Unified χ 2 \chi^{2} discriminators for gravitational wave searches from compact coalescing binaries
Sanjeev Dhurandhar · 2025
Read Unified χ 2 \chi^{2} discriminators for gravitational wave searches from compact coalescing binaries by Sanjeev Dhurandhar (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-18051.txt.
Introduction
Dhurandhar lays out in twenty twenty-five a unified maths for the test statistics used to weed out noise glitches in gravitational wave searches. Real waves from merging black holes and neutron stars are faint and buried in detector noise. Short glitches mimic signals and cut search power, so analysts project data onto spaces away from the signal shape and square the leftover size as a veto score.
He shows every such veto comes from one geometric structure: a bundle of leftover spaces over the family of possible signals. Any choice of leftover space yields a valid veto, and the bundle gathers them all. He builds from this an optimal veto for blip glitches shaped like sine bursts. The frame guides vetoes for other glitch shapes.
Catalog record
| CREATOR | Sanjeev Dhurandhar |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-18051
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Unified χ 2 \chi^{2} discriminators for gravitational wave searches from compact coalescing binaries Sanjeev Dhurandhar Abstract Gravitational wave (GW) signals of astrophysical origin are typically weak. This is because gravity is a weak force, weakest among the four forces we know of. In order to detect GW signals, one must make differential measurements of effective lengths less than thousand’th of the size of a proton. In spite of the detectors achieving extraordinary sensitivity, the detector noise typically overwhelms the signal, so that GW signals are deeply buried in the data. The challenge to the data analyst is of extracting the GW signal from the noise, that is, first deciding whether a signal is present or not then if present, measuring its parameters. However, in the search for coalescing compact binary (CBC) signals, short-duration non-Gaussian noise transients (glitches) in the detector data significantly affect the search sensitivity. Chi-squared discriminators are therefore employed to mitigate their effect. We show that the underlying mathematical structure of any χ 2 \chi^{2} is a vector bundle over the signal manifold 𝒫 {\cal P} , that is, the manifold traced out by the signal waveforms in the Hilbert space of data segments 𝒟 {\cal D} . The χ 2 \chi^{2} is then defined as the square of the L 2 L_{2} norm of the data vector projected onto a finite-dimensional subspace 𝒮 {\cal S} (fibre) of 𝒟 {\cal D} chosen orthogonal to the triggered template waveform. Any such fibre leads to a χ 2 \chi^{2} discriminator and the full vector bundle comprising the subspaces 𝒮 {\cal S} and the base manifold 𝒫 {\cal P} constitute the χ 2 \chi^{2} discriminator. We show that this structure paves the way for constructing effective χ 2 \chi^{2} discriminators against different morphologies of glitches. Here we specifically demonstrate our method on blip glitches, which can be modelled as sine-Gausians, which then generates an optimal χ 2 \chi^{2} statistic for blip glitches. I Introduction A spectacular prediction from Einstein’s general theory of relativity is the existence of gravitational waves [ 1 ] . However, gravitational waves (GW) are inherently weak because gravity is a weak force, the weakest among the four known forces. Minute differential interferometric path length measurements must be made to observe these waves with sufficient confidence. During the past several decades and in the more recent past, breakthroughs in technology have made it possible to build kilometre scale laser interferometric detectors on ground. Past six decades of heroic experimental efforts undertaken by physicists all over the world have finally culminated with the first direct observation of a GW signal announced by the Laser Interferometer Gravitational Wave Observatory (LIGO) project [ 2 , 3 ] . On September 14, 2015, the two LIGO interferometers at Hanford (Washington) and Livingston (Louisiana) simultaneously measured and recorded strain data that indicated the presence of a GW signal emitted by a coalescing binary a system containing two black holes of masses around 30 times the mass of the Sun. Since then many more detections, as many as 90 , have been made in the science runs O1, O2 and O3 [ 4 ] . Currently, the LIGO [ 5 ] , the Italian-French VIRGO [ 6 ] , and the Japanese KAGRA [ 7 ] are in operation, and the O4 run is in progress. The detector on Indian soil, namely, LIGO-India [ 8 ] , is expected to join in these efforts early next decade. Gravitational wave astronomy has truly arrived. However, the data in the ground-based detectors is neither Gaussian nor stationary; non-Gaussian and non-stationary features can be produced by various components of the detector itself or by the environment. In this article, we will only consider CBC signals emitted by a pair consisting of either black holes or neutron stars. These signals are essentially transient, lasting between a fraction of a second to several minutes. Further, the signals can be adequately modeled by using either post-Newtonian approximation methods, perturbation techniques, or numerical relativity. In the case of adequately modelled signals, the preferred data analysis technique is matched filtering. However, the signals depend on several parameters, such as the masses, time of coalescence, etc., and then it becomes necessary to employ a bank of templates densely covering the parameter space. In practice, however, matched filtering alone is not sufficient to identify a signal because the data contain non-Gaussianities and transient noise artifacts, also termed as glitches. Fig. 1 shows a typical example of a glitch that occurs in the data. Figure 1: The figure shows a typical glitch in the time-frequency plane. This is called a blip glitch because of its distinctive features. Even if the overlap of a glitch with a template is small, the glitches themselves can have large amplitudes and they can produce sufficiently loud triggers in the filtered output. Then, such triggers have a chance of being misinterpreted as signals. To remedy this situation, \chi^{2} vetos or discriminators were proposed. The first \chi^{2} discriminator [ 9 ] proposed by Bruce Allen, which we call traditional , is designed to distinguish between a signal and a glitch by producing a high value of the \chi^{2} statistic if the trigger arises from a glitch, while a low value of \chi^{2} indicates the presence of a signal. The statistic is based on how the power of the signal is distributed in various frequency bins. This is done by dividing the data into several frequency bins and checking whether this power distribution is consistent with that of the signal. This is why, it is also called the power \chi^{2} . We briefly review the traditional \chi^{2} due to Bruce Allen’s in section III.3 . However, this is not the only type of \chi^{2} discriminator that can be constructed; it can be constructed from general considerations [ 10 , 11 ] ; these works show that a plethora of \chi^{2} s can be constructed. We call this the generic or the unified \chi^{2} . For example, another \chi^{2} that is also employed is the Sine-Gaussian (SG) \chi^{2} [ 12 ] . The generic \chi^{2} discriminator is discussed in section III and specifically in subsection III.1 . We show that the traditional \chi^{2} can be brought under the common umbrella of the generic \chi^{2} discriminators. In section IV we construct the optimised \chi^{2} which we call Opt \chi^{2} in short. In section V we compare the different \chi^{2} s. Finally, in section VI we conclude. In the next section II , we discuss the context and some preliminaries. II The context In this section we describe the matched filtering technique and how it is employed in the detection of CBC signals. We then discuss the statistical method of deciding on detection and significance. The treatment here follows the frequentist approach. Currently, Bayesian methods are in vogue. However, we do not discuss them here. In fact, gravitational wave astronomy is in its infancy, and it seems somewhat premature to choose priors; we have to rely on our knowledge gained from electromagnetic astronomy, and it could give a misleading picture. The gravitational wave universe could be far different from the universe as we know it from other astronomical windows. II.1 The matched filtering paradigm Consider