Very Low Latency Search for Low Mass Compact
- Slides: 12
Very Low Latency Search for Low Mass Compact Binary Coalescences in the LIGO S 6 and Virgo VSR 2 Data Damir Buskulic for the LIGO Scientific Collaboration and the Virgo Collaboration LIGO-G 0901053, VIR-0012 A-10 January 26 -29, 2010 GWDAW 14
Very low latency CBC searches l l During science run S 6 of LIGO and VSR 2 of Virgo from July '09 Very low latency search (this talk) LIGO – Hanford u NOT baseline search used for publications of detections or upper limits u Low mass range : from 1 to 34 M , » 2 M < total mass < 35 M u Higher threshold analysis u More limited consistency tests u Focus on triple coincidences for multi-detector analysis l LIGO – Livingston Two main goals u Extract single detector triggers for real time detector characterization Virgo » Monitoring of trigger rate and data quality ➜ see talk by F. Robinet u Online multi-detector search » Quickly identify and localize in the sky interesting triple coincident candidates that deserve an electromagnetic follow-up January 26 -29, 2010 GWDAW 14 2
Why electromagnetic follow-ups ? l l NS-NS or NS-BH mergers are plausible progenitors of short, hard gamma ray bursts A GRB + GW coincident observation could u Confirm this hypothesis u Give great confidence in GW detection u Bring additional information about the source » Accurate sky position, host galaxy, redshift… l l Searches triggered by short, hard GRBs are part of LIGO-Virgo analyses, but… GRBs are believed to result from collimated outflows u u Beaming factor reduces chance of observation Many GRBs could be observed only through their afterglows (orphan afterglows) Afterglow ~15 times more likely to be observed Worth triggering afterglow search on GW trigger » Timescale of afterglows – hours – compatible with this approach January 26 -29, 2010 GWDAW 14 3
Follow-up instruments l l External collaborations established with other instruments Target of opportunity observations with Swift (CBC point of view) u Look for afterglows in X-ray, UV and optical domains u Expect ~3 requests during Swift Cycle 6 (starting April 2010) u Most likely to be due to detector noise, could plausibly contain a true signal u One of the triggering candidates could be a test » “Blind” hardware injections to probe detection process l Wide-field optical followups u Look for electromagnetic counterpart using array of wide-field optical telescopes u Examples : TAROT, QUEST, Pi of the Sky January 26 -29, 2010 GWDAW 14 4
The MBTA pipeline l Multi-band template analysis M 2 u Efficient implementation of matched filtering over a bank of templates » Computes matched filtering integral over two frequency bands » Coherently adds SNR from low and high frequency bands u 2 nd order post-Newtonian, time-domain templates u Adaptive mechanisms to follow detector non-stationnarities u Speed, speed… » Limited but computationally inexpensive consistency test – 2 band 2 » No files involved in the data transfer between processes – Use TCP-IP based protocol developed for Virgo DAQ January 26 -29, 2010 GWDAW 14 M 1 M 2 M 1 5
Online implementation l l Bring all data @ Virgo Processing split by: u ITF & Mass range l Single ITF clustering u Cluster triggers separated by less than ~0. 1 second l H 1: h(t) + basic data quality L 1: h(t) + basic data quality V 1: h(t) + basic MBTA process Trigger Generation MBTA process + local clustering Trigger Generation + local clustering + local time clustering MBTA Trigger MBTA proc + local Trigger Gene + local cl + local time cl Single ITF clustering over mass Multi-ITF clustering u Based on time » HL time window: ~20 ms » H/L-V time window: ~40 ms u Clustering can apply time shifts for background estimation More data quality u Provide 1, 2 or 3 sites triggers l Frame Merger + Coincidence step + Coincidence with time shifts Follow up: Cluster quality, direction Background estimate; Plots The triggers journey continues… January 26 -29, 2010 Single ITF c over m Alert to Control rooms GWDAW 14 Web pages for Multi-sites trigger Archive 6
A trigger's journey l l After generation of a coincident trigger Sent to a database ( "Grace. DB" ) u Record candidate event u Record information about followup u Connected to an alert system (not yet enabled) l l Activation of a sky localization procedure Data Quality u Data quality flags and vetoes are produced online and can be used by the very low latency analyses u Before an alert is sent to the outside world, some basic follow-up of the candidate will be done » Procedure can be automated to some extent » Validation should be done by scientist on shift January 26 -29, 2010 GWDAW 14 7
Sky localization l Use triangulation based on time of flight between H 1, L 1, V 1 detectors to locate the source on the sky Scan the sky and identify those points the signal is most likely to come from Example with simulated signal u For better accuracy, use time when signal crosses some reference frequency ~150 Hz instead of end time u Use effective distance measured at each detector to help lifting the symmetry ambiguity u Modest pointing accuracy » ~ several degrees for signals at detection threshold January 26 -29, 2010 Use hardware injections to assess performance of sky localization GWDAW 14 8
Latency h(t) available at the same location MBTA Single ITF/mass triggers available Single ITF clustering over mass All single ITF triggers available Multi ITF coincidence Coincident triggers available for follow-up Latency from data to coincident triggers : < 3 min January 26 -29, 2010 GWDAW 14 9
Controlling the trigger rate l Goal : Tune threshold for ~1 trigger/month to be considered for possible follow-up by Swift l Cuts on Time Of Flight and Chirp Mass u Efficiency and performance checked with software injections u Single detector trigger time resolution = u Coincidence windows » Hanford - Livingston: ± 20 ms Hanford/Livingston - Virgo: ± 40 ms l Estimation of the trigger rates u Assuming single ITF trigger rate = 0. 1 Hz u Expect/observe » ~1. 5 events/hour for H 1 L 1 coincidences, » ~3 events/hour for H 1 V 1 or L 1 V 1 u When applying chirp mass cuts » Expect/observe less than 1 H 1 L 1 V 1 triple coincident event/month January 26 -29, 2010 GWDAW 14 u Triple coincidence rate low, but double coincidences can be used to check how well the background can be estimated from trigger rates 10
Figures of merit H 1 Trigger rates triple coincidences L 1 V 1 SNR, Chirp mass, 2 distributions Combined H 1 -L 1 -V 1 range for single detector triggers (example of H 1) MBTA duty cycle (first two months) January 26 -29, 2010 Trigger rates H 1 L 1 V 1 HL HLV 94% 96% 97% 93% 92% GWDAW 14 11
Conclusion l Compact coalescing binaries involving a neutron star are potentially observable also as GRBs and/or their afterglows l Very low latency searches may be a key point in making a joint GW + electromagnetic observation u Allow to trigger search for EM counterpart on GW candidates l A very low latency pipeline (MBTA) was running during S 6/VSR 2 runs u Latency < 3 min until availability of trigger u Duty cycle of the pipeline for three detector coincidences > 90 % over the full run u Expect/observe less than 1 triple coincident event/month l Trigger submission for E. M follow-up will be enabled after Virgo data taking resume. January 26 -29, 2010 GWDAW 14 12
- Plastic compact semi compact slender
- Cmaf vs webrtc
- Low latency dispersion compensator
- Lola low latency
- Low latency microservices in java
- Lola streaming
- Kay ousterhout
- Generic low latency
- Very bad to very good scale
- Multiplication of scientific notation
- Fewfew
- Is a very shallow skillet with very short sloping sides
- What is food quantifiers