Syracuse University SURFACE Physics - Dissertations College of Arts and Sciences 12-2011 Searching for Gravitational Waves from Compact Binary Coalescence Using LIGO and Virgo Data Collin Capano Syracuse University Follow this and additional works at: https://surface.syr.edu/phy_etd Part of the Physics Commons Recommended Citation Capano, Collin, "Searching for Gravitational Waves from Compact Binary Coalescence Using LIGO and Virgo Data" (2011). Physics - Dissertations. 114. https://surface.syr.edu/phy_etd/114 This Dissertation is brought to you for free and open access by the College of Arts and Sciences at SURFACE. It has been accepted for inclusion in Physics - Dissertations by an authorized administrator of SURFACE. For more information, please contact [email protected]. ABSTRACT This thesis describes current efforts to search for gravitational waves from com- pact binary coalescences (CBCs) by the LIGO Scientific Collaboration (LSC) and the Virgo Collaboration. We briefly review the physics of gravitational-wave emission and detection, describing how gravitational waves are emitted from \inspiraling" compact stellar mass objects and how the LSC and Virgo try to detect them using interferom- eters. Next we review the data-analysis principles used to search for potential signals in the detectors' noise. These principles are employed by \ihope," which is the data- analysis pipeline used to search for CBCs. We describe each step in this pipeline and discuss how interferometer data is stored and examined. Next we present the results from a six-month long search which occurred in early 2007, during LIGO's fifth sci- ence run. This is followed by details of tuning studies carried out on LIGO's sixth-, and Virgo's second- and third-, science runs (S6, VSR2, and VSR3), which ran from July 2009 to October 2010. No gravitational waves were detected in these searches. A \blind injection" was performed during S6/VSR3 and detected by our pipeline, however. We detail studies into assigning a statistical significance to this injection. Next we use these studies to show that we can expect to detect gravitational waves with high significance using two detectors in the advanced detector era. Finally, we review some future developments for the CBC pipeline currently being undertaken. SEARCHING FOR GRAVITATIONAL WAVES FROM COMPACT BINARY COALESCENCE USING LIGO AND VIRGO DATA By Collin D. Capano B.S. Syracuse University, 2005 M.S. Syracuse University, 2008 Dissertation Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Physics Syracuse University December 2011 Copyright c 2011 Collin D. Capano All rights reserved. Contents List of Tables x List of Figures xxv Preface xxvi Conventions xxvii Acknowledgments xxviii 1 Introduction 1 2 Physics of Gravitational Waves and the LIGO/Virgo Interferome- ters 4 2.1 Gravitational Waves in General Relativity . .4 2.1.1 Gravitational Waves from a Compact Binary Inspiral . .8 2.1.2 Evolution of the Gravitational Waveform in Newtonian Physics 15 2.1.3 Orbital Dynamics in a Schwarzschild Spacetime . 17 2.1.4 The Post-Newtonian Approximation . 22 2.2 Detection of Gravitational Waves using Interferometers . 27 2.2.1 Antenna Pattern . 28 2.2.2 The LIGO and Virgo Interferometers . 30 2.3 Astrophysical Sources of Gravitational Waves from Compact Binary Coalescence . 35 3 Obtaining Gravitational Wave Triggers from Interferometer Data 38 3.1 Detecting a Gravitational Wave Using a Matched Filter . 38 iv 3.2 Filtering with Multiple Templates . 47 3.3 The χ2 Test................................ 49 3.3.1 Applying χ2 ............................ 54 3.4 Coincidence Testing . 55 4 Ranking Triggers by False Alarm Rate 61 4.1 Poisson Process . 61 4.2 Modelling the Expected Number of Coincident Triggers . 65 4.3 Computing Background Using Time Slides . 70 4.4 Computing FARs with Multiple Templates . 76 4.5 Alternate Method to Compute Combined FARs . 82 4.6 Algorithm for Computing False Alarm Rates . 86 5 The IHOPE Pipeline 87 5.1 Pipeline Requirements . 88 5.2 ihope at Runtime . 93 5.2.1 Science and Veto Segments Retrieval . 94 5.2.2 HIPE . 95 5.2.3 Pipedown . 96 5.2.4 DAX . 98 5.2.5 The Pipeline in Detail . 98 5.3 HIPE in Detail . 98 5.3.1 Data Find . 100 5.3.2 From Continuous to Discrete Data . 100 5.3.3 Data Segmentation . 102 5.3.4 Template Bank . 104 5.3.5 Injections . 105 5.3.6 First Inspiral . 108 5.3.7 First Coincidence . 111 5.3.8 Trigbank . 112 5.3.9 Second Inspiral . 113 5.3.10 Second Coincidence . 115 5.4 Data Storage . 118 5.4.1 The sngl inspiral Table . 120 v 5.4.2 The sim inspiral and process Tables . 121 5.4.3 Coinc Tables . 121 5.4.4 Experiment Tables . 125 5.4.5 The Segment Tables . 128 5.4.6 Other Tables . 129 5.4.7 File Formats . 129 5.5 Pipedown in Detail . 131 5.5.1 ligolw thinca to coinc ..................... 131 5.5.2 ligolw sqlite .......................... 132 5.5.3 ligolw cbc dbsimplify ..................... 133 5.5.4 ligolw cbc repop coinc .................... 134 5.6 ligolw cbc cluster coincs ....................... 135 5.6.1 The Injection Branch . 137 5.6.2 Preparing the Final Database . 138 5.6.3 Computing False Alarm Rates . 139 5.6.4 IFAR Plots . 140 5.6.5 PlotCumHist and PlotSlides .................. 141 5.6.6 PlotFM ............................... 143 5.6.7 PlotROC .............................. 144 5.6.8 PrintLC and MiniFollowups .................. 146 5.6.9 PrintMissed and PrintSims .................. 149 5.7 Tying it All Together: The ihope Page . 151 6 S5 Results 179 6.1 The Data Analysis Pipeline . 180 6.2 Search results . 183 7 S6 Tuning and Results 189 7.1 Hardware Improvements . 190 7.2 S6 Epochs . 190 7.2.1 S6A . 193 7.2.2 S6B . 199 7.2.3 S6C . 202 7.2.4 S6D . 207 vi 7.3 DQ Issues . 213 7.3.1 The H1 LVEA SEISZ Veto: An Example Veto using Loudest Slides217 7.3.2 The \Spike" Glitch . 218 7.4 Results . 223 7.5 The Blind Injection . 226 7.5.1 Observation and FAR Estimation . 227 7.5.2 Conclusions . 233 8 Can We Detect with Two Detectors? A Study for LIGO South 240 8.1 Advantages of LIGO South . 240 8.2 Can We Detect at SNR 8 with Two Detectors? . 241 8.2.1 Deeper Study . 243 8.3 Conclusions . 244 9 Future Developments and Conclusions 249 9.1 Gating . 249 9.2 Single Stage Pipeline and Updates to Pipedown . 250 9.3 Conclusion . 254 Bibliography 264 vii List of Tables 1 Estimated rates of BNS, NSBH, and BBH coalescence in the universe. Rbest indicates best estimate; \low" and \high" indicate pessimistic and optimistic rates, respectively. The component masses used in the esti- mates are 1:4=1:4M for BNS, 1:4=10:0M for NSBH, and 10:0=10:0M for BBH. For details on how these numbers were derived, see [1]. 37 2 Estimated detection rates in initial and advanced LIGO. The compo- nent masses used in the estimates are 1:4=1:4M for BNS, 1:4=10:0M for NSBH, and 10:0=10:0M for BBH. For details on how these num- bers were derived, see [1]. 37 3 The various veto categories used by the CBC group. Vetoes are applied cumulatively; statistical significance of candidates and upper limits are calculated after category 1, 2, and 3 vetoes are applied. 92 4 Commonly used columns of the sngl inspiral table. Not all columns are shown. 153 5 Some of the columns of the sim inspiral table. 154 6 Relevant columns of the process table. Not shown are the comment, jobid, and domain columns as they are rarely used. 155 7 Columns of the process params table. 155 8 The columns of the coinc inspiral table and their purpose. 156 9 The columns of the coinc event map table and their purpose. 156 10 The columns of the coinc event table and their purpose. 157 11 The columns of the coinc definer table and their purpose. 157 12 The columns of the time slide table and their purpose. 157 13 The columns of the experiment table and their purpose. 158 14 The columns of the experiment summary table and their purpose. 158 viii 15 The columns of the experiment summary table and their purpose. 159 16 The columns of the segment definer table. 159 17 The columns of the segment table. Note: the nanosecond columns are not used in the segment database, and so are currently not used by ligolw segment query, ligolw segments from cats, nor any Pipedown programs. This may change in the future. 159 18 Detailed results from the BNS search. The observation time is the time used in the upper limit analysis. The cumulative luminosity is the luminosity to which the search was sensitive above the loudest event for each coincidence time. The errors in this table are listed as one- sigma logarithmic error bars (expressed as percentages) in luminosity associated with each source error. 185 19 Overview of results from BNS, BBH and NSBH searches. Dhorizon is the horizon distance averaged over the time of the search. The cumulative luminosity is the luminosity to which the search was sensitive above the loudest event for times when all three LIGO detectors were operational. The first set of upper limits are those obtained for binaries with non- spinning components. The second set of upper limits are produced using black holes with a spin uniformly distributed between zero and the maximal value of Gm2=c....................... 186 20 The analyzed time (live time) in each epoch, and the total for S6/VSR2/3.