Searching for EM counterparts to neutron star mergers

David Coward

22017 NewCompStar

Presentation Outline

Advanced(LIGO(low0latency(searches(

Jonah(Kanner((for(the(LIGO(Scien;fic(Collabora;on(and(the(Virgo(Collabora;on(

(

LIGO%G1600392,

AAS(Mee;ng(San(Diego,(CA(June(13,(2016(

• Motivation for coincident observations of gravitational waves and optical

counterparts from neutron star mergers

• emissions from a short gamma ray burst

• How an automated follow-up should work: GRBs as a superb example

• first attempts of an EM follow-up of a gravitational wave source

• Considerations for the first EM follow-up of a neutron star merger

32017 NewCompStar

• Combining GW and EM observations of transients will unleash a new era of astronomy

• Optical data will pinpoint the location of the GW sources, enabling an understanding of their environment and formation history (even if they are not connected to GRBs)

• Gamma ray bursts: the “smoking gun” for the existence of binary neutron star mergers?

• Rates of short gamma ray bursts and NS mergers are compatible but uncertain:

• realistic estimates of some tens of mergers per year, and a GRB optical afterglow coincidence rate of at least several yr−1 (see Coward et al. 2012).

Motivation for chasing EM signatures from neutron star mergers

42017 NewCompStar

gamma ray burst emissions

An ideal automated follow-up: GRB example

Swift 2-3 degree error

XRT location X-ray light curves 2 arcmin localisation

Robotic Telescopes: Transient optical source identified Photometry of early emissions > 10 seconds

Redshift obtained from spectroscopy of the host galaxy or directly from the afterglow about 1 hour

First GRB optical counterpart localised in 1997 (in X-ray)…after thousands

of GRBs were detected in the early 1990s….many years to achieve localisation and redshifts

6

Anatomy of a GRB (note for short GRB emission time is < 1 sec)

Gamma ray (prompt emission) + x-ray + optical afterglow (Zadko Telescope)

72017 NewCompStar

Advanced(LIGO(low0latency(searches(

Jonah(Kanner((for(the(LIGO(Scien;fic(Collabora;on(and(the(Virgo(Collabora;on(

(

LIGO%G1600392,

AAS(Mee;ng(San(Diego,(CA(June(13,(2016(

LOCALIZATION AND BROADBAND FOLLOW-UP OF GW150914 3

100 101 102

t � tmerger (days)

Initial GWBurst Recovery

InitialGCN Circular

Updated GCN Circular(identified as BBH candidate)

Finalsky map

Fermi GBM, LAT, MAXI,IPN, INTEGRAL (archival)

SwiftXRT

SwiftXRT

Fermi LAT,MAXI

BOOTES-3 MASTER Swift UVOT, SkyMapper, MASTER, TOROS, TAROT, VST, iPTF, Keck,Pan-STARRS1, KWFC, QUEST, DECam, LT, P200, Pi of the Sky, PESSTO, UH

Pan-STARRS1VST TOROS

VISTA

MWA ASKAP,LOFAR

ASKAP,MWA

VLA,LOFAR

VLA,LOFAR VLA

Figure 1. Timeline of observations of GW150914, separated by band and relative to the time of the GW trigger. The top row showsGW information releases. The bottom four rows show high-energy, optical, near-infrared, and radio observations, respectively.Optical spectroscopy and narrow-field radio observations are indicated with darker tick marks and boldface text. Table 1 reportsmore detailed information on the times of observations made with each instrument.

search are reported in Abbott et al. (2016c).

3. SKY MAPS

We produce and disseminate probability sky maps using asequence of algorithms with increasing accuracy and compu-tational cost. Here, we compare four location estimates: theprompt cWB and LIB localizations that were initially sharedwith observing partners plus the rapid BAYESTAR localiza-tion and the final localization from LALInference. All fourare shown in Fig. 2.

cWB performs a constrained maximum likelihood estimateof the reconstructed signal on a sky grid (Klimenko et al.2016) weighted by the detectors’ antenna patterns (Essicket al. 2015) and makes minimal assumptions about the wave-form morphology. With two detectors, this amounts to re-stricting the signal to only one of two orthogonal GW polar-izations throughout most of the sky. LIB performs Bayesianinference assuming the signal is a sinusoidally modulatedGaussian (Lynch et al. 2015). While this assumption may notperfectly match the data, it is flexible enough to produce reli-able localizations for a wide variety of waveforms, includingBBH inspiral-merger-ringdown signals (Essick et al. 2015).BAYESTAR produces sky maps by triangulating the times,amplitudes, and phases on arrival supplied by all the CBCpipelines (Singer & Price 2016). BAYESTAR was not avail-able promptly because the low-latency CBC searches werenot configured for BBHs; the localization presented here isderived from the offline CBC search. LALInference performsfull forward modeling of the data using a parameterized CBCwaveform which allows for BH spins and detector calibra-tion uncertainties (Veitch et al. 2015). It is the most accuratemethod for CBC signals but takes the most time due to thehigh dimensionality. We present the same LALInference mapas Abbott et al. (2016e), with a spline interpolation proce-

dure to include the potential effects of calibration uncertain-ties. The BAYESTAR and LALInference maps were sharedwith observers on 2016 January 13 (GCN 18858), at the con-clusion of the O1 run. Since GW150914 is a CBC event, weconsider the LALInference map to be the most accurate, au-thoritative, and final localization for this event. This map hasa 90% credible region with area 630 deg2.

All of the sky maps agree qualitatively, favoring a broad,long section of arc in the southern hemisphere and to a lesserextent a shorter section of nearly the same arc near the equa-tor. While the majority of LIB’s probability is concentratedin the southern hemisphere, a non-trivial fraction of the 90%confidence region extends into the northern hemisphere. TheLALInference sky map shows much less support in the north-ern hemisphere which is likely associated with the strongerconstraints available with full CBC waveforms. The cWB lo-calization also supports an isolated hot spot near ↵ ⇠ 9h, � ⇠5�, where the detector responses make it possible to indepen-dently measure two polarization components. In this region,cWB considers signals not constrained to have the ellipticalpolarization expected from a compact binary merger.

Quantitative comparisons of the four sky maps can be foundin section 2 of the Supplement (Abbott et al. 2016g). Themain feature in all of the maps is an annulus with polar an-gle ✓HL determined by the arrival time difference �tHL be-tween the two detectors. However, refinements are possibledue to phase as well as amplitude consistency and the mildlydirectional antenna patterns of the LIGO detectors (Kasli-wal & Nissanke 2014; Singer et al. 2014). In particular, thedetectors’ antenna patterns dominate the modulation aroundthe ring for un-modeled reconstructions through a correlationwith the inferred distance of the source (Essick et al. 2015).As shown in Fig. 2, the algorithms all infer polar angles thatare consistent at the 1� level.

Latency for EM follow-up of the first GW source

Low latency (but low sensitivity all sky) from “shadowing” same sky as LIGO. Fermi (all sky) candidate also from shadowing.

82017 NewCompStar

LOCALIZATION AND BROADBAND FOLLOW-UP OF GW150914 5

Figure 3. Footprints of observations in comparison with the 50% and 90% credible levels of the initially distributed GW local-ization maps. Radio fields are shaded in red, optical/infrared fields are in green, and the XRT fields are indicated by the bluecircles. The all-sky Fermi GBM, LAT, INTEGRAL SPI-ACS, and MAXI observations are not shown. Where fields overlap, theshading is darker. The initial cWB localization is shown as thin black contour lines and the LIB localization as thick black lines.The inset highlights the Swift observations consisting of a hexagonal grid and a selection of the a posteriori most highly rankedgalaxies. The Schlegel et al. (1998) reddening map is shown in the background to represent the Galactic plane. The projection isthe same as in Fig. 2.

instruments, depth, time, and sky coverage. Some optical can-didate counterparts were followed up spectroscopically and inthe radio band as summarized in Table 2. The overall EMfollow-up of GW150914 consisting of broad-band tiled ob-

servations and observations to characterize candidate coun-terparts are described in detail in Sections 3 through 5 of theSupplement (Abbott et al. 2016g).

Findings from these follow-up observations have been re-

Compared to the first identification of a GRB optical counterpart,

huge localisation uncertainty for the first GW detections

There is reason to be optimistic

Significant global facilities

are eager to participate

in the first detections

strong international interest: e.g. China

The Good, the Bad and the Ugly

92017 NewCompStar

Low0latency(Transient(Searches(

4(

Mo;va;on(•  Validate(found(events(•  Inform(noise(hun;ng(•  Enable(Mul;0messenger(astronomy(

Transient(Searches(

and(follow0up(

Event((Database(

Human((Valida;on( Alerts(

to(partners(

Triggers(reach(database(in(minutes( ((( ((

Validation bottleneck for O1 and improved for O2

For GRBs this is automated from the SNR of detector, seconds latency

102016 Astronomical Society of Australia Annual Meeting

GCN type alert to

EM partners

Human

download sky probability maps

Human receives

alerts

From Alert to follow-up, the brute force pipeline

Human

schedules

telescope

LIGO Validation (Human)

Optimized

algorithms for

scheduling

5e-11 2.41e-07 1.21e-06 5.06e-06 2.04e-05 8.12

15:00:00.020:00.040:00.016:00:00.000.0

16:00:00.0

20:00:00.0

24:00:00.0

28:00:00.0

Tile observation test on GW150914

• Probability map• Tile observation

GW150914 Bayestar Probability Map

Example implementation of code

• Repeated imaging of the same sky location in the optical at similar sensitivities.

• considerable delay (days) before a GW Alert sent, because of human vetoing of the signal

• Despite over 60 participating facilities, less than a third managed to acquire data

• negative declination localisation region is less sampled.

• Evidence for a geographic bias in follow-up capabilities. Australia can capitalise on this window!

Issues from the first EM follow-up of a GW alert

LOCALIZATION AND BROADBAND FOLLOW-UP OF GW150914 5

Figure 3. Footprints of observations in comparison with the 50% and 90% credible levels of the initially distributed GW local-ization maps. Radio fields are shaded in red, optical/infrared fields are in green, and the XRT fields are indicated by the bluecircles. The all-sky Fermi GBM, LAT, INTEGRAL SPI-ACS, and MAXI observations are not shown. Where fields overlap, theshading is darker. The initial cWB localization is shown as thin black contour lines and the LIB localization as thick black lines.The inset highlights the Swift observations consisting of a hexagonal grid and a selection of the a posteriori most highly rankedgalaxies. The Schlegel et al. (1998) reddening map is shown in the background to represent the Galactic plane. The projection isthe same as in Fig. 2.

instruments, depth, time, and sky coverage. Some optical can-didate counterparts were followed up spectroscopically and inthe radio band as summarized in Table 2. The overall EMfollow-up of GW150914 consisting of broad-band tiled ob-

servations and observations to characterize candidate coun-terparts are described in detail in Sections 3 through 5 of theSupplement (Abbott et al. 2016g).

Findings from these follow-up observations have been re-

132016 Astronomical Society of Australia Annual Meeting

Summary

• EM counterparts to NS mergers are transients: implications - a global distribution of follow-up facilities is required.

• Several deep and wide field facilities do not cover the temporal space required.

• Coordination between follow-up facilities optimal: non-trivial (political and technically)

• Australia: follow-up opportunities - Southern sky and temporal niche.

• Western Australia is close to the geographic antipode of LIGO, follow-up of the brightest GW sources observable from the southern hemisphere.

• Automated alerts and robotic pipelines for optimal imaging may be essential

• EM counterparts to GWs from neutron star mergers will revolutionise astronomy: Let’s go!

14

ASKAP 36 identical antennas, each12 metres in diameter, 0.7 to 1.8 GHz, FoV 30 square degrees

MWA 2048 dual-polarization dipole antennas 80-300 MHz frequency, FoV 200 - 2500 sq. deg.

SkyMapper optical 1.35m robotic, FoV about 3 x 3 sq. deg.

Zadko Telescope - optical, 1m robotic dedicated follow-up of transients, FoV 0.5 x 0.5 deg

Australian radio and optical facilities for EM follow-up

152016 Astronomical Society of Australia Annual Meeting

Exciting Post Doctoral position at UWA

Theme: Multi-Messenger astronomy

GW data analysis

Coincident detection of gamma ray bursts and GWs

Fast Radio Bursts

16

Important updateOzGrav is born:

September 2016: Australian Government provides funding for a Centre of excellence for gravitational wave Astronomy

A consortium of national and international partners, OzGrav, (including Italy participants) is born, to consolidate Australia’s role in gravitational astronomy

Goal: to unite individual GW research groups into a single collaboration

Research themes include:

• multi-messenger astronomy

• GW source modelling

• Pulsar timing

• data analysis

• 3rd generation detector technology

172017 NewCompStar

6(6(

Triggers(and(significance((

Waveform(reconstruc;on(and(parameters(

Source(posi;on(reconstruc;on(

From shaking mirrors to sky positions

many “events” in O1 data rejected because the False Alarm Rate (FAR) too high

….some of these could be mergers… beyond horizon distance

EM counterparts ?

NS mergers detected

within a smaller horizon

compared to BH mergers

300 Mpc

LIGO + Virgo needed to

reduce localisation

182016 Astronomical Society of Australia Annual Meeting

Probability for imaging optical counterpart of a NS binary merger

assuming short GRB afterglow

Probability depends on FoV, sensitivity limit and localisation

Assuming a 100 square degree uncertainty as an example