Background estimation in searches for binary inspiral Patrick Brady - - PowerPoint PPT Presentation

G030637-00-Z

Background estimation in searches for binary inspiral

Patrick Brady Inspiral Working Group LIGO Scientific Collaboration

11/10/2003 GWDAW -- 2

G030637-00-Z

1

Inspiral search pipeline

• Data Quality cuts applied up

front

» Analyze only science mode data

• Multi-interferometer follow up
• n coincident triggers.

» (Talk by Bose)

• Single interferometer vetoes

applied to triggers from each interferometer

» (Talk by Christensen and Shawhan)

• Background Estimation

» Time slide Hanford data relative to Livingston data

11/10/2003 GWDAW -- 3

G030637-00-Z

Coincidence criteria

• Triggers from Livingston are used to search the

corresponding Hanford data

• Check for coincidence between the Livingston and

Hanford triggers.

• The coincidence algorithm was:

» 1. Triggers must be coincident in time to within 11 ms. » 2. Each mass parameter in the template must be the same to within 0.03 solar masses. » 3. Compare distances measured at Livingston (DL) and Hanford (DH)

11/10/2003 GWDAW -- 4

G030637-00-Z

Tuning the amplitude cut

• Errors in distance estimates

expected to decrease with increasing SNR

• ε and κ tunable constants

|DL - DH|/DL < ε/ρH + κ Relax amplitude cut to allow detection of these Milky Way events

Missed injections in coincidence

DH/DL

11/10/2003 GWDAW -- 5

G030637-00-Z

Time-lag results: preliminary results using small bank

Livingston SNR LLO triggers in coincidence with LHO LHO triggers in coincidence with LLO Hanford SNR

Take a closer look at these candidates

11/10/2003 GWDAW -- 6

G030637-00-Z

Time-lag analysis: SNR in each detector

Large SNR in Livingston coincident with low SNR at Hanford

Livingston SNR Hanford SNR

Coherent two detector SNR=10

11/10/2003 GWDAW -- 7

G030637-00-Z

Time-lag analysis: triggers versus injections

Livingston SNR Hanford SNR

Raise SNR threshold in Hanford to 6 Simulated injections

• f binary inspirals in

Andromeda

11/10/2003 GWDAW -- 8

G030637-00-Z

Time-lag analysis: full bank & adjusted thresholds

• No. of distinct

candidates after 1 second clustering Two out of 18 time-lags have coincidence in templates of same mass

11/10/2003 GWDAW -- 9

G030637-00-Z

Time-lag analysis: full bank & adjusted thresholds

11/10/2003 GWDAW -- 10

G030637-00-Z

Concluding remarks

• Where to next:

» Complete ~100 time-shifts on playground data » Check for problems with background estimation using time-shifts » Run time-shift analysis on all data to be used in search

– determine network SNR threshold – compare to extrapolation from playground

• Search for binary inspiral

» Use background estimate from time-shifts on full data set » Search for candidate events » Determine rate limit based on efficiency, background, and number

• f detections