Coincidence analysis between periodic source candidates in C
Coincidence analysis between periodic source candidates in C 6 and C 7 Virgo data C. Palomba (INFN Roma) for the Virgo Collaboration • I report on the ongoing work done in collaboration with Pia Astone and Sergio Frasca • Blind analysis of the data of runs C 6 and C 7 to search for gravitational signals emitted by isolated rotating neutron stars • Selection of candidates in the two data sets and coincidences between them. • Injection of simulated signals GWDAW 11 - Potsdam, 19/12/2006 1
‘Blind’ search • Assumes source position, frequency and spin-down are not known • The vast majority of neutron stars is not visible in the EM band • It is rather unlikely that known NS (pulsars) emit detectable signals • Local population of neutron stars must be taken into account • Blind searches cannot be performed with optimal methods due to the huge number of points in the parameter space • Hierarchical procedures strongly cut the needed computing power at the cost of a small reduction in sensitivity 2
Hierarchical method for ‘blind’ searches h-reconstructed data Data quality SFDB Average spect rum estimation peak map hough transf. candidates coincidences coherent step events candidates Frasca, Astone, Palomba, CQG 22, S 1013 2005 Astone, Frasca, Palomba, CQG 22, S 1197 2005 Palomba, Astone, Frasca, CQG 22, S 1255 2005 Presentation at MG 11 The procedure involves two or more data sets belonging to a single or more detectors 3
Parameter space • observation time • frequency band • frequency resolution • number of FFTs • sky resolution • spin-down resolution ~1013 points in the parameter space are explored for each data set 4
Candidates selection • On each Hough map (corresponding to a given frequency and spin-down) candidates are selected putting a threshold on the CR • The choice of the threshold is done according to the maximum number of candidates we can manage in the next steps of the analysis • In this analysis we have used • Number of candidates found: C 6: 922, 999, 536 candidates C 7: 319, 201, 742 candidates 5
• ‘Effective’ (i. e. after selection of candidates) sensitivity loss respect to optimal analysis: C 6: 2. 4 C 7: 1. 8 • False alarm probability: C 6: C 7: MC 1 st violin mode • Still candidates excess at many frequencies, even if some 6 cleaning has been done
red line: theoretical distribution 7
‘disturbed’ band ‘quiet’ band Many candidates appear in ‘bumps’ (at high latitude), due to the short observation time, and ‘strips’ (at low latitude), due to the symmetry of the problem 8
Coincidences • To reduce the false alarm probability; reduce also the computational load of the coherent “follow-up” • Done comparing the set of parameter values identifying each candidate • Coincidence windows: • Number of coincidences: 2, 700, 232 • False alarm probability: band 1045 -1050 Hz 9
Detection of injected signals • 66 signals injected in C 7 data, with frequency in [50, 550]Hz and no spin-down, to study efficiency and accuracy in parameter estimation of the incoherent step • We make coincidences between candidates found in C 7 data + injections and the injected signals • Coincidence windows: 1964 candidates many sources undetected green curve: expected C 7 sensitivity 10
• To check if the short observation time plays a role, we dilate time by a factor 80 (and reduce spin-down of injected signals by the same amount) 5257 candidates • Good agreement with the expected sensitivity. • Accuracy in latitude is only slightly affected by the length of the observation time • Longer time interval increases the detection efficiency 11
• Given two or more data sets, we can suitable mix them in order to produce data sets covering a larger time interval time • If the two sets are created with nearly equal sensitivity, we have a further gain time See presentation at MG 11 for more details 12
Good correlation between signals amplitude and CR of candidates Strongest sources are detected with more accurate position Worse accuracy for low frequency sources (due to lower resolution) 13
• As already noted, sources at low ecliptic latitude are detected with worse accuracy. This is basically independent on the observation time. • We expect to have an improvement by using the adaptive Hough transform, which breaks the symmetry respect to the ecliptic. 14
Conclusions • Well established procedure for going from data to candidates • We make coincidences to reduce false alarm and computational load of the coherent step • Need for stretch of data covering a time interval as large as possible to have better detection efficiency • Uncertainty in latitude will be reduced by using the adaptive Hough transform • Need to extend the injections to non zero spin-down 15
Spare slides 16
17
18
N=1 N=2 N=3 N=4 19
20
- Slides: 20