Документ взят из кэша поисковой машины. Адрес оригинального документа : http://zebu.uoregon.edu/~uochep/talks/talks04/BURST-~1.pdf Дата изменения: Mon Feb 7 21:58:44 2005 Дата индексирования: Tue Oct 2 10:51:27 2012 Кодировка:

Triggered (Pulsed) Burst Search Method Update
I. Leonor (UofO), on behalf of R. Rahkola (UofO)


Motivation to look for pulsed bursts
!

! !

Some modeled GWs assoc. with external triggers exhibit pulsed bursts ( 3 bursts, each ~few ms) Wish to avoid matched filtering, explore larger phase space Use cross-correlation to reduce local noise

Dimmelmeier-Font-MЭller SN wvfrm A1B1G1

LIGO Scientific Collaboration

Ott et. al. SN wvfrm ww-s15a10000.4

2


Method I: cross-correlation w/ pulse-train mask
!

Multiply two IFO timeseries w/ fixed (0?) offset

!

Construct pulse-train mask:
Parameters: » #_pulses » pulse_width » pulse_spacing » start_time (scanned)

LIGO Scientific Collaboration

3


Method II: piece-meal integration
!

Integrate cross-correlated timeseries over all pulses
» Note: Do not integrate between pulses

value of `s' for parameter choice (#_pulses, pulse_width, pulse_spacing, start_time)

» »

» » »

»
st ar t_ tim e

»

LIGO Scientific Collaboration

This is the `full-mask' statistic (s) Can also construct `inverse-mask', `singlepulse', and `ratio' statistics Computed for all start_time parameters Is randomly scattered about 0 w/ only noise When a signal is present in both IFOs, statistic leans slightly positive. Sensitive to non-zero mean in data (|s| increases with pulsewidth and #_pulses) Aggressive, zerophase high-pass filter improves this 4


Method III: Constructing SNR
! ! !

Assume µs = 0 Variance 2s ~ (#_pulses)
» SNR = s / (#_pulses)

!

Can use off-source segments to set nominal background level Choice of parameters has little effect (so far)

LIGO Scientific Collaboration

5


SNR behavior
!

Parameter choices:
» #_pulses: 3 » pulse-width: 5ms » pulse-spacing: 15ms

LIGO Scientific Collaboration

6