LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project1 Aspects of the LIGO burst search analysis Alan Weinstein, Caltech For the LSC Burst ULWG LIGO Science Seminar November 12, 2002 Outline:  LSC Burst ULWG  Burst search goals  Analysis pipeline  LIGO S1 data  Event Trigger Generators  Burst simulations  Tuning thresholds  Applying vetos  Finding coincident events  Averaging over source direction  “Representative” S1 results

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project2 Burst ULWG Status  The LSC Burst Upper Limits working group has prepared a draft report on a burst search with E7 data (no final results) LIGO-T Z  Intense effort in progress to generate a report on burst search with S1 data, with real results, by Nov 1 ;)  This report will be available soon; and we can hope to work towards publication by S2.  SO, no real results can be presented here & now.  AND, this isn’t even an “official” status report; just my impressions of where things stand, with emphasis on aspects on the analysis that I know the most about.  Expect more, and more interesting, burst search reports soon!

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project3 Burst upper limits working group  At least 35 members from LIGO and LSC  Chairs: Sam Finn and Peter Saulson  Web page:  Active, archived list:  Elog of E7 & related activities:  Weekly telecons, sub-group meetings  Subgroups and principles: »LDAS-based burst filters (“DSO’s”) – Erik Katsavounidis, Ed Daw, Julien Sylvestre »Data conditioning – Sam Finn »IFO diagnostics – David Shoemaker »Trigger and Vetos – John Zweizig, Stefan Ballmer »Simulations, efficiency evaluation – Alan Weinstein, Laura Cadonati »Coincident Event analysis tools – Daniel Sigg, Laura Cadonati »External Triggers – Szabi Marka, Rauha Rahkola

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project4 Burst Group membership Rana Adhikari, Warren Anderson, Stefan Ballmer, Barry Barish, Biplab Bhawal, Jim Brau, Kent Blackburn, Laura Cadonati, Joan Centrella, Ed Daw, Ron Drever, Sam Finn, Ray Frey, Ken Ganezer, Joe Giaime, Gabriela Gonzalez, Bill Hamilton, Ik Siong Heng, Masahiro Ito, Warren Johnson, Erik Katsavounidis, Sergei Klimenko, Albert Lazzarini, Isabel Leonor, Szabi Marka, Soumya Mohanty, Benoit Mours, Soma Mukherjee, David Ottoway, Fred Raab, Rauha Rahkola, Peter Saulson, Robert Schofield, Peter Shawhan, David Shoemaker, Daniel Sigg, Amber Stuver, Tiffany Summerscales, Patrick Sutton, Julien Sylvestre, Alan Weinstein, Mike Zucker, John Zweizig Almost certainly incomplete – apologies!

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project5 Burst search goals

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project6 Burst search goals  Search for short-duration bursts with unknown waveforms »Short duration: < 1 second; more typically, < 0.2 seconds. »Matched filtering techniques are appropriate for waveforms for which a model exists. Explicitly exclude, here! »Although the waveforms are a-priori unknown, we must require them to be in the LIGO S1 sensitivity band (~ Hz)  Search for gravitational wave bursts of unknown origin »Bound on the rate of detected gravitational wave bursts, viewed as originating from fixed strength sources on a fixed distance sphere centered about Earth, expressed as a region in a rate v. strength diagram.  Search for gravitational wave bursts associated with gamma-ray bursts »The result of this search is a bound on the strength of gravitational waves associated with gamma-ray bursts. »Work in progress by Marka, Rahkola, etc – not reported on today!

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project7 Search goals (2)  Rely on multi-detector coincidences (in time and in burst properties) to eliminate most or all fake bursts. »Use S1 triple-coincidence data (H2, H1, L1) »GEO data being analyzed in parallel, and final results will include them. Not yet! »Have not yet decided how to make use of double-coincidences, or triple coincidences with 4 detectors…  Given the S1 sensitivity, we do not expect to detect GWs »Assume that any coincident bursts are fake, estimate background through time- lag analysis, set upper limits only »Plan to work much harder on demanding consistency amongst observed coincident bursts before claiming detection »Analysis pipeline designed to work for both upper limits and detection, but have not worked out detailed (blind) criteria for detection

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project8 GEO - goals  Demonstrate ability to run GEO and LIGO data through equivalent data processing pipelines, up to and through the construction of temporal coincidences.  GEO data can be passed through an analysis pipeline on both sides of the Atlantic.  The incorporation of GEO data into the LIGO analysis (quadruple coincidence) in the “Nov. 1” upper limit is viewed to be unrealistic; however, demonstrating the mechanics of the analysis with an hour of GEO may be feasible.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project9 Analysis Pipeline

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project10 Analysis Pipeline Event Tool For E7 and S1, “GW” channel is *:LSC-AS_Q This channel is passed to LDAS, conditioned & whitened, then passed to one of several “Event Trigger Generators” (ETG’s, aka DSOs) Event Triggers (burst candidates) are stored in the LDAS metaDataBase (sngl_burst table) with start time, duration, SNR, frequency band, amplitude/power, … A subset of IFO diagnostic and PEM channels are passed to the DMT, where one or more monitors generate “Veto Triggers” Both GW event and veto trigger generation require careful optimization of thresholds Veto: Event triggers that overlap in time with veto triggers are excluded – loss of livetime. Event triggers and veto triggers are each characterized by a start time and a duration. Resulting list of “IFO Triggers” are the final product of a single IFO analysis.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project11 Pipeline, continued Bring together IFO triggers from multiple interferometers, require temporal coincidence: coincident in time if their start times coincide to better than the greater of the light travel time between the IFO sites and the start time resolution. Other parameters that characterize the events (e.g., characteristic amplitude, frequency or bandwidth) may also be required to be consistent. The result is a list of coincident events (divide by livetime to get coincident event rate.) Store in LDAS metaDataBase (multi_burst table) Accidental coincidence (background or fake) rate is evaluated via time-delayed coincidence (“lag plots”). Coincident event rate and background rate estimate are combined (Feldman-Cousins) to get excess event rate (or rate upper limit). This is then interpreted in terms of the efficiency for detecting astrophysical signals. Event Tool

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project12 Analysis Pipeline - comments Use of LDAS for GW channel and DMT for “veto” channels is arbitrary This choice is largely historical Both tools have their strengths and weaknesses, and their loyal user base I think it would have been cleaner to have one tool for both applications Veto channels, and their efficacy, change from run to run. Vetos much less efficacious for S1 than for E7. This is a good thing! Much overlap with Inspiral UL, DetChar working groups. For current analysis, coincidence processing done with EventTool in ROOT For current analysis, only the simplest coincidence criteria (time, crude frequency overlap) are applied. Much more work is needed! Event Tool

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project13 Post-LDAS/DMT event processing Result of LDAS job on GW channel (with or without simulated injection) is an event trigger in metaDatabase table sngl_burst, with start time, duration, frequency band, SNR, amplitude… Result of DMT job on veto channel is a veto trigger in metaDatabase table gds_trigger, with similar information. Manipulate these triggers, apply vetos, find coincidences, etc, using Event Class (Sigg, Ito) in ROOT Plots from L Cadonati use this tool Can alternatively use matlab, PAW, or your favorite tool.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project14 LIGO S1 data

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project15 LIGO S1 data  August 23 – September 9, 2002: 408 hrs (17 days).  L1: duty cycle 41.7% ; Total Locked time: 170 hrs H1: duty cycle 57.6% ; Total Locked time: 235 hrs H2: duty cycle 73.1% ; Total Locked time: 298 hrs L1 H1 H2  Double coincidence: L1 and H1 duty cycle 28.4%; Total coincident time: 116 hrs Double coincidence: L1 and H2 duty cycle 32.1%; Total coincident time: 131 hrs Double Coincidence: H1 and H2 duty cycle 46.1%; Total coincident time: 188 hrsL1H1L1H2H1H2  Triple Coincidence: L1, H1, and H2 duty cycle 23.4% ;Total coincident time: 95.7 hrsL1H1H2 Triple coincidence

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project16 Playground data  Gaby Gonzalez and Peter Saulson chose a representative sample of 13 locked segments, from the triple coincidence segments. They add up to 9.3 hours.  All tuning of ETG and veto trigger thresholds done on playground data only.  Will we include these 9.3 hours in the full analysis and results?

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project17 Non-stationarity, and Epoch Veto  BLRMS noise in GW channel is not stationary.  Detector response to GW (calibration) is not stationary.  Bursty-ness of GW channel is not stationary.  Fortunately, these appear to vary much less in S1 than in E7, thanks to efforts of detector and DetChar groups.  Much of this is driven by gradual misalignment during long locked stretches.  Under much study!  Shall we veto some epochs of the data based on excess RMS noise? Burst rate? Degradation of response?

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project18 BLRMS - Epoch Veto ?  BLRMS noise is far from stationary.  Playground data (pink vertical lines) are not very representative.  Veto certain epochs based on excessive BLRMS noise in some bands?

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project19 Veto on BLRMS? Histograms: 6-min segment BLRMS. vertical lines at 1  and 3  Veto segments beyond, eg, 3  ? (L. Cadonati)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project20 Calibration of detector response X ext can be a GW or a calibration line; Can be seen in AS_Q Or in DARM_CTRL. C(f) is not stationary!

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project21 Monitoring calibration lines Veto epochs with low  ?

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project22 Event trigger generators

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project23 Event Trigger Generators  Three LDAS filters (ETG’s or DSOs) are now being used to recognize candidate signals: »POWER - Excess power in tiles in the time-frequency plane –Flanagan, Anderson, Brady, Katsavounidis »TFCLUSTERS - Search for clusters of pixels in the time-frequency plane. –Sylvestre »SLOPE - Time-domain templates for large slope or other simple features –Daw, Yakushin  We will continue to carry all three in the hopes that the different approaches will have different strengths for different waveform morphologies, so that an “OR” will give us higher efficiency for GWs for a given fake rate. But, the jury is still out.  All three of these ETGs ran online during the 2 nd half of S1  NEW filter introduced last week: WaveBurst (Klimenko, Yakushin) »Performs t-f analysis in wavelet domain, working with pairs of IFOs.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project24 tfclusters Compute t-f spectrogram, in 1/8-second bins Threshold on power in a pixel, get uniform black-pixel probability Simple pattern recognition of clusters in B/W plane; threshold on size, or on size and distance for pairs of clusters

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project25 Data flow in LDAS User pipeline request  frameAPI  datacondAPI  mpiAPI  wrapperAPI  LAL code  eventmonAPI  metadataAPI  metaDB

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project26 Data conditioning in datacond All of our burst filters are expected to work best with (at least, approximately) whitened data. This is not matched filtering: don’t need to know detector response function to find excess power. In datacondAPI, we (approximately) whiten and HP (at 150 Hz) the data with pre-designed linear filters. These filters were designed for E7, and have not yet been optimized for S1; but I don’t think this is critical. No attempt (yet) at line removal: but we believe that this will eventually be very necessary.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project27 Burst simulations

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project28 Burst Simulations - GOALS  Test burst search analysis chain from: »IFO (ETM motion in response to GW burst)  »GW channel (AS_Q) data stream into LDAS  »search algorithms in LDAS  »burst triggers in database  »post-trigger analysis (optimizing thresholds and vetoes, clustering of multiple triggers, forming coincidences)  Evaluate pipeline detection efficiency for different waveforms, amplitudes, source directions, and different search algorithms  In burst search, simulated signal is injected at an early stage in LDAS (in datacondAPI)  Compare simulated signals injected into IFO with signals injected into data stream: make sure we understand IFO response

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project29 Generic statements about the sensitivity of our searches to poorly-modeled sources can straightforwardly be made from the t-f “morphology”… longish-duration, small bandwidth (ringdowns, Sine-gaussians) longish-duration, large bandwidth (chirps, Gaussians) short duration, large bandwidth (merger) In-between (Zwerger-Muller or Dimmelmeier SN waveforms) These SN waveforms are distance-calibrated; all others are parameterized by a peak or rms strain amplitude ZM SN burst chirp merger ringdown ZM SN bursts Bandwidth vs duration burst waveforms: t-f character

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project30 Ad-hoc signals: (Sine)-Gaussians These have no astrophysical significance; But are well-defined in terms of waveform, duration, bandwidth, amplitude SG 554, Q = 9

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project31 Damped sinusoid in 10 seconds of data from H2:LSC-AS_Q from E7 playground A series of damped sinusoids can be used as a “swept sine” calibration of burst search efficiency damped sinusoid (“ringdown”) waveform

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project32 Zwerger-Müller SN waveforms  Rationale: » »These are astrophysically-motivated waveforms, computed from detailed simulations of axi-symmetric SN core collapses. »There are only 78 waveforms computed, in different classes: –“regular” infall, bounce, ringdown –Multiple bounces, on centrifugal barrier of rapidly-rotating core –“rapid collapse” – rapid change in adiabatic index »Work is in progress to get many more, including relativistic effects, etc. »These waveforms are a “menagerie”, revealing only crude systematic regularities. They are wholly inappropriate for matched filtering or other model-dependent approaches. »Their main utility is to provide a set of signals that one could use to compare the efficacy of different filtering techniques.  Parameters: »Distance D »Signals have an absolute normalization »Almost all waveforms have duration < 0.2 sec

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project33 Zwerger Muller SN waveforms The 78 ZM waveforms differ in initial angular momentum (A parameter), governing degree of differential rotation initial rotational energy ( B, or  = E rot /E pot ). A,B roughly govern how many bounce peaks Within each (AB) set, the adiabatic index  r is varied from to 1.28 (stiff to soft), and the peak amplitude of the wave depends strongly on this parameter, as seen in the right.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project34 Relativistic core collapse waveforms Dimmelmeier, H., Font, J. A., and Müller, E., "Gravitational waves from relativistic rotational core collapse", Astrophys. J. Lett., 560, L163-L166, (2001), ph/ ph/

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project35 ZM supernova ringdown Hermite-gaussian chirp Menagerie of burst waveforms buried in E2 noise, including calibration/TF

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project36 What we need to know about the IFOs  Transfer function for injection from GDS into ETMx/y » (counts/nm * pendulum TF)  Response function from ETMx To LSC-AS_Q Both of these are available from calibrations; but their variation in time has not yet been included in the analysis For tfclusters & power, need IFO noise spectrum. Currently, this is estimated from the data read in to the LDAS job. This can, and does, bias the result. It’s not a big bias, for small signals; but a better way should be developed…

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project37 Tuning thresholds and evaluating efficiency

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project38 Tuning of thresholds  Trigger thresholds for GW ETG’s and vetos were tuned using playground data only  For ETG’s, simulations used to obtain an efficiency  Figure of merit to minimize: background rate / efficiency »Could also use background rate / efficiency 3  POWER and TFCLUSTERS use adaptive thresholds: using power distribution over 6-minute intervals as a baseline, threshold on excess power with fixed probability, assuming Gaussian statistics (POWER) or a Rice distribution (TFCLUSTERS).  SLOPE uses an absolute threshold; trigger rate varies dramatically when noise power varies.  Can choose conservative (low) thresholds when running data through LDAS, limited only by ability of LDAS to handle large trigger rate; cut tighter in EventTool post-processing.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project39 Search code triggers vs time for Z-M waveform injected at 75 seconds SN at 0.1 pc (ouch!) 0.2 pc 1.0 pc slope tfclusters slope tfclusters Time  Trigger “power”  threshold * With signal; o without signal injected. NO VETOES APPLIED. Vetoes get rid of most of these triggers!

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project40 Efficiency for injected signals In this example, tfclusters is run on S1 playground data, with many injections of SG-554 Hz bursts for each amplitude. Efficiency (at fixed threshold) is obtained by averaging over many injections at different times in playground. Deadtime due to vetos not counted in efficiency. Can also evaluate triple coincidence efficiency, assuming optimal response of all 3 detectors (unrealistic) – black curve. Note that power (on which we threshold) tracks input peak strain amplitude well (true for all 3 ETG’s).

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project41 Tuning ETG thresholds Left: efficiency for triple-coincident-detection of SG burst with different peak strains, versus background event rate, as threshold is varied Right: upper limit figure of merit: UL = (2.44+sqrt(b)) / livetime / efficiency Threshold chosen near UL minimum (15 in this case). This is for tfclusters; slope is similar in character. Repeat for a variety of different simulated waveforms; choose a threshold that works ok for all of them. L. Cadonati

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project42 Sine-Gaussians - efficiencies tfclusters slope L. Cadonati

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project43 1 ms Gaussians tfclusters slope L. Cadonati

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project44 Comparison between E7 and S1 L. Cadonati

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project45 Applying vetos

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project46 Veto Channels  LSC-AS_I  LSC-REFL_Q  LSC-REFL_I  LSC-POB_Q  LSC-POB_I  LSC-MICH_CTRL  LSC-PRC_CTRL  LSC-MC_L  LSC-AS_DC  LSC-REFL_DC  IOO-MC_F  IOO-MC_L  PSL-FSS_RCTRANSPD_F  PSL-PMC_TRANSPD_F Run DMT monitors glitchMon and absGlitch on many different channels Focus on IFO channels PEM channels not observed to be useful Look for channels and thresholds that: are correlated in time with GW channel glitches significantly reduce single-IFO background burst rate, while producing minimal deadtime Event Tool

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project47 Vetoes from absGlitch absGlitch first filters the time series. (Here, 30 Hz HP.) Finds times when signal crosses fixed threshold. Calculates size and duration, recorded to database.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project48 Efficacy of vetoes in E7 – L1 PSL glitch is in time with AS_Q; really cleans up L1. broad tail of events is cleaned up by L1 veto. L1 had lots of PSL glitching, so bulk of histogram is affected. S. Ballmer

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project49 Efficacy of vetoes in E7 – H2 MICH glitch some use at H2. broad tail of events is cleaned up by L1 veto. H2 was much cleaner to start with, so only tail is removed. S. Ballmer

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project50 S1 - H1 veto efficiency / deadtime H1:LSC-POB_I H1:LSC-PRC_CTRL H1:LSC-REFL_I H1:LSC-REFL_Q Effect of REFL_I veto improves for larger power tfcluster triggers! S. Ballmer

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project51 S1-H2 veto efficiency / deadtime H2:LSC-POB_I H2:LSC-PRC_CTRL H2:LSC-REFL_I H2:LSC-REFL_Q None much better than random… S. Ballmer

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project52 Veto for S-H1 Veto: H1:LSC-REFL_I filtered at 30 Hz HP. Threshold=11.6 Tuned to remove large glitches every ~ 20 minutes, very evident on REFL_I. Except for these large glitches, the veto is random- behaved. This is visible in the power histogram for TFCLUSTER below.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project53 Vetos for S1 - L1, H2 Veto: L1:LSC-AS_DC filtered at 30 Hz HP. Threshold=21 H2:LSC-PRC_CTRL filtered at 30 Hz HP. Threshold= This veto is barely time-correlated with GW channel; not used.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project54 Cuts on coincident events

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project55 Cuts on coincident events  Require temporal coincidence  tfclusters ETG (currently) finds clusters in t-f plane with 1/8 second time bins.  Can’t establish coincidences to better than that granularity.  Currently, coincidence window is 0.5 seconds.  slope ETG has no such limitation; currently, coincidence window is 0.05 seconds.  More work required to tighten this to a fraction of  10 msec light travel time between LHO/LLO

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project56 Frequency test of temporal coincidences In addition to temporal coincidence of events, we require that tfclusters give a central frequency at the two IFOs that are within 500 Hz of each other. Frequency information not yet available in slope.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project57 Coherence test for coincident bursts Distribution of coherence ( Hz) For H2-L1 coincidences Distribution, in presence of simulated faint bursts With 10 msec time delay With no time delay Simulated faint bursts are coherent Between H2-L1 (E7), in ( Hz) band Cross-correlation (coherence) statistic (CCS).

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project58 CCS Efficiency vs fake rate reduction Vary threshold on CCS. For each threshold,evaluate efficiency vs fake rate reduction. Very effective, even for delayed coincidences. Next step: use CCS to derive estimate of time delay. Note: time delay in, eg, tfclusters, can not be measured to better than 1/8 sec. CCS not yet implemented in burst search. No time delay 10 msec time delay between H2,L1

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project59 Averaging over the sky

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project60 Averaging over source direction and polarization  Generate single-IFO efficiency curve vs signal amplitude, assuming optimal direction / polarization. »Different for each data epoch »Different for each IFO »Different for each waveform. »Different for each ETG / threshold  Assuming source population is isotropic, determine single-IFO efficiency versus amplitude, averaged over source direction and polarization, using single-ifo response function.  This is easily accomplished with simple Monte Carlo; no need to go back to detailed LDAS simulations.  But, this is wrong, if both polarizations are present, with different waveforms.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project61 Coincidence efficiency, averaged over source direction & polarization  Efficiency for coincidences is the product of single-IFO efficiencies, evaluated with the appropriate response to GWs of a given source direction / polarization.  Once again, this can be easily accomplished with simple Monte Carlo, using knowledge of detector position and orientation on Earth. No need to go back to detailed LDAS simulations.  Must estimate any additional loss of efficiency due to post-coincidence event processing.  Again, a different plot for each data epoch, waveform, ETG/threshold, set of coincident detectors.  Finally, obtain event rate upper limit, divide by efficiency curve to get rate vs strain amplitude.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project62 “Representative” S1 results

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project63 Results from S1 playground (preliminary!) triple coincidences TFCLU STER TFCLUSTE R w/ veto SLOPE SLOPE w/ veto Live Time (seconds) Triple coincidences (no clustering) clustered frequency match cut0000 background / / / / % upper limit (Feldman Cousins) on events in the playground % upper limit (Feldman Cousins) on the rate (mHz) L. Cadonati Best we can hope for: 2.44/29128 = 0.08 mHz. Embargoed! TO: LIGO Scientific Collaboration FROM: R. Weiss November 12, 2002 CONCERNING: Release of scientific data from collaboration runs … In keeping with our responsibility, there should be no release of scientific data and astrophysical results until the Collaboration has agreed on the validity of the methods and the results…

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project64 S1 triple- coincidences Rate upper limits. Best we can do is observe 0 triple-coincident events: ~> 2.44/ sec ~ 0.01 mHz Embargoed! TO: LIGO Scientific Collaboration FROM: R. Weiss November 12, 2002 CONCERNING: Release of scientific data from collaboration runs … In keeping with our responsibility, there should be no release of scientific data and astrophysical results until the Collaboration has agreed on the validity of the methods and the results…

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project65 Rate upper limit vs strength from LIGO S1 triple-coincidence MUCH more work to do!  Complete “Nov 1” report  Continue optimization of vetos, epoch veto  Continue optimization of ETGs, thresholds  Optimize post-coincidence consistency cuts  Incorporate time-varying calibration  Consider more waveforms  Include GEO!  Consider 2/3, 3/4, … coincidences.  … Representative example excluded

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project66 Search goals (3)  Form of the result: For fixed analysis pipeline, thresholds, etc, can state: »We observe less than X events (at 90% CL) in Y hours of coincident observation. »These limits apply to waveform Z1 with peak amplitude greater than hZ1 (in strain units), and waveform Z2 with peak amplitude greater than hZ1 (in strain units), … »And, to ZM supernova waveforms, at distance of rZM Mpc. »In more detail: rate upper limit versus peak amplitude for waveform Z1, Z2, …  In order to evaluate our detection efficiency, we must have some model of a waveform. »Choose families of waveforms which are either generic (eg, Gaussians) or astrophysically-motivated (Zwerger-Muller supernova waveforms) »Assume isotropic source direction and polarization

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project67 Burst search interpretations Untriggered search: 1. “Instrumental” interpretation Search for coincident transients in our ifos, with no prejudice about the form of the signals or the nature of their sources. Calibrate against fixed-strength waveforms arriving at ifos. 2. Astrophysically-motivated interpretation Look for transients with features suggested by our (limited) understanding of supernovae, black holes, etc. Calibrate against fixed-luminosity waveforms distributed in space. Triggered search: 3. Coincidences with GRB triggers Analyzed by technique of Finn, Mohanty, and Romano. Are the outputs of our ifos different just before GRBs? Test via ifo-ifo cross-correlation.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project68 Plan of attack for Burst search in E7 and beyond  The group is preparing: »LDAS-based burst filters for generating GW triggers in database – see next bullet »DMT-based database triggers for environmental and/or IFO-based vetos - Zweizig »ROOT-based database-trigger (“event”) analysis to find coincidences, apply quality cuts, apply vetos, evaluate fake rates and efficiencies – Sigg, Ito  Three DSOs under active development: »Excess POWER filter – Katsavounidis/Brady »TFCLUSTERS – Sylvestre »SLOPE filter – Daw  E7 activities: »All three DSOs running during E7, on LDAS at LHO and LLO »POWER and TFCLUSTERS ran continuously throughout E7 »Scientist studies of the origin of bursts in AS_Q - Saulson »Accumulation of GRB triggers into database - Marka  First studies of E7 Burst and DMT trigger rates – Katsavounidis, 1/12/02  Currently using E7 playground data, and calibrated signal injection (at DatacondAPI) to evaluate fake rates and efficiencies, begin tuning of DSO algorithms, thresholds, cuts, parameters  Deep Mine search for effective vetos – Zweizig, Saulson, Shoemaker

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project69 Activities for next 3 months  Finalize and tune parameters for burst search filters »All tuning done with “playground” data, as compiled by Gonzalez/Chin/Shawhan/Brady, and simulations, only  Finalize use of ALLEGRO and GRB data  Identify, finalize and tune DMT vetos  Use simulations to evaluate efficiency for various burst morphologies  Exercise Coincident Event Tool to find coincidences, establish fake rates, apply vetos, evaluate efficiency and livetime  Run through full E7 data sample  Develop and commission burst search status displays for feedback to operators in control room  Narrow-band search: Establish rates/limits on bursts vs h peak or h rms, f central,  f  Astrophysics-based search: Establish rates/limits on bursts vs h peak, waveform  Search for external trigger coincidences (GEO, ALLEGRO, GRB)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project70 Other activities  Very fruitful “Scientific Interpretation Summit“ at U. Maryland (hosted by Joan Centrella), Dec ; many astrophysicists (GSFC, etc) in attendance  Paper drafts in preparation, and secretaries: »Search for narrow-bandwidth waveforms - Saulson »Search for astrophysics-inspired waveforms - Finn »Coincidences with external (GRB) triggers - Kalogera

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project71 Burst pipeline  Triggers generated by LDAS filters, written to DB.  Vetoes generated by DMT monitors looking at PEM channels and at internal ifo diagnostic signals, written to DB.  Event Tool reads DB to define candidate events: »Ignore triggers at times that are vetoed »Merge multiple adjacent triggers into well-defined events »Analyze events from all ifos to determine which are coincident »Draw histograms, analyze statistics of coincidences.  Calibration of efficiency by injection of simulated signals into real (playground) data.  Calibration of false coincidence rates by searching time-shifted data (“lag plots”).

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project72 Z-M waveforms (un-normalized)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project73 Z-M waveforms (un-normalized)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project74 Z-M waveforms (un-normalized)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project75 Z-M menagerie

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project76 Ringdowns  Rationale: »Just a way to represent a burst with limited duration, abrupt rise and gradual fall, with some wiggles. »Very well-defined peaked PSD  Parameters: »Peak h »Decay time  »Ring frequency f ring

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project77 Ringdown PSD

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project78 Hermite-Gaussians  Rationale: »Just a way to represent a burst with limited duration, gradual rise and fall, with some wiggles. »Can also do sine-gaussians, etc »Many beats in the PSD  Parameters: »Peak h »Gaussian width in time,  »Hermite order (number of wiggles)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project79 Hermite-Gauusian (6 th order) PSD

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project80 Chirps  Rationale: »In case the inspiral filters are not operational for some reason… »Just a way to represent a burst with limited duration, gradual rise and abrupt fall, with wiggles. »Well-defined power-law PSD »Simplest Newtonian form; not critical to get phase evolution right since we’re not doing matched filtering  Parameters: »Peak h (or distance D) »Duration  t »f(-  t), or chirp mass M

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project81 Chirp PSD

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project82 ZM waveforms buried in white noise

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project83 H-G, chirps, and ringdowns buried in white noise

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project84 Supernova waveforms  Numerical simulation of super novae  axially symmetric (Newtonian) hydrodynamical simulations of collapsing rotating stellar cores  78 gravitational waveforms »Various waveforms »Don’t cover all conditions »Not suitable for templates »Common characteristics : Short bursts T. Zwerger, E. Müller, Astronomy & Astrophysics, 320 (1997), 209.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project85 ZM waveform duration vs bandwidth

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project86 Response function

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project87 Actuator APlant PSensor S Feedback F=OGI noise Lh(t) PDpreampmixerLP filterAI filterADC ~ DAC AA Filter Coil driver coil Mirror pendulum Input matrix LSC matrix Output matrix VI DAC counts to mirror ADC sensors x mirrors x DOF FVVIx mirrors E  VVADC IFO GDS computer ADC DOF GIO We measure: We want: We need (?) Local damping Calibration

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project88 Burst rate upper limits vs. veto threshold Explore the upper limit on TFCLUSTERS coincident event rate, as a function of veto thresholds. (L1 PSL glitch threshold is important; H2 MICH threshold is less so.)

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project89 Julien Sylvestre’s Ph.D. thesis results Julien has carried to completion a full pipeline analysis of E7 data. DSO: TFCLUSTERS veto generation: custom code (“GIDE”), applied to PSL at L1, MICH_CTRL at H2 Interpreted using specific astrophysical models for calibrated waveforms. Set upper limits on rate density for models of neutron star bar mode instabilities, core collapses, and black hole binary mergers.

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project90 Monte Carlo of detector “events”  Can generate, in ROOT, “events” from multiple IFOS, like: »Locked IFO segments (segment), from ad hoc PDFs »Noise events from sngl_burst triggers, random times at specified rates, ucorrelated between IFO’s, random h_amp from ad hoc pdf »GW signal event sngl_burst triggers, correlated between IFO’s with “proper” time delay »“Veto” events, random times at specified rates, ucorrelated between IFO’s, random durations from ad hoc pdf (what DB table?)  Search for coincidences, fill multi-burst triggers

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project91 MetaDB tables currently defined

LIGO- G02XXXX-00-Z AJW, Caltech, LIGO Project92 Sngl_burst DB table schema