The Progenitors of Short-Hard GRBs from an Extended Sample of Events Avishay Gal-Yam Hubble Fellow CALTECH

Outline - Observational breakthrough An extended sample from the IPN - Population analysis – hosts - Final remarks

Short GRBs (SHBs) - Papers: Nakar, Gal-Yam, Piran & Fox 2005, ApJ Fox et al. 2005, ApJL, b Fox et al. 2005, Nature, Oct. 6, Berger et al. 2005, Nature, Nov. issue, Gal-Yam et al. 2006, ApJ, Submitted, astro-ph/ Nakar et al. 2006, ApJ, in press, astro-ph/ Also: Bloom et al Prochaska et al Tanvir et al …

People: E. Nakar (Caltech) T. Piran (HUJI), D. Fox (Penn State), E. Ofek (Caltech) (Extended) Caltech GRB group (Kulkarni, Frail, Berger, Cenko…)

The new SHB (Swift-HETE-Bursts) era Arcmin -> arcsec localizations -> hosts 4 relevant bursts, as of August 2005

GRB B: Swift Detection Very faint GRB X-ray: T+62 s detects 11 photons(!) No optical, no radio. Very faint limits. Giant elliptical galaxy in a cluster. z=0.22. Host? No SN Gehrels et al T 90 =40 ms

HST Imaging: No Supernova Fox et al Error radius = 9.3 arcsec 4 HST Epochs May 14 to June sources in XRT error circle Giant elliptical (Bloom et al 2005) L=1.5L * SFR<0.1 M  yr -1

SHB : HETE Detection A Hard spike, 84 keV A Soft bump Roughly equal energy in each component Villasenor et al T 90 =70 ms

SHB localization X-ray afterglow Optical afterglow Secure association with a galaxy: late-type, z=0.16, moderate SFR (MW), lots of old stars (>1 GY) - Covino et al No SN

GRB : Swift Detection Hard spike/soft bump X-ray, optical, IR and radio afterglow detected Secure association with elliptical galaxy, z=0.26 No SN (Berger et al. 2006) T 90 =40 ms keV keV T 90 =3 s 250 ms 100 s Gal-Yam et al. 2005

SHB : Swift Detection Another Swift SHB Deep imaging identifies a high-z cluster (Gladders et al. 2006) Initial spectroscopy shows z=0.722 (Berger 2005, Prochaska et al. 2005) (but perhaps revised higher, z=1.8)

Recent Swift/HETE sample: Low redshift Most in early types No SNe So: Different from long GRBs Similar to SNe Ia Long lived progenitors “A merger origin”?

Extended Sample from IPN Goal: Search for luminosity over-density inside or around old, small, IPN error boxes (either single luminous galaxies or galaxy overdensities - clusters) Sample: 4 small error boxes Method: multicolor imaging (P60, LCO100), spectoscopy (P200) Comparison with SDSS luminosity functions and galaxy densities.

SHB Smallest IPN box (<1 arcmin 2 ) 4 reddish galaxies with similar colors – very high density (~1% prob.) 6.5’ away from Abell 1892 (z=0.09), <1% chance association in our entire sample Therefore assume: z=0.09 (nearest SHB), host likely early type, in cluster 26 years … !

SHB Small IPN box (5.6 arcmin 2 ) A single luminous galaxy (R=17.57) Sb galaxy at z=0.14, ~1.4 L* in SDSS r,i Chance association 2% for this error box, 7% for entire sample Therefore assume: z=0.14, intermediate spiral host

SHBs , SHB : Small IPN box (6 arcmin 2 ) Several galaxies Inconsistent redshifts (0.31, 0.39), insignificant overdensity z>0.25 (1  ) SHB : insignificant overdensity z>0.25 (1  ) Z=0.31

Extended SHB sample RedshiftHostSHB 0.22E050509b 0.16Sbc/Sc E >=0.72E/S E/S Sb > >

Comparison with SNe Ia SNe Ia occur in galaxies of all types But most of them occur in late-type galaxies ! The normalized SN rate in Irr galaxies is 20 times that in E galaxies Comparison of the observed relative rates disfavors an identical distribution (P=7%) It appears like SHBs come from older progenitors (several Gy) (also: Zheng & Ramirez-Ruiz 2006) Mannucci et al Gal-Yam et al. 2006

Do we rule out NS-NS mergers? No. Main limit is small numbers (both SHBs and observed NS-NS pairs) But, you really need to stretch to fit the data Observations disfavor DNS models This is not changed by z>0.72 and (Soderberg et al. 2006) Belczynski et al E Soderberg et al. 2006

Concluding remarks Similar results from redshift distribution (Nakar, Gal-Yam & Fox 2006) Recent SHBs useless for similar analysis due to sample incompleteness (~1/10 with data) and obvious strong selection effects (in favor of gas-rich systems) Additional observational effort imperative for further progress Possible high local rates of SHBs (See Nakar, Gal-Yam & Fox) mean the LIGO (I or II) may provide conclusive evidence for compact binary models

Thanks

The rate and progenitor lifetime of SHBs (Nakar, Gal-yam & Fox 2005) Goals: Using the extended sample to constrain the local rate and the progenitors lifetime of short GRBs. Evaluate the compatibility of these results with the compact binary progenitor model. Explore the implications for gravitational wave detection of these events with LIGO.

Method: Comparing the observed redshift and luminosity distributions to predictions of various models of intrinsic redshift and luminosity distributions. (this method is an extension of a method used by Piran 1992; Ando 2004; Guetta & Piran 2005,2006)

Cosmology + Detector ~400 BATSE bursts with unknown z Several bursts with known z Consistency test IntrinsicObserved Redshift distribution Luminosity function

Intrinsic Observed Cosmology Detector If  (L) is a single power-law: In the case of BATSE SHBS a single power-law fits the data very well:  (L)  L -2±0.1

Porciani & Madau 2001 Star formation rate Progenitor lifetime distribution + = Intrinsic redshift distribution

Results

 typical > 4[1]Gyr (95% [99.9%] c.l.) or if f(  )    then  > -0.5[-1] ( 95% [99.5%] c.l. ) *Similar results are obtained when we take z >0.72 **Similar results are obtained by Guetta & Piran 2006 Progenitor lifetime

Main uncertainties and limitations Luminosity function – any “knee” shaped broken power-law results in similar constraints. A short lifetime may be consistent with an “ankle” shaped broken power-law (expected in case of two populations of SHBs). Star-Formation History – the results are valid for any of the three Porciani & Madau (2001) SFH functions (all peak at z≈1.5). Detector Thresholds – important only if the luminosity function is not a single power-law (only Swift SHbs can be used). The results are valid for a range of reasonable threshold functions of Swift. Small sample – the results are only at a level of ~3  and might be affected by unaccounted selection effects or wrong measurements.

Observed Local Rate (and robust lower limit) -BATSE observed rate was  170 yr -1 -At least ¼ of these bursts are at D < 1Gpc Similar result is obtained by Guetta & Piran 2006

Total Local rate We consdider here  (L)  L -2 with a lower cutoff L min f b – beaming correction (30-50 Fox et al ???) L min – The current observation are insensitive to L min < erg/sec. Evidence for population of SHBs within ~100Mpc (Tanvir et al. 2005) suggests L min < erg/sec

Local rate – upper limit SHB progenitors are (almost certainly) the end products of core-collapse supernovae (SNe). The rate of core-collapse SNe at z~0.7 is 5×10 5 Gpc -3 yr -1 (Dahlen et al. 2004), therefore:

Observed NS-NS systems in our galaxy Based on three systems, Kalogera et al. (2004) find: in our galaxy And when extrapolating to the local universe: This rate is dominated by the NS-NS system with the shortest lifetime –  ~100 Myr (the double pulsar PSR J ). Excluding this system the rate is lower by a factor 6-7 (Kalogera et al. 2004).

SHBs and NS-NS mergers For the two to be compatible there should be a hidden population of old long-lived NS-NS systems. Can it be a result of selection effects? Maybe, but we cannot think of an obvious one. Caveat: small number statistics NS-NS (Kalogera et al. 2004) : 200< R NS-NS < 3000 Gpc -3 yr -1 Dominated by binaries that merge within ~100 Myr SHBs (Nakar et al. 2005) : 10< R SHB < 5·10 5 Gpc -3 yr -1 Dominated by old progenitors >4 Gyr

Detection of SHB increases LIGO range by a factor of (Kochanek & Piran 1993): Timing information (~1.5) Beaming perpendicular to the orbital plane (~1.5) Localization information

LIGO-I: Probability for simultaneous detection Swift detects and localizes ~10 SHBs yr -1. If L min ~10 47 erg/s and  (L)  L -2 then ~3% of these SHBs are at D<100 Mpc and ~1% at D<50 Mpc R ( merger+SHB ) ~ 0.1 yr -1 Notes: This result depends weakly on beaming In this scenario R SHB ~ 1000 f b Gpc -3 yr -1 Comparison with Swift and IPN non-localized bursts may significantly increase this rate

Thanks!

Single power-law fit to  (L) Maximum likelihood:  (L)  L -2±0.1  2 /d.o.f  1  a good fit The extended sample (8 SHB) can be used

Tanvir et al Long GRBs SHBs SHBs (E-Sbc galaxies) SHBsGalaxies at D<100Mpc At least 5% of BATSE SHBs are at D<100Mpc Our model predicts that 3% of the SHBs are at D<100Mpc if L min  erg/s

Broken power-law fit to  (L) Only Swift bursts can be used (unknown detector response for the rest). However we can carry a comparison with the two-dimensional L-z distribution For each f(  ),  1 and  2 we fit L * to BATSE dN/dP

 typical > 3Gyr or  >-0.5

Probability for blind search detection LIGO-I: Taking a speculative but reasonable SHB rate of 10 4 Gpc -3 yr -1 predicts a detection rate of: R ( NS-NS) ~ 0.3 yr -1 R ( BH * -NS) ~ 3 yr -1 *M BH ~ 10M  LIGO-II: The SHB rate lower limit of 10 Gpc -3 yr -1 implies: R ( NS-NS) ≥ 1 yr -1 R ( BH * -NS) ≥ 10 yr -1

GRB missions MissionOperationalSHB rate (yr -1 ) localized non-localized Swift ~10? HETE ~1- IPNyes~1~10 * GLAST2007-~30 * - *my rough estimate

LIGO-II: Probability for simultaneous detection This year 3 bursts detected at D < 1Gpc GLAST is expected to detect several SHBs at D<500Mpc every year LIGO-II range for simultaneous detection is ~700 Mpc (NS-NS) and ~1.3 Gpc (BH-NS) Simultaneous operation of LIGO-II and an efficient SHB detector could yield at least several simultaneous detections each year. Non-detection will exclude the compact merger progenitor model

Offset 39 ± 13 kpc 3.5 ± 1.3 kpc 2.4 ± 0.9 kpc Prochaska et al., 2005