LIGO-XXX

Short GRBs and Mergers: Astrophysical constraints on

a BH-NS and NS-NS origin

Richard O’Shaughnessy[V. Kalogera, C. Kim, K. Belczynski, T. Fragos]

PSU, May 14, 2007

LIGO-XXX

Context?

If you want to…• Test GR soon (=LIGO)

– Need high SNR, clean merger waveforms– EM coincidence helps search (orientation, time)

• Understand short GRBs– May not all be mergers– Merger waves or absence distinguishes

• Improve models for binary stellar evolution– Merger rate is constraint– Short GRB observations could be a constraint…– Consistency with non-GRB observations?

LIGO-XXX

Outline• Short GRBs : A Review

– Intersection with LIGO

• Population synthesis predictions– Milky Way astro-ph/0610076; 0609465

– Universe

• Could short GRBs be mergers? – Detection rates consistent?– Redshift distribution, hosts?

• Where are we now? What happens next? – EM: Swift + GLAST and biases– GW: Initial and Enhanced LIGO?

LIGO-XXX

Short GRBs: A Review

Short GRBs• One of two (?) classes• Cosmological distances

– Low redshift selection effect?• Hard: often peaks out of band• Flux power law

dP/dL ~ L-2

--> most (probably) unseen

Many sources at limitof detector (BATSE)

[Berger et al, astro-ph/0611128]

LIGO-XXX

Short GRBs: A ReviewMerger motivation?• No SN structure in afterglow

• In both old, young galaxies

•Occasional host offsets

GRB 051221 (Soderberg et al 2006)• Young NSs are some (known) Energetics suggest not all

GRB 050709 (Fox et al Nature 437 845)

LIGO-XXX

Short GRBs: Review

• Gravitational waves essential– Central engine? : Certainty requires gravitational waves

• See inspiral• Check masses

– Coincident observation powerful [e.g., merger-burst delay time; opening angle constraints; masses; NS radius; …]

– Nondetection still useful[e.g., find fraction of short bursts from NS alone nearby]

• Short GRBs : potentially powerful tool?– Constrain channels: Short GRBs >> 10/yr; #(NS-NS)=4

LIGO-XXX

But are Short GRBs Mergers?Direct test from one event?

EM: Fireball problem -- can’t see central engine Enormous merger model diversity (e.g opening angles)

GW: LIGO range short vs usual GRB distance [see later]

Statistics - look for obvious inconsistencies?Predict #, distribution of mergers Implies #, distribution of short GRBs Compare + reject inconsistent models

LIGO-XXX

StarTrack and Population SynthesisPopulation synthesis:• Evolve representative sample• See what happens

Variety of results Depending on parameters used…• Range of number of binaries per input mass

Priors matter a priori assumptions about what parameters likely influence expectations

O’Shaughnessy et al (in prep)

More binaries/mass

Plot: Distribution of mass efficiencies seen in simulations

LIGO-XXX

StarTrack and Population SynthesisPopulation synthesis:• Evolve representative sample• See what happens

Variety of results Depending on parameters used…• Range of number of binaries per input mass • Range of delays between birth and merger

Priors matter a priori assumptions about what parameters likely influence expectations

O’Shaughnessy et al (in prep)

Merging after 2ndsupernova

Merging after 10 Gyr

Plot: Probability that a random binary merges before time ‘t’, for each model

: changed priors since last paper

LIGO-XXX

Popsyn and Milky Way

Population synthesis• Controlled uncertainties --> wide but limited range of predictions

Milky Way: A test• ~ steady state system (average merger rate)

• Compare to observations (several Kim et al) (NS-NS binaries + known selection effects)

– Observation: shaded– Theory: dotted curve– Systematics : dark shaded

• Limited set (9%) consistent– Complicated, extended 7d volume– Lots of physics can be mined

More binaries/mass

astro-ph/0610076

LIGO-XXX

Milky Way Binary PulsarsObservations

– 7 NS-NS binaries – 4 WD-NS binaries

Selection effects“How many similar binaries exist, given we see one?”

Examples• Lifetime :

– age + merger time < age of universe• Lifetime visible :

– time to pulsar spindown, stop?• Fraction missed - luminosity:

– many faint pulsars Distribution of luminosities ~ known

• Fraction missed - beaming: – Not all pointing at us!

Kim et al ApJ 584 985 (2003)Kim et al astro-ph/0608280Kim et al ASPC 328 261 (2005)

Kim et al ApJ 614 137 (2004)

Example: Lmin correction:

One seen --> many missed

Rate estimate Kim et al ApJ 584 985 (2003)

(steady-state approximation)

Number + ‘lifetime visible’ + lifetime + fraction missed

=> birthrate + error estimate (number-> sampling error)

Note: • Only possible because many single pulsars seen:

Lots of knowledge gained on selection effectsApplied to reconstruct Ntrue

from Nseen

LIGO-XXX

Inhomogeneous universe: The reality

– Time-dependent, multicomponent SFR

– Use delay time distribution

(dP/dt ~ 1/t)

– Long delays matter

Sample multicomponent predictions:

• Merger rate in spirals

(NS-NS)

From recentPlot:

Birth time forpresent-day mergers

Popsyn and Universe

Merging after 2ndsupernova

Merging after 10 Gyr

LIGO-XXX

Can short GRBs be mergers?Binary pulsars

• Many (isolated) observed• Minimum luminosity ~

known• Observed number --> rate (+ ‘small’ error)

Short GRBs• Few observations• Minimum luminosity ~ unknown• Observed number --> rate upper bound

Conclusion:The number (rate) of short GRB observations is

a weak constraint on models

observed

Plots:Cartoon on Lmin

LIGO-XXX

Can short GRBs be mergers?

Test 1: Are there enough mergers? … so far, usually yes:• Plot: All-sky detection rate vs predictions, if

+ No bursts fainter than seen+ All sky coverage & no beaming

… but surprising if detectors

“fine-tuned” many should be missed

BH-NS

NS-NS

LIGO-XXX

Can short GRBs be mergers?

Test 2: Are they distributed consistently

in redshift? (BH-NS shown)

BH-NS?:• Predictions:

– 500 pairs of simulations

– Range of redshift distributions

• Observations:– Solid:

certain

– Shaded:

possible

O’Shaughnessy et al (in prep)

KeySolid: 25-75%Dashed: 10-90%Dotted: 1%-99%

LIGO-XXX

Can short GRBs be mergers?

Test 2: Are they distributed consistently

in redshift? (BH-NS shown)

• Predictions that agree?– Compare cumulative distributions:

maximum difference < 0.48 everywhere

– Compare to well-known GRB redshifts since 2005• dominated by low redshift

O’Shaughnessy et al (in prep)

[95% Komogorov-Smirnov given GRBs]

[consistent selection effects]

Result:Distributions which agree = mostly at low redshift

LIGO-XXX

Can short GRBs be mergers?

Test 2: Are they distributed consistently

in redshift? (NS-NS shown)

• Predictions & observations

KeySolid: 25-75%Dashed: 10-90%Dotted: 1%-99%

O’Shaughnessy et al (in prep)

• Matching redshifts

• Observed NS-NS (Milky Way)

• All agree? - possible - special parameters needed (~1/100)

LIGO-XXX

Can short GRBs be mergers?

NS-NS?:• Physical interpretation

– Observations : GRBs • Dominated by recent events

– Expect: • Recent spirals dominate or

• or Ellipticals dominate, with long delays

O’Shaughnessy et al (in prep)

-Observations: Galactic NS-NS• High merger rate

-Expect-High merger rate in spirals

• Consistent so farMostly inellipticals

Mostly inspirals

Plot: fs : fraction of mergers in spirals (z=0)

LIGO-XXX

Short GRBs: Where are we with Swift?

Good: No longer clueless• Hosts : variety, most star forming• Redshifts : Mostly nearby

Bad• Afterglow searches biased against high redshift (Berger 2007)• Swift search biased against short bursts (Gehrels, Ringberg)

– Few events– Detection rate hard to interpret

• Narrow, strange sky coverage• No peak energiesSurprises• Afterglows look odd• Classification no longer trivial (e.g., long bursts w/ short spikes; long close bursts w/ no

SN; etc)

See Nakar 2007

astro-ph/0701748

[Berger et al, astro-ph/0611128]

GLAST will help

- less biased selection

- improve Swift triggering (e.g., lower flux thresholds) -> more statistics

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

LIGO-XXX

What will LIGO see?

When will LIGO see mergers? [supporting slide] [based on single-IFO NS range (e.g., 15 Mpc)]

Extrapolating from Milky Way: Whole universe: about the same(revised version astro-ph/0610076) (mature draft to be submitted; “raw data”

[totally unconstrained distributions] shown)

Initial

Advanced

NS-NS

BH-NS

Enhanced LIGO

(3x single-IFO range of 15 Mpc??)

R(NS-NS) ~ 0.05 - 1 / year

LIGO-XXX

Detection: A scenario for 2014Scenario: (Advanced LIGO)• Observe n ~ 30 BH-NS events [reasonable]• Rate known to within d log R ~2[1/n1/2ln(10)]~ 0.16

• Relative uncertainty down by factor d log R/ log R ~ 0.16/1

16% ~ 9% : Comparable to all EM observations

Repeat for BH-BH, NS-NS• Independent channels (each depends differently on model params)->

Volume [0.09 (0.16)3] ~ (3 x 10-4) !!Params [0.09 (0.16)3]1/7 ~ 0.32

LIGO-XXX

LIGO Nondetections Useful

SGRs (=B-driven bursting neutron stars) are GRBs • Known galactic/nearby source : SGR 1806• Unknown (small?) contribution to nearby short GRB rate

LIGO can “distinguish”:• Short GRB nearby (e.g., <15 Mpc)

– Merger : Detectable– SGR : Marginally/not detectable

• Application– Assist host galaxy searches (i.e., minimum distance to merger)– estimate SGR contribution

critical to apply spiral fraction as constraint

LIGO-XXX

Conclusions

• Useful comparison method despite large uncertainties

• Preliminary results

– Via comparing to pulsar binaries in Milky Way

– Via comparing to short GRBs?• Conventional popsyn works : weak constraints-> standard model ok • Expect GRBs in either host : spirals form stars now

– Spirals now favored; may change with new redshifts!

• Short GRBs = NS-NS? easier : few consistent ellipticals

• Short GRBs = BH-NS? harder : fewer observations

• Observational recommendations

LIGO-XXX

Supporting slides follow• LIGO and short GRBs : Nondetection still useful

• Swift detection biases

LIGO-XXX

Outline

• Predictions and Constraints: Milky Way– Observations (pulsars in binaries) and selection effects

– Prior predictions versus observations

– Constrained parameters

– Physics behind comparisons : what we learn

– Revised rate predictions

– What if a detection?

• Why Ellipticals Matter• Predictions and Constraints Revisited

LIGO-XXX

Accepted modelsConstraint-satisfying volume

7d volume: • Hard to visualize!• Extends over ‘large’ range: characteristic extent(each parameter): 0.091/7~0.71

7d grid

= 7 inputs to StarTrack

9% of models work

LIGO-XXX

Outline

• Predictions and Constraints: Milky Way• Why Ellipticals Matter

– Two-component star formation model

• Predictions and Constraints Revisited– Prior predictions– Reproducing Milky Way constraints

LIGO-XXX

From recentPlot:Birth time for

present-day mergers

Importance of early SFRLong delays allow mergers in ellipticals now• Merger rate from starburst: R ~ dN/dt~1/t• SFR higher in past:

• Result: – Many mergers now occur in ancient binaries Nagamine et al astro-ph/0603257\

From oldancient SFR = ellipticals (mergers, …)

LIGO-XXX

Outline

• Predictions and Constraints: Milky Way

• Why Ellipticals Matter

• Predictions and Constraints Revisited

• GRBs– Review + the short GRB merger model

– Short GRB observations, the long-delay mystery, and selection effects

– Detection rates versus Lmin

– Predictions versus observations:• If short GRB = BH-NS

• If short GRB = NS-NS

– Gravitational waves?

• Conclusions

LIGO-XXX

ConclusionsFuture (model) directions:• More comparisons

– Milky Way• Pulsar masses• Binary parameters (orbits!)• Supernova kick consistency?

– Extragalactic• Supernova rates

Some examples:Belczynski et al. (in prep)

• Broader model space–Polar kicks?–Different maximum NS mass

[important: BH-NS merger rate sensitive to it!]

–Different accretion physics

Goal: - show predictions robust to physics changes - if changes matter, understand why (and devise tests to constrain physics)