Table of contents

The yearly Gravitational Wave Data Analysis Workshop (GWDAW) is now a well established meeting. As usual, the scope of the meeting covered the status of ground-based and space-based detectors, sources and population modelling, detector characterization, event searches, multi-detector analysis and new analysis methods.

The 9th GWDAW followed the trend observed in previous years: more presentations that included real data and more sophisticated analysis techniques. However, the biggest change observed was a significant increase in the number of presentations dealing with the analysis of data coming from multiple detectors. Our field needs to combine these data to best extract the astrophysical information. This yearly meeting, bringing together all the worldwide players in the field, is a chance for us to make progress in this direction, and we will certainly see more results in future GWDAW meetings.

The importance of the 9th GWDAW meeting can be assessed not only by the record attendance and number of submitted contributions, but also by looking at the work schedules of the various groups. People are now using the GWDAW dates to set deadlines for their activities in order to present their results at the next GWDAW.

This special issue of Classical and Quantum Gravity contains the most up-to-date papers on the topics covered by the meeting. Besides being a good reference that will fit proudly into the series of previous GWDAW proceedings, it provides valuable details about current work.

As organizers of this workshop, we would like to thank not only the sponsors that made this meeting possible, but also all the participants for coming, sharing their results and enjoying the event with us.

The following institutions have sponsored the 9th GWDAW:

Laboratoire d'Annecy-le-Vieux de Physique des Particules (LAPP)

European Gravitational Observatory (EGO)

Université de Savoie

Groupement de Recherche: Phénomènes cosmiques de haute énergie (GDR-PCHE)

Institut National de Physique Nucléaire et de Physique des Particules (IN2P3)

Conseil Régional Rhône Alpes

Conseil Général de la Haute Savoie

Assemblée des Pays de Savoie

The gravitational wave interferometer Virgo is presently in its commissioning phase. The status of the detector will be presented, focusing attention on the results obtained during this last year of commissioning activity, running the interferometer in the recombined configuration (a Michelson interferometer with Fabry–Perot cavities in both the arms) and finally recycling the light beam into the interferometer.

TAMA300, an interferometric gravitational-wave detector with a 300 m baseline length, has been developed and operated with sufficient sensitivity to detect gravitational-wave events within our galaxy and sufficient stability for observations. The interferometer was operated stably and continuously for intervals longer than 24 h with a strain noise level of 2 × 10⁻²¹ Hz^−1/2. We carried out nine observation runs, and have obtained over 2700 h of data so far. These data are being analysed to look for gravitational-wave signals. In this paper, we review the design of the TAMA300 interferometer and the observation results, and briefly review the data-analysis activities. In addition, recent efforts to improve the detector are also reported.

A number of gravitational wave detectors throughout the world are currently moving from the final stages of commissioning to a more continuous observational mode. Together, these detectors form a global network which will search for gravitational waves from various astrophysical sources, such as continuous wave signals from rotating neutron stars, transient signals from, for example, inspiralling compact objects and supernovae explosions, and stochastic gravitational wave signals from the early universe. GEO 600 is a long baseline laser-interferometric gravitational wave detector which employs advanced optical and suspension techniques to reach its design sensitivity. Almost all of the major installation work at GEO 600 is already completed and the detector is currently being commissioned to prepare it for extended observation periods. The commissioning process involves many activities in the areas of noise reduction, calibration, operational stability and characterization. This report highlights some of the major commissioning steps that have contributed to the increase in sensitivity of the instrument over the period from December 2003 to December 2004. In addition, recent extensions to the on-line calibration scheme used in GEO 600 are briefly discussed.

The Laser Interferometer Space Antenna (LISA) is expected to detect gravitational radiation from a large number of compact binary systems. We present a method by which these signals can be identified and have their parameters estimated. Our approach uses Bayesian inference, specifically the application of a Markov chain Monte Carlo method. The simulation study that we present here considers a large number of sinusoidal signals in noise, and our method estimates the number of periodic signals present in the data, the parameters for these signals and the noise level. The method is significantly better than classical spectral techniques at performing these tasks and does not use stopping criteria for estimating the number of signals present.

Laser Interferometer Space Antenna (LISA) is a proposed mission to detect and study gravitational radiation in the frequency range from 10⁻⁴ to 10⁻¹ Hz. In the low part of its frequency band, the LISA data will contain a stochastic signal consisting of an incoherent superposition of hundreds of millions of gravitational wave signals radiated by inspiraling white-dwarf binaries present in our own galaxy. In order to estimate the LISA response to this background, we have simulated a population of white-dwarf binaries recently synthesized by one of us. Our approach relies on an analytic expressions of the LISA Time-Delay Interferometric responses to the gravitational radiation emitted by such systems, and it allows us to implement a computationally efficient and accurate simulation of the background in the LISA data. We find the amplitude of the galactic white-dwarf binary background in the LISA data to be modulated in time with a period of 1 year, reaching a minimum equal to about twice that of the LISA noise for a period of about 2 months around the time when the Sun–LISA direction is roughly oriented towards the Autumn equinox. This modulation means that the galactic white-dwarf background that will be observable by LISA is a cyclostationary random process with a period of 1 year. We summarize the theory of cyclostationary random processes and present the corresponding generalized spectral method needed to characterize such a process in the LISA data. We find that, by measuring the generalized spectral components of the white-dwarf background, LISA will be able to infer properties of the distribution of the white-dwarf binary systems present in our galaxy.

One of the main sources of gravitational waves for the LISA space-borne interferometer is galactic binary systems. The waveforms for these sources are represented by eight parameters of which four are intrinsic and four are extrinsic to the system. Geometrically, these signals exist in an 8D parameter space. By calculating the metric tensor on this space, we calculate the number of templates needed to search for such sources. We show in this study that below a particular monochromatic frequency of f₀ ∼ 1.6 × 10⁻³ Hz we can ignore one of the intrinsic parameters and search over a 7D space. Beyond this frequency, we have a change in dimensionality of the parameter space from 7 to 8 dimensions. This sudden change in dimensionality results in a change in the scaling of template number as a function of the monochromatic frequency from ∼f^1.25₀ to ∼f^5.88₀.

In this paper we present new estimates of the coalescence rate of neutron star binaries in the local universe, and we discuss its consequences for the first generations of ground-based interferometers. Our approach based on both evolutionary and statistical methods gives a galactic merging rate of 1.7 × 10⁻⁵ yr⁻¹, in the range of previous estimates 10⁻⁶ to 10⁻⁴ yr⁻¹. The local rate which includes the contribution of elliptical galaxies is two times higher, in the order of 3.4 × 10⁻⁵ yr⁻¹. We predict one detection every 148 and 125 years with initial VIRGO and LIGO, and up to six events per year with their advanced configuration. Our recent detection rate estimates from investigations on VIRGO future improvements are quoted.

Observations of binary inspirals with the proposed Laser Interferometer Space Antenna (LISA) will allow us to place bounds on alternative theories of gravity and to study the merger history of massive black holes (MBH). These possibilities rely on LISA's parameter estimation accuracy. We update previous studies of parameter estimation for inspiralling compact binaries of MBHs, and for inspirals of neutron stars into intermediate-mass black holes, including non-precessional spin effects. We work both in Einstein's theory and in alternative theories of gravity of the scalar–tensor and massive-graviton types. Inclusion of non-precessional spin terms in MBH binaries has little effect on the angular resolution or on distance determination accuracy, but it degrades the estimation of intrinsic binary parameters such as chirp mass and reduced mass by between one and two orders of magnitude. The bound on the coupling parameter ω_BD of scalar–tensor gravity is significantly reduced by the presence of spin couplings, while the reduction in the graviton-mass bound is milder. LISA will measure the luminosity distance of MBHs to better than ∼10% out to z ≃ 4 for a (10⁶ + 10⁶)M_⊙ binary, and out to z ≃ 2 for a (10⁷ + 10⁷)M_⊙ binary. The chirp mass of a MBH binary can always be determined with excellent accuracy. Ignoring spin effects, the reduced mass can be measured within ∼1% out to z = 10 and beyond for a (10⁶ + 10⁶)M_⊙ binary, but only out to z ≃ 2 for a (10⁷ + 10⁷)M_⊙ binary. Present-day MBH coalescence rate calculations indicate that most detectable events should originate at z∼ 2–6: at these redshifts LISA can be used to measure the two black-hole masses and their luminosity distance with sufficient accuracy to probe the merger history of MBHs. If the low-frequency LISA noise can only be trusted down to 10⁻⁴ Hz, parameter estimation for MBHs (and LISA's ability to perform reliable cosmological observations) will be significantly degraded.

One of the fundamental and yet untested predictions of inflationary models is the generation of a very weak cosmic background of gravitational radiation. We investigate the sensitivity required for a space-based gravitational wave laser interferometer with peak sensitivity at ∼1 Hz to observe such signal as a function of the model parameters and compare it with indirect limits that can be set with data from present and future cosmic microwave background missions. We concentrate on signals predicted by slow-roll single-field inflationary models and instrumental configurations such as those proposed for the LISA follow-on mission: big bang observer.

We describe a method for calibrating the ALLEGRO resonant detector. The resulting response function can be used to transform the observed data backwards to gravitational strain data. These data are the input to a cross-correlation analysis to search for stochastic gravitational waves.

We could reconstruct the strain of gravitational wave signals from acquired data in the time domain by using the infinite impulse response filter technique in TAMA300. We would like to analyse the waveform in the time domain for burst-like signal, merger phase waveform of binary neutron stars, and so on. We established the way to make a continuous time-series gravitational wave strain signal. We compared the time-domain reconstruction with the Fourier-space reconstruction. Both coincided within 3% in the observation range. We could also produce the voltage signal which would be recorded by the data-acquisition system from a simulated gravitational wave. This is useful for some analyses of simulations and signal injections. We could extract the waveform of the hardware injection signal in an observational run in the time domain. The extracted waveform was similar to the injection signal.

Amplitude calibration procedures have been developed by the LIGO Scientific Collaboration (LSC) for use in determining the strain sensitivity of the three LIGO interferometers. These frequency-domain procedures rely on a fiducial calibration taken at a reference time t₀. The calibration is then propagated to all other times during the science run via calibration factors (denoted by α and β), which are derived from sinusoidal length excitations in interferometer cavity lengths. We briefly review the standard calibration methods that were employed in the first two LIGO science runs (S1 and S2), and then describe improvements in calibration procedures implemented during the third science run S3.

It is computationally expensive to search the large parameter space associated with a gravitational wave signal of uncertain frequency, such as might be expected from the possible pulsar generated by SN1987A. To address this difficulty we have developed a Markov Chain Monte Carlo method that performs a time-domain Bayesian search for a signal over a 4 Hz frequency band and a spindown of magnitude of up to 1 × 10⁻⁹ Hz s⁻¹. We use Monte Carlo simulations to set upper limits on signal amplitude with this technique, which we intend to apply to a gravitational wave search.

The space of phase parameters (sky position, frequency, spindowns) of a coherent matched-filtering search for continuous gravitational waves from isolated neutron stars shows strong global correlations ('circles in the sky'). In the local limit this can be analysed in terms of a parameter-space metric, but the global properties are less well studied. In this work, we report on our recent progress in understanding these global correlations analytically for short to intermediate (less than a month, say) observation times and neglecting spindowns. The location of these correlation circles in parameter space is found to be determined mostly by the orbital velocity of the earth, while the spin motion of the detector and the antenna patterns only contribute significantly to the amplitude of the detection statistic along these circles.

In this paper we present the hierarchical search for periodic gravitational sources that we propose for the data analysis of the Virgo and Explorer and Nautilus resonant gravitational antennae. All the equations that can be used to 'tune' the algorithm are presented.

Astrophysical models indicate that low mass x-ray binaries (LMXBs) are very promising as sources of continuous, quasi-periodic gravitational waves at high frequencies, such as those at which resonant bar detectors are operating. In this preliminary paper we quickly derive the expected amplitude of the gravitational wave signals. We show that it is likely that there may be some objects emitting in the AURIGA band. We are producing a database of possible sources and identifying the most promising ones. Our preliminary result indicates that we can achieve a good sensitivity with an observation time of about 10 days thanks to the improved sensitivity of the upgraded AURIGA detector. We also show what will be possible to perform with the advanced resonant detectors (DUAL). Finally, we briefly outline the general features of the analysis method we intend to apply postponing to a future paper the detailed discussion.

I study, via a Monte Carlo simulation, a population of neutron stars evolving through the emission of gravitational waves. A starting population, with ages uniformly distributed back to 100 Myr (or 500 Myr), is evolved in the galactic gravitational potential to the present time. Neutron stars' natal kick velocity and the Gould Belt, with an enhanced formation rate, are taken into account. Neutron stars' ellipticity is assumed to be distributed exponentially with mean value treated as a parameter. The detectability of the emitted gravitational signals, by the first and planned second generation of interferometric detectors, is estimated.

We present the noise analysis package library (NAP), a new tool to perform noise studies and data conditioning on the data produced by gravitational wave detectors. We describe its design and report the results of the application of parametric spectral estimation, whitening, line removal with adaptive notch filters and noise removal using a multicoherence procedure on the data taken by the Virgo interferometer during the C5 engineering run.

The LIGO detectors collected about four months of data in 2003–2004 during two science runs, S2 and S3. Several environmental and auxiliary channels that monitor the instruments' physical environment and overall interferometric operation were analysed in order to establish the quality of the data as well as the presence of transients of non-astrophysical origin. This analysis allowed a better understanding of the noise character of the instruments and the establishment of correlations between transients in these channels and the one recording the gravitational wave strain. In this way, vetoes for spurious burst events were identified. We present the methodology we followed in this analysis and the results from the S2 and S3 veto analysis within the context of the search for gravitational wave bursts.

LIGO recently conducted its third scientific data run, S3. Here, we summarize the veto and data quality studies conducted by the LIGO Scientific Collaboration in connection with the search for binary inspiral signals in the S3 data. The veto results presented here come from studies on the S3 playground data. LIGO's interferometer channels and physical environmental monitors were monitored, and events in these channels coincident with inspiral triggers were examined.

During the commissioning of the Virgo interferometer, a search for environmental noise contributions to the dark fringe signal was undertaken. Dedicated tests have been performed to identify major sources of disturbances and to understand the coupling mechanism with the interferometer. The major effect is due to seismic/acoustic noise coupling to the laser beam before the input mode cleaner, then propagating as beam power noise to the ITF dark fringe output signal. In this paper we illustrate the tests performed and preliminary results of our investigation.

The high sensitivity required of gravitational wave detectors is such that environmental noise becomes a very important consideration. Local environmental noise may be partially or wholly eliminated by cross-correlating data from two geographically separated detectors; such a strategy will not, however, be able to eliminate environmental noise that is global in nature. Thus, the characterization of global environmental noise is an important task. We demonstrate an analysis pipeline to correlate measurements taken by globally separated physical environment monitoring instruments located at LIGO, VIRGO and The Australian National University (ANU), and successfully detect expected features.

We describe the cross-correlation measurements being carried out on data from the LIGO Livingston Observatory and the ALLEGRO resonant bar detector. The LIGO data are sampled at 16 384 Hz while the ALLEGRO data are base-banded, i.e., heterodyned at 899 Hz and then sampled at 250 Hz. We handle these different sampling parameters by working in the Fourier domain, and demonstrate the approximate equivalence of this measurement to a hypothetical time-domain method in which both data streams are upsampled.

The INSPIRAL program is the LIGO Scientific Collaboration's computational engine for the search for gravitational waves from binary neutron stars and sub-solar mass black holes. We describe how this program, which makes use of the FINDCHIRP algorithm, is integrated into a sophisticated data analysis pipeline that was used in the search for low-mass binary inspirals in data taken during the second LIGO science run.

We present the status of the joint search for gravitational waves from inspiraling neutron star binaries in the LIGO Science Run 2 and TAMA300 Data Taking Run 8 data, which were taken from 14 February to 14 April 2003, by the LIGO and TAMA Collaborations. In this paper, we discuss what has been learned from an analysis of a subset of the data sample reserved as a 'playground'. We determine the coincidence conditions for parameters such as the coalescence time and chirp mass by injecting simulated Galactic binary neutron star signals into the data stream. We select coincidence conditions so as to maximize our efficiency of detecting simulated signals. We obtain an efficiency for our coincident search of 78% and show that we are missing primarily very distant signals for TAMA300. We perform a time-slide analysis to estimate the background due to accidental coincidence of noise triggers. We find that the background triggers have a very different character from the triggers of simulated signals.

The LIGO scientific collaboration is currently engaged in the first search for binary black hole inspiral signals in real data. We are using the data from the second LIGO science run and we focus on inspiral signals coming from binary systems with component masses between 3 and 20 solar masses. We describe the analysis methods used and report on preliminary estimates for the sensitivities of the LIGO instruments during the second science run.

Detecting gravitational ringdown waves provides a probe for direct observation of astrophysical black holes. The masses and angular momenta of black holes can be determined from the waveforms by using the black-hole perturbation theory. In this paper we present data analysis methods to search for black-hole ringdowns of fundamental quasi-normal modes with interferometric gravitational wave detectors, and report an application to the TAMA300 data. Our method is based upon matched filtering by which we calculate cross-correlations between detector outputs and reference waveforms. In a search for gravitational signals, fake reductions and event identifications are of most importance. We developed two methods to reject spurious triggers in filter outputs in the time domain and examined their reduction powers. It is shown that by using the methods presented here the number of fake triggers can be reduced by an order with a false dismissal probability of 5%. We also discuss the possibility of using the higher order quasi-normal modes for event selection.

We report on an investigation of Virgo Commissioning run 4 data, dedicated to searching signals of the kind supposed to be emitted by inspiral neutron star binary systems. Given the still relatively limited sensitivity, the goal was to test some of the elements of the analysis chain, using simulated events, hardware and software injected in the data; the test allowed us to also characterize the detector stability during the run, for the purposes of the inspiral event search.

This paper reports on a project that is the first step the LIGO Scientific Collaboration and the Virgo Collaboration have taken to prepare for a mutual search for inspiral signals. The project involved comparing the analysis pipelines of the two collaborations on data sets prepared by both sides, containing simulated noise and injected events. The ability of the pipelines to detect the injected events was checked, and a first comparison of how the parameters of the events were recovered has been completed.

CorrPower is a cross-correlation-based algorithm to be used in the LIGO burst analysis. The code looks for excesses of coherent power in multiple interferometers, unifying techniques previously implemented in the LIGO triggered and untriggered burst searches. CorrPower performs three functions: (1) a continuous scan of the data, (2) an r-statistic waveform consistency test on candidates produced by the burst event analysis (Cadonati L 2004 Class. Quantum Grav.21 S1695–703, LIGO Scientific Collaboration 2005 Preprint gr-qc/0505029), and (3) a search around the time of external triggers, a natural evolution of the analysis used for the gravitational-wave signature of GRB030329 (LIGO Scientific Collaboration 2005 Phys. Rev. D 72 042002, Mohanty S et al2004 Class. Quantum Grav.21 S1831–7, Mohanty S et al2004 Class. Quantum Grav.21 S765–74). The paper describes the techniques implemented in the continuous data scan and presents its expected performance on simulated data from two or three ideal detectors with white Gaussian noise.

Tracksearch is an algorithm to detect unmodelled gravitational wave signals in interferometric data which was first proposed almost ten years ago by Anderson and Balasubramanian. It is one of the only methods proposed which is well suited to look for unmodelled gravitational wave signals which have hundreds of cycles or more. This paper continues the work they began. In particular, we introduce a new trigger statistic for tracksearch, the integrated power, and compare it to the track length statistic used by Anderson and Balasubramanian. Our initial findings suggest that the integrated power will perform equivalently to or better than track length in almost every case. Furthermore, the integrated power statistic appears to be far less sensitive to suboptimal parameter choices, indicating that it may be more suitable for use on real gravitational wave data.

Post-Newtonian expansions of the binding energy and gravitational wave flux truncated at the same relative post-Newtonian order form the basis of the standard adiabatic approximation to the phasing of gravitational waves from inspiralling compact binaries. Viewed in terms of the dynamics of the binary, the standard approximation is equivalent to neglecting certain conservative post-Newtonian terms in the acceleration. In an earlier work, we had proposed a new complete adiabatic approximant constructed from the energy and flux functions. At the leading order, it employs the 2PN energy function rather than the 0PN one in the standard approximation, so that, effectively, the approximation corresponds to the dynamics where there are no missing post-Newtonian terms in the acceleration. In this paper, we compare the overlaps of the standard and complete adiabatic templates with the exact waveform in the adiabatic approximation of a test-mass motion in the Schwarzschild spacetime, for the VIRGO and the advanced LIGO noise spectra. It is found that the complete adiabatic approximants lead to a remarkable improvement in the effectualness at lower PN (<3PN) orders, while standard approximants of order ⩾3PN provide a good lower bound to the complete approximants for the construction of effectual templates. Faithfulness of complete approximants is better than that of standard approximants except for a few post-Newtonian orders. Standard and complete approximants beyond the adiabatic approximation are also studied using the Lagrangian templates of Buonanno, Chen and Vallisneri.

We propose a new method for the detection of spectral lines in random noise. It mimics the processing scheme of matching filtering, i.e., a whitening procedure combined with the measurement of the correlation between the data and a template. Thanks to the original noise spectrum estimate used in the whitening procedure, the algorithm can easily be tuned to various types of noise. It can thus be applied to the data taken from a wide class of sensors. This versatility and its small computational cost make this method particularly well suited for real-time monitoring in gravitational wave experiments. We show the results of its application to Virgo C4 commissioning data.

In the hierarchical search for periodic sources the construction of the short FFT database (SFDB) plays a key role for the next steps of the search, in terms of easy and fast access to the needed information and of data quality. The last information is crucial when combining data over long time periods, given the presence of non-stationarities in the noise. We will outline here the procedure we use to construct the SFDB and in particular the peak map, which is the first step of the hierarchical procedure, describing the tools we use to remove disturbances, which would enhance the noise floor. We will also describe the data and information we evaluate and store to characterize each FFT. Particular emphasis is given to the procedure used to construct the estimator of the average noise spectral density, which is needed for good detection efficiency in the identification of peaks. We will give some examples, using non-stationary data of the resonant detector Nautilus and simulated signals added to the noise.

In this study, we apply post-Newtonian (T-approximants) and resummed post-Newtonian (P-approximants) to the case of a test particle in equatorial orbit around a Kerr black hole. We compare the two approximants by measuring their effectualness (i.e., larger overlaps with the exact signal) and faithfulness (i.e., smaller biases while measuring the parameters of the signal) with the exact (numerical) waveforms. We find that in the case of prograde orbits, T-approximant templates obtain an effectualness of ∼0.99 for spins q ⩽ 0.75. For 0.75 < q < 0.95, the effectualness drops to about 0.82. The P-approximants achieve effectualness of >0.99 for all spins up to q = 0.95. The bias in the estimation of parameters is much lower in the case of P-approximants than T-approximants. We find that P-approximants are both effectual and faithful and should be more effective than T-approximants as a detection template family when q > 0. For q < 0, both T- and P-approximants perform equally well so that either of them could be used as a detection template family. However, for parameter estimation, the P-approximant templates still outperform the T-approximants.

In this paper, a hierarchical Bayesian learning scheme for autoregressive neural network models is shown which overcomes the problem of identifying the separate linear and nonlinear parts modelled by the network. We show how the identification can be carried out by defining suitable priors on the parameter space which help the learning algorithms to avoid undesired parameter configurations. Some applications to synthetic and real world experimental data are shown to validate the proposed methodology.

Combining non-parametric tests and change point detection leads to signal detectors that do not require knowledge of the noise distribution, or estimated models of the same, when fixing a threshold corresponding to a specified false alarm rate. This can lead to a simple and robust detection strategy in non-stationary and non-Gaussian noise. However, the price paid is in reduced sensitivity when dealing with ideal stationary, Gaussian noise, though the best sensitivity achievable by this class of detectors is an open issue. A non-parametric change point detector for gravitational wave burst signals has been introduced by Mohanty (2000 Phys. Rev. D 61 122002). Here, we report on modifications that lead to a significant jump in the sensitivity of this detector. This result suggests that there may still be a lot of room for improvement in the sensitivity of non-parametric change point detectors.

We have analysed three data sets, each two days long, of the EXPLORER resonant bar detector. We have searched for continuous gravitational-wave signals from spinning neutron stars. Our data analysis technique was based on the maximum likelihood detection method. We briefly describe the theoretical methods that we used in our search and we present results of the search. The main outcome of our analysis is an upper limit of 1 × 10⁻²² for the dimensionless amplitude of a continuous gravitational-wave signal. The upper limit is for any source location in the sky, any polarization of the wave and for signals of frequency from 921.00 Hz to 921.76 Hz and with spin down from −2.36 × 10⁻⁸ Hz s⁻¹ to +2.36 × 10⁻⁸ Hz s⁻¹.

Real data produced by gravitational wave detectors are affected by non-stationarities which must be properly weighted in order to reduce their effect. In the incoherent step of the hierarchical method for the periodic sources search, based on the Hough transform, two kinds of non-stationarities must be taken into account: one connected to non-stationary disturbances and another, with a period of one sidereal day, due to the rotation of the Earth, which changes the orientation of the detector and therefore the signal amplitude. In this paper, we describe the adaptive Hough transform in which these two issues are suitably treated. We discuss its statistical properties and some implementative details.

We use the Hough transform to analyse data from the second science run of the LIGO interferometers, to look for gravitational waves from isolated pulsars. We search over the whole sky and over a large range of frequencies and spindown parameters. Our search method is based on the Hough transform, which is a semi-coherent, computationally efficient, and robust pattern recognition technique. We also present a validation of the search pipeline using hardware signal injections.

We present upper limits on the amplitude of gravitational waves from 28 isolated pulsars using data from the second science run of LIGO. The results are also expressed as a constraint on the pulsars' equatorial ellipticities. We discuss a new way of presenting such ellipticity upper limits that takes account of the uncertainties of the pulsar moment of inertia. We also extend our previous method to search for known pulsars in binary systems, of which there are about 80 in the sensitive frequency range of LIGO and GEO 600.

We present the results of observations with the TAMA300 gravitational-wave detector, targeting burst signals from stellar-core collapse events. We used an excess-power filter to extract gravitational-wave candidates, and developed two methods to reduce fake events caused by non-stationary noises of the detector. These analysis methods were applied to real data from the TAMA300 interferometric gravitational wave detector. We compared the data-processed results with those of a Monte Carlo simulation with an assumed galactic-event distribution model and with burst waveforms expected from numerical simulations of stellar-core collapses, in order to interpret the event candidates from an astronomical viewpoint. We set an upper limit of 5.0 × 10³ events s⁻¹ on the burst gravitational-wave event rate in our galaxy with a confidence level of 90%.

We present a comparative study of six search methods for gravitational wave bursts using simulated LIGO and Virgo noise data. The simulated data were generated according to the design sensitivity of the two 4 km LIGO interferometers and the 3 km Virgo interferometer. The searches were applied on replicas of the data sets to which eight different signals were injected. Three figures of merit were employed in this analysis: (a) receiver operator characteristic curves, (b) necessary signal-to-noise ratios for the searches to achieve 50% and 90% efficiencies and (c) variance and bias for the estimation of the arrival time of a gravitational wave burst.

We present the current status of the TAMA300 data analysis using the alternative linear fit (ALF) filter, which is one of the burst filters used to extract burst gravitational wave (GW) signals. Although burst GW analyses of the TAMA300 data have already made progress using an excess-power filter, it is important to use various methods in order to avoid missing burst GW signals, because the precise waveforms are unknown. In the present work, we set the parameters of the ALF filter and then calculated the detection efficiency for GW burst events in our galaxy. We obtained a comparable efficiency to that of the excess-power filter. In addition, we calculated the trigger rate of the TAMA300 data. The result shows that there were so many trigger events which could not be identified as being GW signals or fake events that methods to reduce fake events are required.

Coherent detection techniques are beginning to play more and more prominent roles in searches for gravitational-wave bursts with networks of interferometric detectors. Such techniques often involve cross-correlations of data streams from different detectors and therefore rely on similarity of their signals, which occurs only when the detectors are closely aligned. A simple extension of the cross-correlation test which can be applied even to completely misaligned interferometers is presented here. In this method, searches for bursts in one of the detectors are performed with noisy templates built out of the data streams from other detectors in the network. The efficiency of this algorithm is studied with numerical simulations. We show that by properly mixing the signals from misaligned detectors one can achieve a degree of their correlation which is close to that of a perfectly aligned pair of detectors. The redundancy in the signal mixing is used to improve the detection efficiency.

A network of gravitational wave detectors is called redundant if, given the direction to a source, the strain induced by a gravitational wave in one or more of the detectors can be fully expressed in terms of the strain induced in others in the network. Because gravitational waves have only two polarizations, any network of three or more differently oriented interferometers with similar observing bands is redundant. The three-armed LISA space interferometer has three outputs that are redundant at low frequencies. The two aligned LIGO interferometers at Hanford WA are redundant, and the LIGO detector at Livingston LA is nearly redundant with either of the Hanford detectors. Redundant networks have a powerful veto against spurious noise, a linear combination of the detector outputs that contains no gravitational wave signal. For LISA, this 'null' output is known as the Sagnac mode, and its use in discriminating between detector noise and a cosmological gravitational wave background is well understood. But the usefulness of the null veto for ground-based detector networks has been ignored until now. We show that it should make it possible to discriminate in a model-independent way between real gravitational waves and accidentally coincident non-Gaussian noise 'events' in redundant networks of two or more broadband detectors. It has been shown that with three detectors, the null output can even be used to locate the direction to the source, and then two other linear combinations of detector outputs give the optimal 'coherent' reconstruction of the two polarization components of the signal. We discuss briefly the implementation of such a detection strategy in realistic networks, where signals are weak, detector calibration is a significant uncertainty, and the various detectors may have different (but overlapping) observing bands.

Over the course of a two-week period, starting on Christmas Eve 2003, the recently upgraded AURIGA and the LIGO observatory were simultaneously acquiring data. This first coincidence run between the two projects triggered a new collaborative effort in the search for gravitational wave bursts. This paper introduces the goals of the AURIGA–LIGO joint analysis and the methods that have been formulated to address the challenges of a coincidence between detectors with different spectral sensitivities, bandwidths and antenna patterns. Two approaches are presented, both based on the exchange of event triggers between AURIGA and LIGO: a set of directional coincidence searches, which exploit measured amplitude information, and a cross-correlation search in the LIGO interferometers around the time of the AURIGA events, with minimal assumptions on the signal characteristics.

Gamma-ray burst (GRB) central engines are potentially strong gravitational wave (GW) sources. GRBs occur at the rate of ∼1/day and several currently operating and planned observatories will maintain coverage of the gamma-ray window over the next decade. Data analysis strategies are required that can exploit multiple GRB triggers collected over a long span of time to expand the science reach of GW detectors beyond that possible with a single direct coincidence search. Such a strategy has been proposed earlier by Finn, Mohanty and Romano (FMR). In this paper, we formally derive multi-GRB search algorithms from principles of Frequentist statistical decision theory. The FMR approach is shown to be a special case where the duration of the GW signal from a GRB is of the same order as its expected offset from the gamma-ray counterpart. A new ad hoc statistic, based on plausible arguments, is also constructed and compared with the formally derived one.

The inspirals of stellar-mass compact objects into supermassive black holes constitute some of the most important sources for LISA. Detection of these sources using fully coherent matched filtering is computationally intractable, so alternative approaches are required. In a previous paper (Wen L and Gair J R 2005 Class. Quantum Grav.22 S445), we outlined a detection method based on looking for excess power in a time–frequency spectrogram of the LISA data. The performance of the algorithm was assessed using a single 'typical' trial waveform and approximations to the noise statistics. In this paper we present results of Monte Carlo simulations of the search noise statistics and examine its performance in detecting a wider range of trial waveforms. We show that typical extreme mass ratio inspirals can be detected at distances of up to 1–3 Gpc, depending on the source parameters. We also discuss some remaining issues with the technique and possible ways in which the algorithm can be improved.

When testing multiple hypotheses in a survey—e.g. many different source locations, template waveforms, and so on—the final result consists of a set of confidence intervals, each one at a desired confidence level. But the probability that at least one of these intervals does not cover the true value increases with the number of trials. With a sufficiently large array of confidence intervals, one can be sure that at least one is missing the true value. In particular, the probability of false claim of detection becomes non-negligible. In order to compensate for this, one should increase the confidence level, at the price of reduced detection power. False discovery rate control (Benjamini Y and Hochberg Y 1995 J. R. Stat. Soc. B 57 289–300) is a relatively new statistical procedure that bounds the number of mistakes made when performing multiple hypothesis tests. We shall review this method, discussing exercise applications to the field of gravitational wave surveys.

The detection of burst-type events in the output of ground gravitational wave observatories is particularly challenging due to the expected variety of astrophysical waveforms and the issue of discriminating them from instrumental noise. Robust methods, that achieve reasonable detection performances over a wide range of signals, would be most useful. We present a burst-detection pipeline based on a time–frequency transform, the S transform. This transform offers good time–frequency localization of energy without requiring prior knowledge of the event structure. We set a simple (and robust) event extraction chain. Results are provided for a variety of signals injected in simulated Gaussian statistics data (from the LIGO–Virgo joint working group). Indications are that detection is robust with respect to event type and that efficiency compares reasonably with reference methods. The time–frequency representation is shown to be affected by spectral features such as resonant lines. This emphasizes the role of pre-processing.