 2 ν t−1/2 t−1/2 t−3/2 [t−4/5] [t−2/7] [t−12/7] 0 A 10

νa νc νm

8 10 12 14 16 18 10 10 10 10 10 10 Afterglows

Relativistic external shock model for Afterglows

Paradigmatic Model b 1/3 −(p−1)/2 slow cooling ν ν t>t Emerges 4 0 10 F G 2 ν 2 J) 10 µ E −p/2 ν

Flux ( 0 0 −3/2 −1/2 H 10 t t t

−2 10 νa νm νc

8 10 12 14 16 18 10 10 10 10 10 10 ν (Hz) Sari, Piran, Narayan 98 Fig. 1.— Synchrotron spectrum of a relativistic shock with a power-law distribution of electrons. (a) The case of fast cooling, which is expected at early times (tt0). The evolution is always adiabatic. The four segments are identiﬁed as E, F, G, H. –4–

a fast cooling −1/2 t

J) ν µ 2 D 10 Flux ( 2 ν t−1/2 t−1/2 t−3/2 [t−4/5] [t−2/7] [t−12/7] 0 A 10

νa νc νm

8 10 12 14 16 18 10 10 10 10 10 10 Afterglows

Relativistic external shock model for Afterglows

Paradigmatic Model b 1/3 −(p−1)/2 slow cooling ν ν t>t Emerges 4 0 10 F G 2 ν 2 J) 10 Collimation µ E −p/2 ν

Theory: Rhoads 97 Flux ( 0 0 −3/2 −1/2 H 10 t t t Early Events: −2 10 •GRB 971214 Kulkarni+98 νa νm νc

•GRB 990510 8 10 12 14 16 18 Harrison+99, Stanek+99 10 10 10 10 10 10 ν (Hz) Sari, Piran, Narayan 98 Summary: Frail+01Fig. 1.— Synchrotron spectrum of a relativistic shock with a power-law distribution of electrons. (a) The case of fast cooling, which is expected at early times (tt0). The evolution is always adiabatic. The four segments are identiﬁed as E, F, G, H. NATURE | VOL 435 | 12 MAY 2005 |

- 3rd Swift localized event - 1st long-wavelength afterglow detected for Swift

- “Forward shock” flashes (eg., GRB990123 Akerlof+00 )

http://w.astro.berkeley.edu/~jbloom/Autotel/Workshop_Talks/Bloom_grb.pdf Collaboration: Butler, N., Watson, A. M., Kutyrev, A., Lee, W. H., Richer, M. G., Fox, O., Prochaska, J. X., Bloom, J., Cucchiara, A., Troja, E., Littlejohns, O., Ramirez-Ruiz, E., de Diego, J. A., Georgiev, L., Gonzalez, J., Roman-Zuniga, C., Gehrels, N., Moseley, H., Capone, J., Golkhou, V. Z., Klein, C., Toy, V.

http://butler.lab.asu.edu/RATIR/ Progenitors No. 3, 2002 OFFSET DISTRIBUTION OF GAMMA-RAY BURSTS 1141

•Long-soft GRBs (LSB) from massive stars LSB Locations (“collapsars”) Model: MacFayden & Woosley 99

•LSB locations correlated with the light of star NS-NS predictions forming galaxies Bloom, Kulkarni, Djorgovski 02, Fruchter+06

Fig. 7.—Oﬀset distribution of GRBs compared with delayed merging remnant binaries (NS-NS and BH-NS) prediction. The models, depicted as smooth curves, are the radial distributions in various galactic systems that have been projected by a factor of 1.15 (see text). The letters denote the model distributions 1 from Table 2 of Bloom et al. (1999d); a* is the galactic model that we consider the most representative of GRB hosts galaxies (vcirc =100kmsÀ , rbreak =1 9 kpc, re =1.5kpc,Mgal = 9.2 10 M ). The cumulative histogram is the observed data set. The inset shows the distribution of KS statistics (based on the maximum deviation from theÂ predicted and observed distribution) of 1000 synthetic data sets compared with model a*. Even with conservative assumptions (see text) the observed GRB distribution is inconsistent with the prediction: in only 0.3% of synthetic data sets is PKS 0.05. Instead, the collapsar–promptly bursting remnant progenitor model appears to be a better representation of the data (see Fig. 8).

(projected) galaxy, then the transient position will appear prove quite useful in this regard, as will IR imaging with the unrelated to any galaxies (the wrong host will be assigned, Next Generation Space Telescope. of course) but Pch will always appear high no matter how Weaker evidence for a star formation connection exists in deep the host search is. We try to account for this effect in that no GRB to date has been observed to be associated our modeling (Appendix C) by synthetically replacing with an early-type galaxy (morphologically or spectroscopi- observed (small) offsets that are associated with a high value cally), though in practice it is often difficult to discern galaxy of Pch with new generally larger offsets drawing from the type with the data at hand. Indeed most well-resolved hosts expected distribution of offsets for a particular galactic appear to be compact star-forming blue galaxies, spirals, or model. This then biases the distribution of PKS statistics morphological irregulars. toward higher values (by definition) but the median values Above we have demonstrated that GRBs follow the UV of PKS are largely unaffected (see Table 4). (rest frame) light of their host galaxies. However, the comparison has been primarily mediated by a single parameter, 7.2. Massive Stars (Collapsars) and Promptly Bursting the half-light radius and the median normalized offset. We Binaries (BH-He) now take this comparison one step further. For the GRB As discussed, collapsars produce GRBs in star-forming hosts with high signal-to-noise ratio HST detections (e.g., regions, as will BH-He binaries. The localization of GRB GRB 970508, 971214, and 980703) our analysis shows that 990705 near a spiral arm is, of course, tantalizing smaller the surface brightness is well approximated by an exponen- scale evidence of the GRB–star formation connection. tial disk. We use this finding as the point of departure for a Ideally, the burst sites of individual GRBs could be studied simplifying assumption about all GRB hosts: we assume an in detail with imaging and spectroscopy and should, if the exponential disk profile such that the surface brightness of collapsar–promptly bursting binary origin is correct, reveal the host galaxy scales linearly with the galactocentric radius that the burst sites are H ii regions. Unfortunately, the dis- in the disk. We further assume that the star formation rate tances to GRBs preclude a detailed examination of the spe- of massive stars scales with the observed optical light of the cific burst sites on a resolution scale of tens of parsecs (the host; this is not an unreasonable assumption given that typical size for a star-forming region) with current instru- HST STIS imaging probes rest-frame UV light, an excellent mentation. Adaptive optics laser–guide star imaging may tracer of massive stars, at GRB redshifts. Progenitors

•Spectroscopic Confirmation: GRB030329 Stanek+03

•Early Photometric evidence for a supernova connection GRB980326 Bloom+98 GRB970228 Reichart+98

cf., Woosley & Bloom 06; Hjorth & Bloom 12 for review GRB 050509b First Well-Localized Short-Hard GRB by 366 BLOOM ET AL. Vol. 638 Video from AAS Swift (BAT/XRT) press release (May Acore-collapsesupernova(SN)producesnoelectromag- axies, we found—based on a positional argument—plausible 31, 2005) netic radiation until its envelope is completely consumed by the evidence that the progenitor is likely associated with 2MASX explosion (although see Khokhlov et al. 1999). This phase ends, J12361286+2858580, a bright elliptical galaxy at z 0:2248. however, with a brilliant flash of X-ray or extreme ultraviolet We have argued that the observations find natural explanation¼ photons as the shock reaches the stellar surface. The ‘‘breakout’’ with a compact merger system progenitor. If so, then short-hard flash is delayed in time, and vastly reduced in energy, relative to GRBs provide a bridge from electromagnetic to gravitational the neutrino transient produced by core collapse. However, it wave astronomy: indeed, had GRB 050509b occurred a factor conveys useful information about the explosion. Shock break- of 3 closer in luminosity distance, it might have produced a out flashes were predicted by Colgate (1968) as a source for (the detectable chirp signal with the next-generation Laser Inter- then-undetected) gamma-ray bursts. The explosion of SN 1987A ferometer Gravitational-Wave Observatory (LIGO II).26 The Astrophysical Journal,638:354–368,2006February10 A # 2006. The Americanstimulated Astronomical a reanalysis Society. All rights of reserved. SN Printed breakout in U.S.A. flashes by Ensman & Brightening emission from most types of supernovae would Burrows (1992) and, more recently, by Blinnikov et al. (1998, have been seen in our imaging, so the lack of such emission 2000). These studies represent an increase in sophistication to- appears inconsistent with the notion that short bursts are due to ward the full numerical treatment of this complicated, radiation- collapsars or variants thereof.Ourafterglowmodelingisalso CLOSINGhydrodynamic IN ON A problem. SHORT-HARD In principle, BURST the XRT PROGENITOR: data could constrain CONSTRAINTSconsistent FROM with, EARLY-TIME but does not require, OPTICAL a circumburst medium the existenceIMAGING of a AND shock SPECTROSCOPY breakout produced OF by both A POSSIBLE a red super- HOSThaving GALAXY lower OF density GRB than 050509b that inferred in long-duration GRBs; giant explosion like SN 1993J (Van Dyk et al. 2002) and a blue if true, this would suggest that the progenitor produces a GRB J. S. Bloom,1 J. X. Prochaska,2 D. Pooley,1,3 C. H. Blake,4 R. J. Foley,1 S. Jha,1 E. Ramirez-Ruiz,2,3,5 supergiant explosion5,6 analog to SN 1987A,1 but the X-ray7 lumi- 8 in an environment9 that is baryon-poor1 compared to that expected nosity isJ. sensitive Granot, to theA. V. uncertain Filippenko, distributionS. Sigurdsson, of the extragalac-A. J. Barth, H.-W.for collapsars. Chen, M. Moreover, C. Cooper, we have seen no evidence for ongoing E. E. Falco,4 R. R. Gal,10 B. F. Gerke,11 M. D. Gladders,12 J. E. Greene,4 J. Hennanwi,1,13 tic gas columnL. and C. the Ho, specific12 K. Hurley, XRT observing14 B. P. Koester, epochs.15 W. Li,1 L. Lubin,star10 formationJ. Newman, in13,16 the putative host, so there are likely no remain- Using our ESI optical imaging,D. A. Perley, we can1 G. also K. Squires, limit the17 presenceand W. M. Wood-Vaseying massive4 stars. GivenarXiv the [v1] short Tue, active 24 May life 2005 of 18:27:28 a neutron GMT star of brightening due to a SN or SN-likeReceiv emissionedv 2005 May at 24; 8.17 accepted days 2005 after Septemberhaving 3 a high magnetic field, this also disfavors the magnetar the GRB to RC 25:0 mag. A normal, unextinguished Type Ia hypothesis. (thermonuclear) SN at z 0:22 would have R ABSTRACT22 mag around The nondetection of brightening emission may place limits 6.7 days after explosion¼ (t 8:17 days in the observer’s frame). on the presence of a thermal ‘‘minisupernova’’ from nonrela- AverysubluminousSNIalikeSN1991bg(Filippenkoetal.The localization of the short-duration,¼ hard-spectrum gamma-ray burst GRBtivistic 050509b ejecta by the of aSwift compactsatellite merger was a system (Li & Paczyn´ski watershed event. We report the discovery of the probable host galaxy, a bright elliptical galaxy at z 0:2248. This is the 1992) would have R 24 mag, still somewhat brighter than our 1998; Rosswog &¼ Ramirez-Ruiz 2002). In this scenario, the small first known redshift and host of a short-hard GRB and shows that at least some short-hard GRBs are cosmological in origin.limit. ExtinctionWe began imaging would the obviously GRB field make 8 minutes the after SN fainter, the burst but and the continueddense for 8 days. mass We (m presentej) ejected a reanalysis during coalescence of the expands as it is heated Milky Way contribution is small (AV 0:06 mag; Schlegel by radioactivity of the decompressed ejecta. Using the scalings XRTafterglow and report the absolute position of the GRB. Based on positional coincidences, the GRB and the elliptical areet al. likely 1998), to be and physically the outskirts related, unlike of an elliptical any known galaxy connection in a between cluster a long-durationof Li & Paczyn GRB and´ski an (1998) early-type and galaxy. crudely assuming that 10% of the Similarlyshould have unique, essentially GRB 050509b no dust. likely While also some originated core-collapse from within su- a rich clusterbolometric of galaxies light with at peak detectable is radiated diffuse in the R band, the R-band 1=2 X-raypernovae emission. could We be demonstrate as faint as that (or while fainter the burst than) was our underluminous, limit, the the ratiobrightness of the blast should wave peak energy at toobserver the -ray time t 1:2(mej/0:01 M ) energypresence is consistent of such a with supernova that of long-duration in the outskirts GRBs. of an Based elliptical on this gal- analysis, ondays the location after the of the burst, GRB with (40 absolute13 kpc magnitude MR 18:5 Æ À À fromaxy would the putative be truly host), extraordinary on the galaxy (see type van (elliptical), den Bergh an etdthelackofacoincidentsupernova,wesuggestthatthere al. 2005). 1:25 log (mej/0:01 M ) mag. Assuming that the GRB did indeed isOthers now observational have also reported support for no the evidence hypothesis for a that SN short-hard at later times bursts arise duringoriginate the from merger the of redshift a compactz binary.0:2248, from inspection of Fig- We limit the properties of any Li-Paczyn´ski ‘‘minisupernova’’ that is predicted to arise on 1daytimescales.Other¼ (Hjorth et al. 2005a; Bersier et al. 2005). ure 2, with nondetections at MR 16 mag at t 1days,we progenitor models are still viable, and new Swift bursts will undoubtedly help to further clarify the progenitor picture.À 3 The location of this short burst (and future short bursts) pro- can very roughly exclude mej > few ; 10À M .AlthoughtheLi Subjectvides a headin usefulg discriminantg: gamma rays: for bursts distinguishing between different &Paczyn´ski (1998) model was intended as a simplistic sketch of Onlineprogenitor material: modelscolor of short figures bursts. Simplistically, we would ex- the phenomenon, this limit on mej is somewhat surprising given pect evaporating black holes to occur near the center of deep the amount of escaping nonrelativistic material expected in com- potential wells (as discussed in the context of Galactic BHs; pact mergers (Rosswog & Ramirez-Ruiz 2002). Indeed, we con- Cline et al.1. 1999); INTRODUCTION thus, the offset from G1 seems to disfavorwith thisdurations ofsider a few this hundred lack of milliseconds a minisupernova and harder as weak spectra. evidence against a z Despite remarkable progress in understanding the nature and The distribution in duration (Mazets et al. 1981; Norris et al. ¼ hypothesis. A giant flare from a magnetar would need toprogenitors have an of long-duration0:22 origin from GRBs, a comparativelycompact merger little system. has been Still, these limits are 1984) and hardness (Kouveliotou et al. 1993) reveals evidence 3 isotropic luminosity (L ;iso) larger by a factor of 10 learnedand an about thesubject origin to of short-hard considerable bursts, uncertainty primarily because in a number of uncertain for two distinct populations of classic gamma-ray2 bursts (GRBs): E ;iso larger by a factor of 10 compared to the initialvery spike few of such burstsparameters have had of ejecta. rapid and For precise instance, localizations. if the velocity of the ejecta long-duration bursts, with typical durations around 30 s and peak the 2004 December 27 giant flare from SGR 1806 20The (the modeledwere bursting to be rate at0.01 redshiftc insteadz of0 of 0.3 long-softc (as assumed bursts by Li & Paczyn´ski energies at 200 keV; and the minority, short-duration bursts, À ¼ difference in the factor between the two quantities arisesoutnumbers since short-hard1998), bursts then the by about peak a of factor the thermal of 3.5 in emissionthe Burst would occur after GRB 050509b lasted only 30 ms, which is 10 timesand shorter Transient Sourceabout 1 Experiment month and (BATSE)would not catalog have been (Schmidt detected with the current 1 Departmentthan of the Astronomy, initial 601 spike Campbell of the Hall, giant University flare of from California, SGR 1806-20).2001); this assumeslimits. the same bursting rate as a function of red- Berkeley, CABursts 94720-3411. from magnetars might be expected from galaxies ofshift a later and does notWe include conclude the effect by emphasizing of beaming, which, that in if dif-the NS-NS or BH-NS 2 Universitytype of than California G1, Observatories/ where neutron Lick Observatory, stars would University be formed of copiously:ferent for longprogenitor and short bursts, hypothesis would imply for short-hard that the intrinsic bursts, the host galaxies California, Santa Cruz, CA 95064. relative rates differ from those observed. While a number of bursts 3 Chandramagnetic Fellow. field decay would cut the active lifetime for megaflare may be a range of Hubble types (e.g., Livio et al. 1998). Com- 4 have been triangulated through the Interplanetary Network (see Harvard-Smithsonianactivity after Center10 for4 Astrophysics,yr. 60 Garden Street, Cambridge, pact merger systems coalesce in appreciable rates from Myr to MA 02138. Hurley et al. 2005b) on roughly day-long timescales, there has 5 Gyr after a starburst (e.g., Fryer et al. 1999; Bloom et al. 1999). Institute for Advanced Study, Olden Lane, Princeton, NJ 08540. only been one precisely localized short-hard burst relayed to 6 KIPAC, Stanford University, P.O.8. Box CONCLUSIONS 20450, Mail Stop 29, Stanford, ground observersObviously, in less than the 1 hr longer (GRB the 050202, time sinceSwift;Tueller the starburst, the larger the CA 94309. 18 distance a binary system will travel before coalescence. A clear 7 et al. 2005); owing to its proximity to the Sun at the time of DepartmentWe of Astronomy have monitored and Astrophysics, the location 525 Davey of Laboratory, GRB 050509b Penn- at optical prediction from this model is that as more short bursts are lo- sylvania Stateand University, infrared University wavelengths Park, PA 16802. from 8 minutes to 8 days afterlocalization, the sparse ground-based follow-up was undertaken. 8 Department of Physics and Astronomy, 4129 Frederick Reines Hall, Including GRBcalized, 050509b, those this associatedcorresponds with to a later-typeratio of 1:18 galaxies for of a given mass University oftrigger California, and Irvine, found CA no 92697-4575. indication of variability at the locationshort-hard of to long-softshould burst be preferentially detections with closerSwift, to much the smaller star formation centers of 9 Centerthe for Space fading Research, X-ray Massachusetts source, the Institute first of solid Technology, X-ray Cam- detectionthan of the an BATSEthe result. host;19 that is, we expect a more concentrated distribution bridge, MAafterglow 02139-4307. of a short-hard burst. Near the location of this source 10 Department of Physics, 1 Shields Avenue, University of California, Davis, As with long-durationaround a spiralbursts, galaxy the distribution with the of same short mass bursts as an early type. On CA 95616-8677.we and others have found an apparent group of faint blueappears gal- very nearly isotropic (Kouveliotou et al. 1993; Briggs 11 Departmentaxies of at Physics, redshifts 366k LeConte1.3. While Hall, University it is indeed of California, plausible thatet al. this 1996), and their brightness distribution ( V/Vmax 0:35) is Berkeley, CAshort 94720-7300. burst arose from a progenitor connected with those gal- 26 See http://www.ligo.caltech.edu/docs/G/G990111-00.pdf.h i 12 Carnegie Observatories, 813 Santa Barbara Street, Pasadena, CA 91101. 13 Hubble Fellow. 14 Space Sciences Laboratory, 7 Gauss Way, University of California, 18 There have been a few other short bursts (durationP2s)detectedandwell Berkeley, CA 94720-7450. localized, but with soft spectra and hence not members of the short-hard class. 15 Department of Physics, University of Michigan, Ann Arbor, MI 48109-1090. For example, GRB 040924 was a soft, X-ray–rich GRB (Fenimore et al. 2004; 16 Institute for Nuclear and Particle Astrophysics, Lawrence Berkeley Na- Huang et al. 2005). Hereafter, we use the term ‘‘short burst’’ interchangeably with tional Laboratory, Berkeley, CA 94720. short-hard burst. 17 Spitzer Science Center, Mail Stop 314-6, California Institute of Tech- 19 As of 2005 May 20, Swift has localized two short bursts out of a total of nology, Pasadena, CA 91125. 38; see http://swift.gsfc.nasa.gov/docs/swift/archive/grb_table.html. 354 GRB 050509b First Well-Localized Short-Hard GRB by 366 BLOOM ET AL. Vol. 638 Video from AAS Swift (BAT/XRT) press release (May Acore-collapsesupernova(SN)producesnoelectromag- axies, we found—based on a positional argument—plausible 31, 2005) netic radiation until its envelope is completely consumed by the evidence that the progenitor is likely associated with 2MASX explosion (although see Khokhlov et al. 1999). This phase ends, J12361286+2858580, a bright elliptical galaxy at z 0:2248. however, with a brilliant flash of X-ray or extreme ultraviolet We have argued that the observations find natural explanation¼ photons as the shock reaches the stellar surface. The ‘‘breakout’’ with a compact merger system progenitor. If so, then short-hard flash is delayed in time, and vastly reduced in energy, relative to GRBs provide a bridge from electromagnetic to gravitational the neutrino transient produced by core collapse. However, it wave astronomy: indeed, had GRB 050509b occurred a factor conveys useful information about the explosion. Shock break- of 3 closer in luminosity distance, it might have produced a out flashes were predicted by Colgate (1968) as a source for (the detectable chirp signal with the next-generation Laser Inter- then-undetected) gamma-ray bursts. The explosion of SN 1987A ferometer Gravitational-Wave Observatory (LIGO II).26 The Astrophysical Journal,638:354–368,2006February10 A # 2006. The Americanstimulated Astronomical a reanalysis Society. All rights of reserved. SN Printed breakout in U.S.A. flashes by Ensman & Brightening emission from most types of supernovae would Burrows (1992) and, more recently, by Blinnikov et al. (1998, have been seen in our imaging, so the lack of such emission 2000). These studies represent an increase in sophistication to- appears inconsistent with the notion that short bursts are due to ward the full numerical treatment of this complicated, radiation- collapsars or variants thereof.Ourafterglowmodelingisalso CLOSINGhydrodynamic IN ON A problem. SHORT-HARD In principle, BURST the XRT PROGENITOR: data could constrain CONSTRAINTSconsistent FROM with, EARLY-TIME but does not require, OPTICAL a circumburst medium the existenceIMAGING of a AND shock SPECTROSCOPY breakout produced OF by both A POSSIBLE a red super- HOSThaving GALAXY lower OF density GRB than 050509b that inferred in long-duration GRBs; giant explosion like SN 1993J (Van Dyk et al. 2002) and a blue if true, this would suggest that the progenitor produces a GRB J. S. Bloom,1 J. X. Prochaska,2 D. Pooley,1,3 C. H. Blake,4 R. J. Foley,1 S. Jha,1 E. Ramirez-Ruiz,2,3,5 supergiant explosion5,6 analog to SN 1987A,1 but the X-ray7 lumi- 8 in an environment9 that is baryon-poor1 compared to that expected nosity isJ. sensitive Granot, to theA. V. uncertain Filippenko, distributionS. Sigurdsson, of the extragalac-A. J. Barth, H.-W.for collapsars. Chen, M. Moreover, C. Cooper, we have seen no evidence for ongoing E. E. Falco,4 R. R. Gal,10 B. F. Gerke,11 M. D. Gladders,12 J. E. Greene,4 J. Hennanwi,1,13 tic gas columnL. and C. the Ho, specific12 K. Hurley, XRT observing14 B. P. Koester, epochs.15 W. Li,1 L. Lubin,star10 formationJ. Newman, in13,16 the putative host, so there are likely no remain- Using our ESI optical imaging,D. A. Perley, we can1 G. also K. Squires, limit the17 presenceand W. M. Wood-Vaseying massive4 stars. GivenarXiv the [v1] short Tue, active 24 May life 2005 of 18:27:28 a neutron GMT star of brightening due to a SN or SN-likeReceiv emissionedv 2005 May at 24; 8.17 accepted days 2005 after Septemberhaving 3 a high magnetic field, this also disfavors the magnetar the GRB to RC 25:0 mag. A normal, unextinguished Type Ia hypothesis. (thermonuclear) SN at z 0:22 would have R ABSTRACT22 mag around The nondetection of brightening emission may place limits 6.7 days after explosion¼ (t 8:17 days in the observer’s frame). on the presence of a thermal ‘‘minisupernova’’ from nonrela- AverysubluminousSNIalikeSN1991bg(Filippenkoetal.The localization of the short-duration,¼ hard-spectrum gamma-ray burst GRBtivistic 050509b ejecta by the of aSwift compactsatellite merger was a system (Li & Paczyn´ski watershed event. We report the discovery of the probable host galaxy, a bright elliptical galaxy at z 0:2248. This is the 1992) would have R 24 mag, still somewhat brighter than our 1998; Rosswog &¼ Ramirez-Ruiz 2002). In this scenario, the small first known redshift and host of a short-hard GRB and shows that at least some short-hard GRBs are cosmological in origin.limit. ExtinctionWe began imaging would the obviously GRB field make 8 minutes the after SN fainter, the burst but and the continueddense for 8 days. mass We (m presentej) ejected a reanalysis during coalescence of the expands as it is heated Milky Way contribution is small (AV 0:06 mag; Schlegel by radioactivity of the decompressed ejecta. Using the scalings XRTafterglow and report the absolute position of the GRB. Based on positional coincidences, the GRB and the elliptical areet al. likely 1998), to be and physically the outskirts related, unlike of an elliptical any known galaxy connection in a between cluster a long-durationof Li & Paczyn GRB and´ski an (1998) early-type and galaxy. crudely assuming that 10% of the Similarlyshould have unique, essentially GRB 050509b no dust. likely While also some originated core-collapse from within su- a rich clusterbolometric of galaxies light with at peak detectable is radiated diffuse in the R band, the R-band 1=2 X-raypernovae emission. could We be demonstrate as faint as that (or while fainter the burst than) was our underluminous, limit, the the ratiobrightness of the blast should wave peak energy at toobserver the -ray time t 1:2(mej/0:01 M ) energypresence is consistent of such a with supernova that of long-duration in the outskirts GRBs. of an Based elliptical on this gal- analysis, ondays the location after the of the burst, GRB with (40 absolute13 kpc magnitude MR 18:5 Æ À À fromaxy would the putative be truly host), extraordinary on the galaxy (see type van (elliptical), den Bergh an etdthelackofacoincidentsupernova,wesuggestthatthere al. 2005). 1:25 log (mej/0:01 M ) mag. Assuming that the GRB did indeed isOthers now observational have also reported support for no the evidence hypothesis for a that SN short-hard at later times bursts arise duringoriginate the from merger the of redshift a compactz binary.0:2248, from inspection of Fig- We limit the properties of any Li-Paczyn´ski ‘‘minisupernova’’ that is predicted to arise on 1daytimescales.Other¼ (Hjorth et al. 2005a; Bersier et al. 2005). ure 2, with nondetections at MR 16 mag at t 1days,we progenitor models are still viable, and new Swift bursts will undoubtedly help to further clarify the progenitor picture.À 3 The location of this short burst (and future short bursts) pro- can very roughly exclude mej > few ; 10À M .AlthoughtheLi Subjectvides a headin usefulg discriminantg: gamma rays: for bursts distinguishing between different &Paczyn´ski (1998) model was intended as a simplistic sketch of Onlineprogenitor material: modelscolor of short figures bursts. Simplistically, we would ex- the phenomenon, this limit on mej is somewhat surprising given pect evaporating black holes to occur near the center of deep the amount of escaping nonrelativistic material expected in com- potential wells (as discussed in the context of Galactic BHs; pact mergers (Rosswog & Ramirez-Ruiz 2002). Indeed, we con- Cline et al.1. 1999); INTRODUCTION thus, the offset from G1 seems to disfavorwith thisdurations ofsider a few this hundred lack of milliseconds a minisupernova and harder as weak spectra. evidence against a z Despite remarkable progress in understanding the nature and The distribution in duration (Mazets et al. 1981; Norris et al. ¼ hypothesis. A giant flare from a magnetar would need toprogenitors have an of long-duration0:22 origin from GRBs, a comparativelycompact merger little system. has been Still, these limits are 1984) and hardness (Kouveliotou et al. 1993) reveals evidence 3 isotropic luminosity (L ;iso) larger by a factor of 10 learnedand an about thesubject origin to of short-hard considerable bursts, uncertainty primarily because in a number of uncertain for two distinct populations of classic gamma-ray2 bursts (GRBs): E ;iso larger by a factor of 10 compared to the initialvery spike few of such burstsparameters have had of ejecta. rapid and For precise instance, localizations. if the velocity of the ejecta long-duration bursts, with typical durations around 30 s and peak the 2004 December 27 giant flare from SGR 1806 20The (the modeledwere bursting to be rate at0.01 redshiftc insteadz of0 of 0.3 long-softc (as assumed bursts by Li & Paczyn´ski energies at 200 keV; and the minority, short-duration bursts, À ¼ difference in the factor between the two quantities arisesoutnumbers since short-hard1998), bursts then the by about peak a of factor the thermal of 3.5 in emissionthe Burst would occur after GRB 050509b lasted only 30 ms, which is 10 timesand shorter Transient Sourceabout 1 Experiment month and (BATSE)would not catalog have been (Schmidt detected with the current 1 Departmentthan of the Astronomy, initial 601 spike Campbell of the Hall, giant University flare of from California, SGR 1806-20).2001); this assumeslimits. the same bursting rate as a function of red- Berkeley, CABursts 94720-3411. from magnetars might be expected from galaxies ofshift a later and does notWe include conclude the effect by emphasizing of beaming, which, that in if dif-the NS-NS or BH-NS 2 Universitytype of than California G1, Observatories/ where neutron Lick Observatory, stars would University be formed of copiously:ferent for longprogenitor and short bursts, hypothesis would imply for short-hard that the intrinsic bursts, the host galaxies California, Santa Cruz, CA 95064. relative rates differ from those observed. While a number of bursts 3 Chandramagnetic Fellow. field decay would cut the active lifetime for megaflare may be a range of Hubble types (e.g., Livio et al. 1998). Com- 4 have been triangulated through the Interplanetary Network (see Harvard-Smithsonianactivity after Center10 for4 Astrophysics,yr. 60 Garden Street, Cambridge, pact merger systems coalesce in appreciable rates from Myr to MA 02138. Hurley et al. 2005b) on roughly day-long timescales, there has 5 Gyr after a starburst (e.g., Fryer et al. 1999; Bloom et al. 1999). Institute for Advanced Study, Olden Lane, Princeton, NJ 08540. only been one precisely localized short-hard burst relayed to 6 KIPAC, Stanford University, P.O.8. Box CONCLUSIONS 20450, Mail Stop 29, Stanford, ground observersObviously, in less than the 1 hr longer (GRB the 050202, time sinceSwift;Tueller the starburst, the larger the CA 94309. 18 distance a binary system will travel before coalescence. A clear 7 et al. 2005); owing to its proximity to the Sun at the time of DepartmentWe of Astronomy have monitored and Astrophysics, the location 525 Davey of Laboratory, GRB 050509b Penn- at optical prediction from this model is that as more short bursts are lo- sylvania Stateand University, infrared University wavelengths Park, PA 16802. from 8 minutes to 8 days afterlocalization, the sparse ground-based follow-up was undertaken. 8 Department of Physics and Astronomy, 4129 Frederick Reines Hall, Including GRBcalized, 050509b, those this associatedcorresponds with to a later-typeratio of 1:18 galaxies for of a given mass University oftrigger California, and Irvine, found CA no 92697-4575. indication of variability at the locationshort-hard of to long-softshould burst be preferentially detections with closerSwift, to much the smaller star formation centers of 9 Centerthe for Space fading Research, X-ray Massachusetts source, the Institute first of solid Technology, X-ray Cam- detectionthan of the an BATSEthe result. host;19 that is, we expect a more concentrated distribution bridge, MAafterglow 02139-4307. of a short-hard burst. Near the location of this source 10 Department of Physics, 1 Shields Avenue, University of California, Davis, As with long-durationaround a spiralbursts, galaxy the distribution with the of same short mass bursts as an early type. On CA 95616-8677.we and others have found an apparent group of faint blueappears gal- very nearly isotropic (Kouveliotou et al. 1993; Briggs 11 Departmentaxies of at Physics, redshifts 366k LeConte1.3. While Hall, University it is indeed of California, plausible thatet al. this 1996), and their brightness distribution ( V/Vmax 0:35) is Berkeley, CAshort 94720-7300. burst arose from a progenitor connected with those gal- 26 See http://www.ligo.caltech.edu/docs/G/G990111-00.pdf.h i 12 Carnegie Observatories, 813 Santa Barbara Street, Pasadena, CA 91101. 13 Hubble Fellow. 14 Space Sciences Laboratory, 7 Gauss Way, University of California, 18 There have been a few other short bursts (durationP2s)detectedandwell Berkeley, CA 94720-7450. localized, but with soft spectra and hence not members of the short-hard class. 15 Department of Physics, University of Michigan, Ann Arbor, MI 48109-1090. For example, GRB 040924 was a soft, X-ray–rich GRB (Fenimore et al. 2004; 16 Institute for Nuclear and Particle Astrophysics, Lawrence Berkeley Na- Huang et al. 2005). Hereafter, we use the term ‘‘short burst’’ interchangeably with tional Laboratory, Berkeley, CA 94720. short-hard burst. 17 Spitzer Science Center, Mail Stop 314-6, California Institute of Tech- 19 As of 2005 May 20, Swift has localized two short bursts out of a total of nology, Pasadena, CA 91125. 38; see http://swift.gsfc.nasa.gov/docs/swift/archive/grb_table.html. 354 GRB 050509b First Well-Localized Short-Hard GRB by 366 BLOOM ET AL. Vol. 638 Video from AAS Swift (BAT/XRT) press release (May Acore-collapsesupernova(SN)producesnoelectromag- axies, we found—based on a positional argument—plausible 31, 2005) netic radiation until its envelope is completely consumed by the evidence that the progenitor is likely associated with 2MASX explosion (although see Khokhlov et al. 1999). This phase ends, J12361286+2858580, a bright elliptical galaxy at z 0:2248. however, with a brilliant flash of X-ray or extreme ultraviolet We have argued that the observations find natural explanation¼ photons as the shock reaches the stellar surface. The ‘‘breakout’’ with a compact merger system progenitor. If so, then short-hard flash is delayed in time, and vastly reduced in energy, relative to GRBs provide a bridge from electromagnetic to gravitational the neutrino transient produced by core collapse. However, it wave astronomy: indeed, had GRB 050509b occurred a factor conveys useful information about the explosion. Shock break- of 3 closer in luminosity distance, it might have produced a out flashes were predicted by Colgate (1968) as a source for (the detectable chirp signal with the next-generation Laser Inter- then-undetected) gamma-ray bursts. The explosion of SN 1987A ferometer Gravitational-Wave Observatory (LIGO II).26 The Astrophysical Journal,638:354–368,2006February10 A # 2006. The Americanstimulated Astronomical a reanalysis Society. All rights of reserved. SN Printed breakout in U.S.A. flashes by Ensman & Brightening emission from most types of supernovae would Burrows (1992) and, more recently, by Blinnikov et al. (1998, have been seen in our imaging, so the lack of such emission 2000). These studies represent an increase in sophistication to- appears inconsistent with the notion that short bursts are due to ward the full numerical treatment of this complicated, radiation- collapsars or variants thereof.Ourafterglowmodelingisalso CLOSINGhydrodynamic IN ON A problem. SHORT-HARD In principle, BURST the XRT PROGENITOR: data could constrain CONSTRAINTSconsistent FROM with, EARLY-TIME but does not require, OPTICAL a circumburst medium the existenceIMAGING of a AND shock SPECTROSCOPY breakout produced OF by both A POSSIBLE a red super- HOSThaving GALAXY lower OF density GRB than 050509b that inferred in long-duration GRBs; giant explosion like SN 1993J (Van Dyk et al. 2002) and a blue if true, this would suggest that the progenitor produces a GRB J. S. Bloom,1 J. X. Prochaska,2 D. Pooley,1,3 C. H. Blake,4 R. J. Foley,1 S. Jha,1 E. Ramirez-Ruiz,2,3,5 supergiant explosion5,6 analog to SN 1987A,1 but the X-ray7 lumi- 8 in an environment9 that is baryon-poor1 compared to that expected nosity isJ. sensitive Granot, to theA. V. uncertain Filippenko, distributionS. Sigurdsson, of the extragalac-A. J. Barth, H.-W.for collapsars. Chen, M. Moreover, C. Cooper, we have seen no evidence for ongoing E. E. Falco,4 R. R. Gal,10 B. F. Gerke,11 M. D. Gladders,12 J. E. Greene,4 J. Hennanwi,1,13 tic gas columnL. and C. the Ho, specific12 K. Hurley, XRT observing14 B. P. Koester, epochs.15 W. Li,1 L. Lubin,star10 formationJ. Newman, in13,16 the putative host, so there are likely no remain- Using our ESI optical imaging,D. A. Perley, we can1 G. also K. Squires, limit the17 presenceand W. M. Wood-Vaseying massive4 stars. GivenarXiv the [v1] short Tue, active 24 May life 2005 of 18:27:28 a neutron GMT star of brightening due to a SN or SN-likeReceiv emissionedv 2005 May at 24; 8.17 accepted days 2005 after Septemberhaving 3 a high magnetic field, this also disfavors the magnetar the GRB to RC 25:0 mag. A normal, unextinguished Type Ia hypothesis. (thermonuclear) SN at z 0:22 would have R ABSTRACT22 mag around The nondetection of brightening emission may place limits 6.7 days after explosion¼ (t 8:17 days in the observer’s frame). on the presence of a thermal ‘‘minisupernova’’ from nonrela- AverysubluminousSNIalikeSN1991bg(Filippenkoetal.The localization of the short-duration,¼ hard-spectrum gamma-ray burst GRBtivistic 050509b ejecta by the of aSwift compactsatellite merger was a system (Li & Paczyn´ski watershed event. We report the discovery of the probable host galaxy, a bright elliptical galaxy at z 0:2248. This is the 1992) would have R 24 mag, still somewhat brighter than our 1998; Rosswog &¼ Ramirez-Ruiz 2002). In this scenario, the small first known redshift and host of a short-hard GRB and shows that at least some short-hard GRBs are cosmological in origin.limit. ExtinctionWe began imaging would the obviously GRB field make 8 minutes the after SN fainter, the burst but and the continueddense for 8 days. mass We (m presentej) ejected a reanalysis during coalescence of the expands as it is heated Milky Way contribution is small (AV 0:06 mag; Schlegel by radioactivity of the decompressed ejecta. Using the scalings XRTafterglow and report the absolute position of the GRB. Based on positional coincidences, the GRB and the elliptical areet al. likely 1998), to be and physically the outskirts related, unlike of an elliptical any known galaxy connection in a between cluster a long-durationof Li & Paczyn GRB and´ski an (1998) early-type and galaxy. crudely assuming that 10% of the Similarlyshould have unique, essentially GRB 050509b no dust. likely While also some originated core-collapse from within su- a rich clusterbolometric of galaxies light with at peak detectable is radiated diffuse in the R band, the R-band 1=2 X-raypernovae emission. could We be demonstrate as faint as that (or while fainter the burst than) was our underluminous, limit, the the ratiobrightness of the blast should wave peak energy at toobserver the -ray time t 1:2(mej/0:01 M ) energypresence is consistent of such a with supernova that of long-duration in the outskirts GRBs. of an Based elliptical on this gal- analysis, ondays the location after the of the burst, GRB with (40 absolute13 kpc magnitude MR 18:5 Æ À À fromaxy would the putative be truly host), extraordinary on the galaxy (see type van (elliptical), den Bergh an etdthelackofacoincidentsupernova,wesuggestthatthere al. 2005). 1:25 log (mej/0:01 M ) mag. Assuming that the GRB did indeed isOthers now observational have also reported support for no the evidence hypothesis for a that SN short-hard at later times bursts arise duringoriginate the from merger the of redshift a compactz binary.0:2248, from inspection of Fig- We limit the properties of any Li-Paczyn´ski ‘‘minisupernova’’ that is predicted to arise on 1daytimescales.Other¼ (Hjorth et al. 2005a; Bersier et al. 2005). ure 2, with nondetections at MR 16 mag at t 1days,we progenitor models are still viable, and new Swift bursts will undoubtedly help to further clarify the progenitor picture.À 3 The location of this short burst (and future short bursts) pro- can very roughly exclude mej > few ; 10À M .AlthoughtheLi Subjectvides a headin usefulg discriminantg: gamma rays: for bursts distinguishing between different &Paczyn´ski (1998) model was intended as a simplistic sketch of Onlineprogenitor material: modelscolor of short figures bursts. Simplistically, we would ex- the phenomenon, this limit on mej is somewhat surprising given pect evaporating black holes to occur near the center of deep the amount of escaping nonrelativistic material expected in com- potential wells (as discussed in the context of Galactic BHs; pact mergers (Rosswog & Ramirez-Ruiz 2002). Indeed, we con- Cline et al.1. 1999); INTRODUCTION thus, the offset from G1 seems to disfavorwith thisdurations ofsider a few this hundred lack of milliseconds a minisupernova and harder as weak spectra. evidence against a z Despite remarkable progress in understanding the nature and The distribution in duration (Mazets et al. 1981; Norris et al. ¼ hypothesis. A giant flare from a magnetar would need toprogenitors have an of long-duration0:22 origin from GRBs, a comparativelycompact merger little system. has been Still, these limits are 1984) and hardness (Kouveliotou et al. 1993) reveals evidence 3 isotropic luminosity (L ;iso) larger by a factor of 10 learnedand an about thesubject origin to of short-hard considerable bursts, uncertainty primarily because in a number of uncertain for two distinct populations of classic gamma-ray2 bursts (GRBs): E ;iso larger by a factor of 10 compared to the initialvery spike few of such burstsparameters have had of ejecta. rapid and For precise instance, localizations. if the velocity of the ejecta long-duration bursts, with typical durations around 30 s and peak the 2004 December 27 giant flare from SGR 1806 20The (the modeledwere bursting to be rate at0.01 redshiftc insteadz of0 of 0.3 long-softc (as assumed bursts by Li & Paczyn´ski energies at 200 keV; and the minority, short-duration bursts, À ¼ difference in the factor between the two quantities arisesoutnumbers since short-hard1998), bursts then the by about peak a of factor the thermal of 3.5 in emissionthe Burst would occur after GRB 050509b lasted only 30 ms, which is 10 timesand shorter Transient Sourceabout 1 Experiment month and (BATSE)would not catalog have been (Schmidt detected with the current 1 Departmentthan of the Astronomy, initial 601 spike Campbell of the Hall, giant University flare of from California, SGR 1806-20).2001); this assumeslimits. the same bursting rate as a function of red- Berkeley, CABursts 94720-3411. from magnetars might be expected from galaxies ofshift a later and does notWe include conclude the effect by emphasizing of beaming, which, that in if dif-the NS-NS or BH-NS 2 Universitytype of than California G1, Observatories/ where neutron Lick Observatory, stars would University be formed of copiously:ferent for longprogenitor and short bursts, hypothesis would imply for short-hard that the intrinsic bursts, the host galaxies California, Santa Cruz, CA 95064. relative rates differ from those observed. While a number of bursts 3 Chandramagnetic Fellow. field decay would cut the active lifetime for megaflare may be a range of Hubble types (e.g., Livio et al. 1998). Com- 4 have been triangulated through the Interplanetary Network (see Harvard-Smithsonianactivity after Center10 for4 Astrophysics,yr. 60 Garden Street, Cambridge, pact merger systems coalesce in appreciable rates from Myr to MA 02138. Hurley et al. 2005b) on roughly day-long timescales, there has 5 Gyr after a starburst (e.g., Fryer et al. 1999; Bloom et al. 1999). Institute for Advanced Study, Olden Lane, Princeton, NJ 08540. only been one precisely localized short-hard burst relayed to 6 KIPAC, Stanford University, P.O.8. Box CONCLUSIONS 20450, Mail Stop 29, Stanford, ground observersObviously, in less than the 1 hr longer (GRB the 050202, time sinceSwift;Tueller the starburst, the larger the CA 94309. 18 distance a binary system will travel before coalescence. A clear 7 et al. 2005); owing to its proximity to the Sun at the time of DepartmentWe of Astronomy have monitored and Astrophysics, the location 525 Davey of Laboratory, GRB 050509b Penn- at optical prediction from this model is that as more short bursts are lo- sylvania Stateand University, infrared University wavelengths Park, PA 16802. from 8 minutes to 8 days afterlocalization, the sparse ground-based follow-up was undertaken. 8 Department of Physics and Astronomy, 4129 Frederick Reines Hall, Including GRBcalized, 050509b, those this associatedcorresponds with to a later-typeratio of 1:18 galaxies for of a given mass University oftrigger California, and Irvine, found CA no 92697-4575. indication of variability at the locationshort-hard of to long-softshould burst be preferentially detections with closerSwift, to much the smaller star formation centers of 9 Centerthe for Space fading Research, X-ray Massachusetts source, the Institute first of solid Technology, X-ray Cam- detectionthan of the an BATSEthe result. host;19 that is, we expect a more concentrated distribution bridge, MAafterglow 02139-4307. of a short-hard burst. Near the location of this source 10 Department of Physics, 1 Shields Avenue, University of California, Davis, As with long-durationaround a spiralbursts, galaxy the distribution with the of same short mass bursts as an early type. On CA 95616-8677.we and others have found an apparent group of faint blueappears gal- very nearly isotropic (Kouveliotou et al. 1993; Briggs 11 Departmentaxies of at Physics, redshifts 366k LeConte1.3. While Hall, University it is indeed of California, plausible thatet al. this 1996), and their brightness distribution ( V/Vmax 0:35) is Berkeley, CAshort 94720-7300. burst arose from a progenitor connected with those gal- 26 See http://www.ligo.caltech.edu/docs/G/G990111-00.pdf.h i 12 Carnegie Observatories, 813 Santa Barbara Street, Pasadena, CA 91101. 13 Hubble Fellow. 14 Space Sciences Laboratory, 7 Gauss Way, University of California, 18 There have been a few other short bursts (durationP2s)detectedandwell Berkeley, CA 94720-7450. localized, but with soft spectra and hence not members of the short-hard class. 15 Department of Physics, University of Michigan, Ann Arbor, MI 48109-1090. For example, GRB 040924 was a soft, X-ray–rich GRB (Fenimore et al. 2004; 16 Institute for Nuclear and Particle Astrophysics, Lawrence Berkeley Na- Huang et al. 2005). Hereafter, we use the term ‘‘short burst’’ interchangeably with tional Laboratory, Berkeley, CA 94720. short-hard burst. 17 Spitzer Science Center, Mail Stop 314-6, California Institute of Tech- 19 As of 2005 May 20, Swift has localized two short bursts out of a total of nology, Pasadena, CA 91125. 38; see http://swift.gsfc.nasa.gov/docs/swift/archive/grb_table.html. 354 A tough act to follow

Joshua Bloom, assistant professor of astronomy at the University of California, Berkeley

Vol 437|6 October 2005|doi:10.1038/nature04142 LETTERS

A short g-ray burst apparently associated with an elliptical galaxy at redshift z 5 0.225 N. Gehrels1, C. L. Sarazin2, P. T. O’Brien3, B. Zhang4, L. Barbier1, S. D. Barthelmy1, A. Blustin5, D. N. Burrows6, J. Cannizzo1,7, J. R. Cummings1,8, M. Goad3, S. T. Holland1,9, C. P. Hurkett3, J. A. Kennea6, A. Levan3, C. B. Markwardt1,10, K. O. Mason5, P. Meszaros6, M. Page5, D. M. Palmer11, E. Rol3, T. Sakamoto1,8, R. Willingale3, L. Angelini1,7, A. Beardmore3, P. T. Boyd1,7, A. Breeveld5, S. Campana12, M. M. Chester6, G. Chincarini12,13, L. R. Cominsky14, G. Cusumano15, M. de Pasquale5, E. E. Fenimore11, P. Giommi16, C. Gronwall6, D. Grupe6, J. E. Hill6, D. Hinshaw1,17, J. Hjorth18, D. Hullinger1,10, K. C. Hurley19, S. Klose20, S. Kobayashi6, C. Kouveliotou21, H. A. Krimm1,9, V. Mangano12, F. E. Marshall1, K. McGowan5, A. Moretti12, R. F. Mushotzky1, K. Nakazawa22, J. P. Norris1, J. A. Nousek6, J. P. Osborne3, K. Page3, A. M. Parsons1, S. Patel23, M. Perri16, T. Poole5, P. Romano12, P. W. A. Roming6, S. Rosen5, G. Sato22, P. Schady5, A. P. Smale24, J. Sollerman25, R. Starling26, M. Still1,9, M. Suzuki27, G. Tagliaferri12, T. Takahashi22, M. Tashiro27, J. Tueller1, A. A. Wells3, N. E. White1 & R. A. M. J. Wijers26

Gamma-ray bursts (GRBs) come in two classes1: long (>2 s), soft- GRB survey made with the Burst and Transient Source Experiment spectrum bursts and short, hard events. Most progress has been (BATSE). The 15–150 keV fluence is (9.5 ^ 2.5) 1029 erg cm22, made on understanding the long GRBs, which are typically which is the lowest imaged by BAT so far and is just£ below the short observed at high redshift (z < 1) and found in subluminous GRB fluence range detected by BATSE (adjusted for the different star-forming host galaxies. They are likely to be produced in energy ranges of the two instruments). core-collapse explosions of massive stars2. In contrast, no short Swift slewed promptly and XRT started acquiring data 62 s after GRB had been accurately (<10 00 ) and rapidly (minutes) located. the burst (T 62 s, where T is the BAT trigger time). Ground- Here we report the detection of the X-ray afterglow from—and the processed dataþ revealed an uncatalogued X-ray source near the centre localization of—the short burst GRB 050509B. Its position on the of the BATerror circle containing 11 photons (5.7j significance due sky is near a luminous, non-star-forming elliptical galaxy at a to near-zero background in image) in the first 1,640 s of integration redshift of 0.225, which is the location one would expect3,4 if the time. The XRT position is shown with respect to the Digitized Sky origin of this GRB is through the merger of neutron-star or black- Survey (DSS) field in Fig. 1. A Chandra target-of-opportunity hole binaries. The X-ray afterglow was weak and faded below the observation of the XRT error circle was performed on 11 May at detection limit within a few hours; no optical afterglow was 4:00 UT for 50 ks, with no sources detected in the XRT error circle. detected to stringent limits, explaining the past difficulty in The light curve combining BAT, XRTand Chandra data are shown in localizing short GRBs. Fig. 3. The UVOT observed the field starting at T 60 s. No new The new observations are from the Swift5 satellite, which features optical/ultraviolet sources were found in the XRTþ error circle to the hard X-ray wide-field Burst Alert Telescope (BAT), and rapid V-band magnitude . 19.7 for t , 300 min. Joshua Bloom, assistant professor of astronomy at the University of California, Berkeley spacecraft slewing to point the narrow-field X-ray Telescope (XRT) Swift has provided the first accurate localization of a short GRB. and the Ultraviolet-optical Telescope (UVOT) at the burst. On 9 May No optical afterglow was detected to stringent limits (R-band 2005 at 04:00:19.23 UT, the BAT triggered and located GRB 050509B magnitude . 25 at 25 h; ref. 7). When the XRTerror circle is plotted on board6. The BAT location is shown in Fig. 1 (large red circle) and on the R-band image we obtained8 with the Very Large Telescope the light curves in Fig. 2. The event is a single short spike with (VLT), several faint objects are seen in the error circle, some of which duration of 40 ^ 4 ms. The burst has a ratio of 50–100 keV to 25– are extended and could be high-redshift galaxies9,10. It is possible the 25 keV fluences of 1.4 ^ 0.5, which is consistent with, but in the soft burst occurred in one of these. However, the centre of the XRT error portion of, the short/hard population detected by the first extensive circle lies only 9.8 00 away from the centre of the large E1 elliptical

1NASA/Goddard Space Flight Center, Greenbelt, Maryland 20771, USA. 2Department of Astronomy, University of Virginia, Charlottesville, Virginia 22903-0818, USA. 3Department of Physics and Astronomy, University of Leicester, Leicester, LE1 7RH, UK. 4Department of Physics, University of Nevada, Las Vegas, Nevada 89154-4002, USA. 5Mullard Space Science Laboratory, University College London, Dorking RH5 6NT, UK. 6Department of Astronomy and Astrophysics, Penn State University, University Park, Pennsylvania 16802, USA. 7Joint Center for Astrophysics, University of Maryland, Baltimore County, Baltimore, Maryland 21250, USA. 8National Research Council, 2101 Constitution Ave NW, Washington DC 20418, USA. 9Universities Space Research Association, 10211 Wincopin Circle, Suite 500, Columbia, Maryland 21044-3432, USA. 10Department of Astronomy, University of Maryland, College Park, Maryland 20742, USA. 11Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA. 12INAF—Osservatorio Astronomico di Brera, Via Bianchi 46, I-23807 Merate, Italy. 13Universita degli studi di Milano Bicocca, Piazza delle Scienze 3, I-20126 Milano, Italy. 14Department of Physics and Astronomy, Sonoma State University, Rohnert Park, California 94928, USA. 15INAF—Istituto di Astroﬁsica Spaziale e Cosmica, Via Ugo La Malfa 153, I-90146 Palermo, Italy. 16ASI Science Data Center, Via Galileo Galilei, I-00044 Frascati, Italy. 17SP Systems Inc., 7500 Greenway Center Drive, Greenbelt, Maryland 20770, USA. 18Niels Bohr Institute, University of Copenhagen, DK-2100 Copenhagen, Denmark. 19UC Berkeley Space Sciences Laboratory, Berkeley, California 94720-7450, USA. 20Thu¨ringer Landessternwarte Tautenburg, Sternwarte 5, D-07778 Tautenburg, Germany. 21NASA/Marshall Space Flight Center, NSSTC, XD-12, 320 Sparkman Drive, Huntsville, Alabama 35805, USA. 22Institute of Space and Astronautical Science, JAXA, Kanagawa 229-8510, Japan. 23Universities Space Research Association, NSSTC, XD-12, 320 Sparkman Drive, Huntsville, Alabama 35805, USA. 24Ofﬁce of Space Science, NASA Headquarters, Washington DC 20546, USA. 25Stockholm Observatory, Department of Astronomy, AlbaNova, 106 91 Stockholm, Sweden. 26Astronomical Institute “Anton Pannekoek”, University of Amsterdam, Kruislaan 403, 1098 SJ Amsterdam, The Netherlands. 27Department of Physics, Saitama University, Sakura, Saitama 338-8570, Japan. 851 © 2005 Nature Publishing Group Progenitors

Early indications that SHBs were from a diﬀerent population

Short-Hard GRBs (SHBs):

- More diffusely positioned around galaxies - More massive, earlier-type putative hosts - Consistent with NS-NS/NS-BH merger simulations

Bloom & Prochaska 06, Troja+07; Fong+09; Gehrels, Ramirez-Ruiz & Fox 09; Berger 14 - Coincident GW would be the only smoking gun

Bloom & Prochaska 06

FIGURE 2. Comparison of the cumulative offset distribution of long-soft (left) and short-hard GRBsP. O’Brien, P. Mészáros: this afternoon; W-f Fong tomorrow (right). The SHB sample is taken from Table 2 and the LSB samplecomprisestheﬁrst16burstswith known redshifts and offsets. The histograms are made assuming that the offsets are known precisely (ie., a δ-function at the measured offset), whereas the smooth curvesaccountfortheuncertaintyintheoffset measurements following the formalism of [35]. While SHBs appear qualitatively to be more diffusely located with respect to their putative hosts, the Kolomogorov-Smirnoffprobability that the observed SHB population was drawn at random from the observed LSB population is 38%. That is, the data do not yet support the qualitatively assessment that the locations of SHBs are substaintially different than LSBs. median merger times between 0.1 – 1 Gyr from starburst with systemic kick velocities of order 100 km s−1.Theexpectedoffsetsfrommergersisofcoursedependenton the host properties3.] Second, the locations of LSBs appeared to trace the location of the UV light of hosts, suggesting an intimate connection of LSBs with star formation. A. Fruchter presented results at the conference, making use of HST imaging of new GRBs, and showing a continued connection of LSBs with the light of their hosts. In particular, the locations of LSBs appear to prefer some of thehighestsurfacebrightness components of their hosts.

3 Since the hosts are relatively low mass, suggesting that double NSs could escape the host potenitial before coalescence, a tight concentration is especially disfavored. This inconsistency is entirely dependent upon the dominant channel for double NS production. Other production channels do posit small offsets of GRBs from star formation locations (e.g., [38]).

Constraints on the Diverse Progenitors of GRBs from the Large-Scale Environments May 22, 2018 6 able progenitor models predict offsets and locations withinhoststhatarefarfromrobust. So what may we conclude about how the current host observations of SHBs relative to LSBs reflect the progenitors? A secure statement is that "SHB hosts contain a generally older population of stars than LSB" and a reasonable, but somewhat more directed statement is that "SHBs come from an old stellar population whereasLSBsdonot."What we think is still a matter of debate is whether “the frequency of early-type to late-type hosts for SHB is consistent with NS–NS" (E. Ramirez-Ruiz discussed his views on this in the conference). Similar statements can be made of the offset distribution of SHBs: yes, there appears to be some differences between LSBs and SHBs but nothing stastisti- cally significant at this time. What appears relatively secure, based on the small offsets of three of 5 SHBs from low-mass galaxies is that: Progenitors The progenitors of SHBs cannot have both large systematic kicks (> 100 km s−1)atformationANDinherelargedelaytimesfromstarburst(> 1 Early indicationsGyr) .that SHBs were from a different population Making even stronger statements about the nature of the progShort-Hardenitors is notGRBs only (SHBs): ham- pered by small number statistics but in the lack of robust pre dictions from the models. But the above statement can be reworked in the form of a genericsetofpredictions: - More diffusely positioned around • In Long delay (>1Gyr)progenitorsscenarioswithkickstheoffsetsshouldagalaxies nti- correlate with host mass and correlate with average stellar- Moreage massive, earlier-type putative • In Short delay (< 1Gyr)progenitorstheoffsetsshouldcorrelatewithhostmahosts ss and anti-correlate with average stellar age. - Consistent with NS-NS/NS-BH If the progenitor lifetime of the SHBs is long and kicks are small,merger then the simulations bursts should correspond spatially to the oldest stellar populations in a given galaxies. Bloom & Prochaska For early-type 06, Troja+07; Fong+09; galaxies, the distribution would presumably follow the light of the Gehrels, galaxy. Ramirez-Ruiz In contrast, & Fox 09; Berger 14 the distribution in star-forming galaxies might be more concentrated in the spheroid (e.g., bulge of the Milky Way). - Coincident GW would be the only Those of us that have worked on creating ab initio predictionssmoking for NS–NS gun progenitors all basically agree on the offsets and delays provided we use aproductionchannel dominated by binaries where the helium stars do not fill their respective Roche lobe Bloom & Prochaska 06 before exploding as a SN. But other productions might dominate, leading to predictions of relatively short merger times and small offsets [38, 40]. Both T. Piran and J. Grindlay discussed other production channels in the conference. For example, if SHBs arise from FIGURE 2. ComparisonNS of binaries the cumulative made offset in distribution the centers of long-s ofoft globular (left) and clusters short-hard [41], GRBsP. O’Brien, then the P. offsetsMészáros: should this afternoon; scale W-f Fong tomorrow (right). The SHB samplewith is taken the from halo Table size 2 and of the LSB host sampl (ratherecomprisesthefirst16burstswith than exponential disk size) and the systemic kicks known redshifts and offsets. The histograms are made assuming that the offsets are known precisely (ie., a δ-function at the measuredcould offset), be small. whereas the This smooth results curvesaccountfortheuncertaintyintheoffset in diverse predictions of offsets and host demography for measurements followingthe the same formalism progenitors. of [35]. While SHBs appear qualitatively to be more diffusely located with respect to their putative hosts, the Kolomogorov-Smirnoffprobability that the observed SHB population was drawn at randomFostered from the by observed the presence LSB populat ofion a is 38%. curious That is, 100 the data sec do timescale not yet for prompt emission and support the qualitativelylacking assessment the that detectionthe locations of ofSHBs a Li-Paczy are substaintiallynski´ different minisupernova than LSBs. [42], the theory community is rapidly developing other progenitor models for SHBs such as WD–WD mergers [43] and accretion induced collapse of a NS [44, 45]. In this respect the connection of GRBs median merger timeswith between SNe 0.1 may – 1 not Gyr only from bestarburst a historical with systemic one, there kick velocities may be a deeper analogy: long-soft of order 100 km s−1.Theexpectedoffsetsfrommergersisofcoursedependenton the host properties3.]GRBs Second, are the to locations core-collapased of LSBs appearedsupernovae to trace as short-hard the location GRBs of are to Type Ia supernovae the UV light of hosts,(“LSB:CC::SHB:Ia”). suggesting an intimate Not connection only does of LSBs this with ring star true formation. for the hosts and locations (like Type A. Fruchter presented results at the conference, making use of HST imaging of new GRBs, and showing a continued connection of LSBs with the light of their hosts. In particular, the locationsConstraints of LSBs on appear the Diverse to prefer Progenitors some of th ofehighestsurfacebrightness GRBs from the Large-Scale Environments May 22, 2018 8 components of their hosts.

Constraints on the Diverse Progenitors of GRBs from the Large-Scale Environments May 22, 2018 6 AA52CH02-Berger ARI 30 July 2014 6:55

galaxies. A progenitor population of young magnetars, however, will exhibit a spatial coincidence with star-forming regions and offsets that track an exponential disk distribution. Delayed magnetars will not track young star-forming regions but will track the overall light distribution of their hosts owing to the lack of natal kicks. Finally, a dominant population of dynamically formed NS-NS binaries in globular clusters will track the overall globular cluster distribution around galaxies, extending to tens of kiloparsecs (Salvaterra et al. 2010).

7.1. The Offset Distribution Determining the locations of short GRBs relative to their host centers (offsets) and relative to the underlying light distribution in the rest-frame optical (stellar mass) and UV (star formation) bands requires the high angular resolution and superior depth of the HST. A comprehensive study based on HST observations of 32 short GRB host galaxies was carried out by Fong, Berger & Fox (2010) and Fong & Berger (2013; see also Church et al. 2011). The HST data, combined with ground-based optical afterglow observations (and, in two cases, Chandra observations), provide accurate offsets at the subpixel level and reveal the broad distribution shown in Figure 10.The projected offsets span 0.5–75 kpc with a median of about 5 kpc. These are about four times greater than the median offset for long GRBs (Bloom, Kulkarni & Djorgovski 2002) and about 1.5 times greater than the median offsets of core-collapse and Type Ia SNe (Prieto, Stanek & Beacom 2008). In addition, though no long GRBs and only 10% of SNe have offsets of 10 kpc, the fraction ∼ ! of short GRBs with such offsets is about 25%. For offsets of !20 kpc, the fraction of short GRBs is about 10%, but essentially no SNe exhibit such large offsets. Progenitors

1 Short GRBs 0.9 Short-Hard GRBs (SHBs): Long GRBs 0.8 NS-NS merger models - More diffusely positioned around Core-collapse SNe galaxies 0.7 More massive, earlier-type putative Ia SNe - 0.6 hosts - Consistent with NS-NS/NS-BH 0.5 merger simulations

0.4 Bloom & Prochaska 06, Troja+07; Fong+09; Gehrels, Ramirez-Ruiz & Fox 09; Berger 14

Cumulative fraction 0.3 - Coincident GW would be the only 0.2 smoking gun Annu. Rev. Astron. Astrophys. 2014.52:43-105. Downloaded from www.annualreviews.org Access provided by University of California - Berkeley on 05/20/18. For personal use only. 0.1 Berger 14

10–1 100 101 102 Projected physical ofset δR (kpc) P. O’Brien, P. Mészáros: this afternoon; W-f Fong tomorrow Figure 10 Cumulative distribution of projected physical offsets for short GRBs with subarcsecond positions (red )(Fong, Berger & Fox 2010; Fong & Berger 2013), compared with the distributions for long GRBs (black;Bloom, Kulkarni & Djorgovski 2002), core-collapse supernovae (SNe) ( green;Prieto,Stanek&Beacom2008), Type Ia SNe (blue; Prieto, Stanek & Beacom 2008), and predicted offsets for neutron star (NS) binaries from population synthesis models ( gray)(Bloom,Sigurdsson&Pols1999;Fryer,Woosley&Hartmann1999; Belczynski et al. 2006). Short GRBs have substantially larger offsets than long GRBs and match the predictions for compact object binary mergers. Reprinted from Fong & Berger (2013) with permission.

www.annualreviews.org Short-Duration Gamma-Ray Bursts 71 • Oddballs, or Nature is good at Making Bursts of Gamma rays

Relativistically Beamed X-ray Flashes (XRFs) Tidal Disruption Events - Lower-energy events Sw 1644+57

Soft-gamma Ray Repeaters (SGRs) - Long GRBs without March 5 Events Supernovae ~15 known GRBs as Probes

090423 090423080913 D. Perley 09 050904050904

•ISM/IGM/Host via Absorption Spectroscopy Chen+05, Savaglio+07, Prochaska+07 For Swift: •Reionization (Neutral Fraction vs Redshift) <7% of bursts from z>7 <15%median of bursts z ~ 2.1 from z>5 Miralda-Escudé 98, Bromm & Loeb 02, Kawai+05, Totani+06 •Signposts to Pop III stars in the early universe Bromm+00 median z ~ 2.1

Bloom+08 GRBs as Probes

090423 090423080913 D. Perley 09 050904050904

D. Perley 09 050904 090423

For Swift: <7% of bursts from z>7 <15%median of bursts z ~ 2.1 from z>5

median z ~ 2.1

Screenshot From My Talk at “Swift 5th Birthday” Meeting (18 Nov 2009) Multimessenger

Short-lived afterglow

GRB 080503

Perley, Metzger+08; Also, GRB 130603B, Tanvir+03 Multimessenger

Short-lived afterglow

GRB 080503

Perley, Metzger+08; Also, GRB 130603B, Tanvir+03 Multimessenger

GRB/GW 170817 - NS-NS merger produced a GRB & kilonova

A B C 60° x´ x´

90% 10 x Mpc 50% 90% 30° HL

HLV 5° x 0° D 50% Gemini+F2 JHKs 2017-08-27.97 IPN GBM

-30°

16h 12h 8h 10´´ -60° Kasliwal+17, Science

Abbott+17, ApJ, 848, 2, L13

E. Troja, V. Kalogera: this afternoon; L. Singer, T. Piran: tomorrow From a talk by me in Cefalu, 2008 IPN Vela Series CGRO (BATSE) Swift HETE-2 Fermi BeppoSAX INTEGRAL

High-Energy Era Afterglow Era Multimessenger Era

GRB 980425/SN 1998bw GRB 130603B GRB 090423 GRB 030329/SN 2003dh March 5 GRB 970508 GRB 080319b Events GRB 130427A Event Sw J1644+57 GRB 670702 (SGR 0525-66) GRB 970228 GRB 050509b GW/GRB 170817

& Memorial Paczyński & Rhoads 93 Meegan+92 Li & Paczyński 98 Theory MacFadyen & Woosley 99 Josh Bloom 50 Years of Gamma-Ray Bursts* * With a biased overemphasis on Neil & stuff I was involved in

Josh Bloom UC Berkeley @profjsb