SEARCHING FOR GRAVITATIONAL-WAVES FROM COMPACT BINARY COALESCENCES WHILE DEALING WITH CHALLENGES OF REAL DATA AND SIMULATED WAVEFORMS
Dayanga, Waduthanthree Thilina
MetadataShow full item record
Albert Einstein's general theory of relativity predicts the existence of gravitational waves (GWs). Direct detection of GWs will provide enormous amount of new information about physics, astronomy and cosmology. Scientists around the world are currently working towards the first direct detection of GWs. The global network of ground-based GW detectors are currently preparing for their first advanced detector Science runs. In this thesis we focus on detection of GWs from compact binary coalescence (CBC) systems. Ability to accurately model CBC GW waveforms makes them the most promising source for the first direct detection of GWs. In this thesis we try to address several challenges associated with detecting CBC signals buried in ground-based GW detector data for past and future searches. Data analysis techniques we employ to detect GW signals assume detector noise is Gaussian and stationary. However, in reality, detector data is neither Gaussian nor stationary. To estimate the performance loss due to these features, we compare the efficiencies of detecting CBC signals in simulated Gaussian and real data. Additionally, we also demonstrate the effectiveness of multi-detector signal based consistency tests such ad null-stream. Despite, non-Gaussian and non-stationary features of real detector data, with effective data quality studies and signal-based vetoes we can approach the performance of Gaussian and stationary data. As we are moving towards advanced detector era, it is important to be prepared for future CBC searches. In this thesis we investigate the performances of non-spinning binary black hole (BBH) searches in simulated Gaussian using advanced detector noise curves predicted for 2015-2016. In the same study, we analyze the GW detection probabilities of latest pN-NR hybrid waveforms submitted to second version of Numerical Injection Analysis (NINJA-2) project. The main motivation for this study is to understand the ability to detect realistic BBH signals of currently available template waveforms in LIGO Algorithms Libraries (LAL) such as EOBNR waveform family. Results of the analysis demonstrates, although the detection efficiency is least affected, parameter estimation can be challenging in future searches.Many authors suggested and demonstrated coherent searches are the most sensitive in detecting GW signals using network of multiple detectors. Owing to computational expenses in recent Science data searches of LIGO and Virgo we did not employ coherent search methods. In this thesis we demonstrate how to employ coherent searches for current CBC searches in computational feasible way. As a solution, we thoroughly investigate many aspects of coherent searches using a all-sky blind hierarchical coherent pipeline. Most importantly we presents some powerful insights extracted by running coherent hierarchical pipeline on LIGO and Virgo data. This also includes the challenges we need to address before moving to all-sky all-time fully coherent searches. Estimating GW background play critical role in data analysis. We are still exploring the best way to estimate background of a CBC GW search when one or more signal present in data. In this thesis we try to address this to certain extend through NINJA-2 mock data challenge. However, due to limitations of methods and computer power, for triple coincident GW candidates we only consider loudest two interferometers for background estimation purposes.