Constructing and Evaluating Template Banks for Binary Neutron Star Detection

Matched filtering is the cornerstone of gravitational wave detection from compact binary coalescences. This technique compares detector data with a large library of theoretically predicted waveforms. While the general waveform shape is derived from general relativity, key parameters—such as masses, spins, time of arrival, and phase—are unknown a priori. To ensure robust detection, matched filtering must be performed over a densely spaced template bank covering the relevant parameter space. Efficient template bank design is critical: if too sparse, signals may be missed; if too dense, computational resources are wasted. In this project, I treat waveforms as vectors in a signal space and define a metric that quantifies the distance between signals. Using this framework, I implemented and evaluated a geometric template placement algorithm, comparing its performance and computational cost to existing stochastic methods. To do so, I adapted LALSuite's C-coded routines to generate template banks for non-spinning binary neutron star systems with component masses in the range 1–3 solar masses, using modern design sensitivity power spectral density. Bank coverage was evaluated using PyCBC's bank verification tools and tested with 10,000 noiseless waveform injections. The geometric algorithm generated the bank in just 0.33 seconds and successfully recovered 97.6% of injected signals. I will present the methods, results, and comparative analysis, along with recommendations for future bank design strategies.