1. The Long and the Short of Gamma-Ray Bursts Kevin Hurley UC Berkeley Space Sciences Laboratory

2. EXAMPLE “…I was asked to give this talk in Italy, on a small Venetian island called San Servolo. It sounded so good, I couldn’t refuse. But the long and the short of it is, that when I got there, I discovered that the place used to be an insane asylum!*” *http://www.sanservolo.provincia.venezia.it/english/news/museoManicomio.asp

3. THE BATSE GRB DURATION DISTRIBUTION (McBreen et al. 1994)

4. THE IPN AND SWIFT GRB DURATION DISTRIBUTIONS

5. WHICH DISTRIBUTION IS CLOSEST TO THE TRUE ONE? Distributions are instrument-dependent: –Trigger criteria, detector sizes, efficiencies as a function of energy, sensitivities IPN is a mix of many instruments, and sensitivity is limited (not good) Swift is very sensitive, has good trigger criteria, but its high energy coverage is limited by both the detector and the mask efficiencies (better, but not perfect) BATSE had good trigger criteria, good high energy efficiency, and observed many bursts. Their distribution is probably closest, but by no means without strong selection effects

6. BASED ON DURATION >25 keV ALONE, WHAT IS THE DIVIDING LINE BETWEEN LONG AND SHORT GRBS? The lognormal fit to the two distributions predicts that a burst with T 90 <5 s has a probability <10 -3 of being a long duration burst This is roughly a 3σ criterion A 10 -6 probability criterion would be T 90 <0.9 s But duration should not be used by itself if more data are available ~ ~ ~

7. WHY ARE THE SHORT AND LONG BURST DISTRIBUTIONS SO BROAD?

8. Define the relative width as Then the relative width of the short burst distribution is 2.95, and the relative width of the long burst distribution is 2.91 They are equally broad, in relative terms (Same conclusion using variance 0.5 /mean) Is this because we are using the “raw” durations, i.e. uncorrected for redshift and systematics? FWHM Mode

9. DURATION DISTRIBUTIONS OF 70 GRBs WITH MEASURED T 90 AND REDSHIFTS

10. MEANSTANDARD DEVIATION BEFORE REDSHIFT CORRECTION 116 s335 s AFTER REDSHIFT CORRECTION 65 s312 s 1. The mean is smaller after redshift correction, as expected 2. The standard deviation does not change substantially after redshift correction!

11. IS THE WIDTH OF THE DURATION DISTRIBUTION MAINLY INTRINSIC? A variable has a lognormal distribution when it is the outcome of many multiplicative factors For GRB durations, this might include progenitor mass, surrounding ISM, details of accretion onto the black hole But this is not the only effect that shapes the distributions –S/N considerations for the long duration bursts may introduce a factor ~2 uncertainty in T 90 (Bonnell et al. 1997) –a simple 1+z correction ignores other factors (e.g. pulse width as a function of energy)

12. OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT, IN ORDER OF INCREASING COMPLEXITY (Donaghy et al. astro-ph/0605570) 1.Duration 2.Spectral hardness

13. Hardness is good for distinguishing short SGR bursts (not giant flares) from GRBs But it is a weak criterion for distinguishing long and short GRBs

14. 1.Duration 2.Spectral hardness 3.Extended soft emission OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT, IN ORDER OF INCREASING COMPLEXITY (Donaghy et al. astro-ph/0605570)

15. Mazets et al. 2002 Present in some BATSE bursts (Lazzati et al. 2001; Connaughton 2002) Present in some Konus-Wind bursts (Mazets et al. 2002; Frederiks et al. 2004) Present in some Swift and HETE bursts (Barthelmy et al. 2005; Villasenor et al. 2005) Possibly a defining characteristic of short bursts, but requires good low energy coverage and/or intense burst to detect it

16. OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT (Donaghy et al. astro-ph/0605570) 1.Duration 2.Spectral hardness 3.Extended soft emission 4.Spectral lag

17. Norris 2002 Norris & Bonnell. 2006 Lag vs. flux for 1429 long BATSE bursts Lag vs. flux for 260 short BATSE bursts Average lag ~ 50 ms Average lag ~0.1±0.5 s (consistent with 0) If a burst has small lag with good significance, it is probably short; if it has significant positive lag, it’s probably long If T 90 <2 s and lag is consistent with zero, it’s probably a short burst

18. 1.Duration 2.Spectral hardness 3.Extended soft emission 4.Spectral lag 5.Pulse widths OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT, IN ORDER OF INCREASING COMPLEXITY (Donaghy et al. astro-ph/0605570)

19. PULSE WIDTHS (Norris et al. 1996) Decompose the GRB time history into many individual pulses described by I(t)=A exp[-( ‌ t-t max ‌ / σ rise ) ν ], t < t max I(t)=A exp[-( ‌ t-t max ‌ / σ decay ) ν ], t > t max Pulse widths for short bursts are tens of milliseconds long Pulse widths for long bursts are hundreds of milliseconds long Don’t try this at home!

20. 1.Duration 2.Spectral hardness 3.Extended soft emission 4.Spectral lag 5.Pulse widths 6.Energy radiated in gamma-rays OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT, IN ORDER OF INCREASING COMPLEXITY (Donaghy et al. astro-ph/0605570)

21. EγEγ In general, short bursts radiate 10 48 – 10 49 erg, while long bursts radiate 10 50 – 10 51 erg Redshift obviously required Evidence for beaming useful, but not always required

22. SOME WELL-STUDIED SHORT BURSTS DATEDETECTED BY zE γ, ergBEAMED?HOST, LOCATION 790613IPN0.09?*6x10 49 iso?? 050509BSwift0.225?*2.7x10 48 iso?ELLIPTICAL? 050709BHETE0.163x10 48 14ºSPIRAL, OUTSKIRTS 050724Swift0.2589.9x10 49 iso?ELLIPTICAL, OUTSKIRTS 050813Swift0.7221.7x10 50 iso?? 050906Swift0.031?*1.2x10 47 iso?? 060121HETE1.5/4.6?*3/1.3x10 49 2.3º/0.6ºSPIRAL? *Based on probability argument

23. 1.Duration 2.Spectral hardness 3.Extended soft emission 4.Spectral lag 5.Pulse widths 6.Energy radiated in gamma-rays 7.Location within host galaxy OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT, IN ORDER OF INCREASING COMPLEXITY (Donaghy et al. astro-ph/0605570)

24. LOCATION WITHIN HOST GALAXY Long bursts occur in star-forming regions Location in the outskirts  older stellar population But a location within a galaxy wouldn’t necessarily contradict this GRB050709B; Fox et al. 2005 GRB050724; Berger et al. 2005

25. 1.Duration 2.Spectral hardness 3.Extended soft emission 4.Spectral lag 5.Pulse widths 6.Energy radiated in gamma-rays 7.Location within host galaxy 8.Type of host galaxy/age of stellar population OTHER WAYS OF DETERMINING WHETHER A BURST IS LONG OR SHORT, IN ORDER OF INCREASING COMPLEXITY (Donaghy et al. astro-ph/0605570)

26. SOME WELL-STUDIED SHORT BURSTS DATEDETECTED BY zE γ, ergBEAMED?HOST, LOCATION 790613IPN0.09?*6x10 49 iso?? 050509BSwift0.225?*2.7x10 48 iso?ELLIPTICAL? 050709BHETE0.163x10 48 14ºSPIRAL, OUTSKIRTS 050724Swift0.2589.9x10 49 iso?ELLIPTICAL, OUTSKIRTS 050813Swift0.7221.7x10 50 iso?? 050906Swift0.031?*1.2x10 47 iso?? 060121HETE1.5/4.6?*3/1.3x10 49 2.3º/0.6ºSPIRAL? *Based on probability argument

27. Host galaxy which is not undergoing star formation is probably a good indication of a short burst But short bursts can also occur in galaxies which are still forming stars Gorosabel et al. 2006

28. CONCLUSIONS 10 years ago, distinguishing long and short bursts was easy – hardness and duration were the only criteria required Today, it’s much harder – up to 9 criteria should be used, and some are ambiguous We should probably avoid the terms “short-hard bursts” and “long-soft bursts” entirely Donaghy et al. (2006) divide the criteria into gold (lag, extended emission, location, host type), silver (duration, pulse width, Eγ), and bronze (spectral hardness) standards But the platinum standard may turn out to be detection of gravitational radiation And that’s the long and the short of it