Optimal vetoing methods for removing glitches that can be modelled from interferometric detector data

Abstract

The LIGO and VIRGO Detectors have conducted two observation runs, in which a total of ten binary black-hole mergers and neutron star mergers have been detected. The third observation run has already seen several potential candidates for CBCs. Search algorithms involve cross-correlation of data from detectors with a template and searching with a template bank in parameter space. This method works very well, if noise in the detector data is Gaussian and stationary which is not the case. The data is extremely noisy and contains a lot of glitches which produce false triggers. Tests that have been introduced to act as vetos to discriminate between real signals and glitches are called the χ 2 tests. The most successful among them is called the power χ 2 . But these tests have also been unable to perform well against all the types of glitches. Thus, development of better χ 2 tests is a very important problem. The unified χ 2 formalism is a rigorous mathematical framework which can be used to study and construct new χ 2 tests. I will use this formalism to construct an optimal χ 2 test specifically to discriminate between signals and short, transient glitches that can be modelled using sine-gaussian functions. I will then demonstrate its performance as compared to the power χ 2 test using simulated data.