36th Annual Report (2023-24)

algorithmto search for CBCs ismatched filtering. Shreejit Jadhav, Mihir Srivastava (IIT Kharagpur) and Sanjit Mitra have recently developed a method towards that goal. While the method is not yet as efficient as Matched Filtering, it is a big step towards making a primarily Machine Learning based search. Also, this has the potential to work with the present Matched Filtering based analysis to perform a deeper search (that is, to find low significance events). [Mach. Learn.: Sci. Technol. 4 (2023) 045028] Sigmanet - A neural network to distinguish Binary Black Holes (BBH) vsglitches: Sunil Choudhary, Anupreeta More, Sudhagar S and Sukanta Bose designed SiGMaNet, a neural network, initially, to distinguish massive BBH from Blip glitch, a type of transient non- gaussian feature. This is done by generating projection maps of cross- correlations of triggers (either a BBH or a Blip glitch) with Sine-Gaussian functions. Since the projections of BBH and Blip glitches are different, they could successfully identify BBH from Blip glitches far better than traditional methods [Phys. Rev. D, 107, 024030 (2023)] . In the extension of this work, with SiGMaNet-2, they are broadening its applicability to a wider class of glitches and by training a new neural network on much larger samples [Narayan, More et al. in prep]. Pipeline development for GWB sea r ches : Un t i l now , t he LVK collaboration has been using mainly Matlab-based codes for isotropic GWB searches. The collaboration has now developed a Python-based search library for GWB searches [ApJ 952 25 (2023)]. This new code is faster and also easier to use. This will be useful for the wider GWB community outside of LVK. Much documentation has also been produced for this new code base to support that effort [JOSS 9(94), 5454 (2024)]. This code is now publicly ¬ Gravitational Wave Background (GWB) ¬ available. Shivaraj Kandhasamywas one of the lead developers of the new pipeline. Sukanta Bose and Sanjit Mitra were part of the LVK team that reviewed the code. GWB searches: With the ongoing improvements in the detector sensitivities, the detection of GWB is on the horizon. However, much work still needs to be done to extract all the information from the detected signal. The calibration process, which converts photo-detector signals at the output to GW strain, can affect such parameter estimation. Shivaraj Kandhasamy (IUCAA) and Junaid Yousuf (Univ. Kashmir) studied such effects quantitatively and showed that for the current generation of detectors with calibration uncertainties less than 10%, such effects are negligible [Phys. Rev. D 107, 102002 (2023)]. All-SkyAll-Frequency (ASAF) analysis: Sanjit Mitra and his group developed an analysis pipeline, PyStoch, to search for anisotropic GWB, which utilised years worth of data folded to one sidereal day utilising a mathematical symmetry, which was also developed by the same group. The folded data not only made the present LVK analysis hundreds of times faster, but enabled making skymaps at every frequency bin. The LVK collaboration devoted a full publication on All-Sky All-Frequency (ASAF) analysis, where Deepali Agarwal was the lead analyst, Sanjit Mitra led the search. Sukanta Bose and Shivaraj Kandhasamy were part of the LVK review team. [Phys. Rev. D 105, 122001 (2022)] G r ound - b a s e d i n t e r f e r ome t r i c detectors of GW have taken great strides in detecting nearly a hundred binary mergers, and more detections are in the offing. However, these detectors are sensitive in the high- frequency range above 10 Hz. Because of various noise sources, it is impossible ¬ ¬ Time Delay Interferometry (TDI) for LISA ¬ Effect of calibration uncertainties on 22 RESEARCH HIGHLIGHTS GRAVITATIONAL WAVE SCIENCE IUCAA has perhaps the largest research group in gravitational waves (GW) in India. Consequently the research work done here spans a vast range of areas, namely, data analysis, signal modelling, astrophysics, cosmology, detector characterization, instrumentation, etc., leading to several publications every year. The research outcome is highly regarded worldwide, often for introducing ideas that have made significant impact in the field. In this article, we highlight several research achievements that happened in the past fewyears. Hierarchical search: Matched Filtering is the primary method that is used to find GW signals buried in noisy data. The analysis has to be conducted to search for lakhs of model waveforms (templates), making the process computationally expensive. At this point, due to computational limitations, one needs to invoke major assumptions to reduce the dimensionality and size of the search parameter space. Kanchan Soni, Sanjeev Dhurandhar and Sanjit Mitra have developed a two stage (coincident) hierarchical search, which speeds up the search by more than one order of magnitude [Phys. Rev. D 105, 064005 (2022)]. The search was applied to real data and it recovered all the events published in the LIGO-Virgo- KAGRA (LVK) collaboration's first transient catalogue (GWTC-1) with approximately the same significance. They have also developed a novel t echn i que o f de t e rmi n i ng t he significance of an event accurately yet fast, without which the computational efficiency would be lost [Phys. Rev. D 109, 024046 (2024)]. Towards a Machine Learning based search: While Machine Learning based algorithms are progressively being introduced for GW data analysis for various tasks, and Sanjit Mitra and his group at IUCAA made the first successful implementations of it on real GW data, as of now, the primary Searches CompactBinaryCoalescence (CBC) ¬ ¬