Skip to content

Detection statistics

Reed Essick requested to merge detection-statistics into master

This patch implements a few detection statistics. Specifically, it implements single-IFO and network versions of

  • realizations of statistics (drawn from appropriate distributions)
    • matched filter S/N (maximized over amplitude separately in each IFO)
    • matched filter S/N (maximized amplitude and coalescence phase separately in each IFO)
  • probablity that of detection based on thresholds on observed S/N conditioned on true (optimal) S/N
    • matched filter S/N (maximized over amplitude separately for each IFO)
    • matched fitler S/N (maximized over amplitude and coalescence phase separately in each IFO)
  • write up derivation of these results and include it within the repo (next to the source code)
  • write and perform some basic tests
    • test/test-observed-snr
    • test/test-detection-probability

Note! there are correlations between the single-IFO SNRs and the network SNRs so that one cannot construct quantities like P( (single > X) and (network > Y) ) from what is currently stored: P(single > X) and P(network > Y). However, one can still estimate more complicated detection criteria by direct monte-carlo integration with an appropriate conditional in an EventGenerator. Similarly, the single-IFO probabilities are all independent, so one can compute quantities like P( at least 2 single > X) from the appropriate combinations of the stored single-IFO probabilities.

Fixes #21 (closed).

Edited by Reed Essick

Merge request reports

Loading