Title: Allsky search of Nautilus data: data analysis tools
1All-sky search of Nautilus data data analysis
tools
- Andrzej Królak
- Albert Einstein Institute
- Golm
- and
- Institute of Mathematics
- Warsaw
2Collaboration
- MOU to search NAUTILUS detector 2001data for
periodic gravitational-wave signals. - MOU involves
- 1. ROG collaboration led by Eugenio Coccia
- (contact person Pia Astone).
- AEI continuous GW group led by Maria Alessandra
Papa. - A.K. group in Poland.
3(No Transcript)
4Parameters of the search
- 1 year of data divided into 2-day sections.
- 2 day sections analyzed coherently using F
-statistic. - Bandwidth of 1-2Hz around 1kHz.
- 1 spin down parameter with minimum spin down time
of 1000 years. - All-sky search.
5Upper limit
pD probability of detection C confidence
(say 90) L number data stretches searched
6Upper limit if no signal detected
- Threshold SNR d 8.2 for 1 false alarm
probability - Upper limit for dimensionless GW amplitude for
the search of 2 day stretch of data when the
detector was most sensitive (Sh 5 x 10-22
Hz-1/2) - hUL 5.7x10-23
- Upper limit for dimensionless GW amplitude for
the search of 150 of independent 2 day stretches
of data (Sh 10-21 Hz-1/2) - hUL 1.8x10-23
7Maximum Likelihood Detection
- Response of the detector GW signal
- Optimal filtering F-statistic
- False alarm and detection probability
- Grid of templates
- Numerical calculation of the F-statistic
- Tools Fisher Information Matrix
- Correlation function cell concept
8Response of the detector
9Neglect amplitude modulation and introduce a
linear parameterization of the phase
Approximation of the response to calculate the
false alarm probability and construct the grid of
templates in the parameter space.
We neglect the component of the vector rD that
is perpendicular to the ecliptic. (rD joins the
center of the Earth and the detector).
10F-statistic for a periodic signal
11Number of templates
Spacing
Hypecubic covering
K no. of intrinsic parameters V (hyper)volume
of the parameter space Co- minimum correlation
coefficient G reduced Fisher matrix We choose
Co sufficiently large so that the loss of the
signals is minimized
12Grid of templates in the parameter space
No. of templates 108
13Calculation of the F -statistic
Resampling
FFT
14(No Transcript)
15CODE
- Code is being tested
- The aim is to achieve a real time performance,
- i.e. analyze 2 days of data
- (1 Hz bandwidth, 1 spin- down (1000yr), all-sky)
- in 2 days.
- Should be possible with a couple hundred of
- processors.