Abstract
The article reviews the statistical theory of signal detection in application to analysis of deterministic gravitationalwave signals in the noise of a detector. Statistical foundations for the theory of signal detection and parameter estimation are presented. Several tools needed for both theoretical evaluation of the optimal data analysis methods and for their practical implementation are introduced. They include optimal signaltonoise ratio, Fisher matrix, false alarm and detection probabilities, \({\mathcal F}\)statistic, template placement, and fitting factor. These tools apply to the case of signals buried in a stationary and Gaussian noise. Algorithms to efficiently implement the optimal data analysis techniques are discussed. Formulas are given for a general gravitationalwave signal that includes as special cases most of the deterministic signals of interest.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this review we consider the problem of detection of deterministic gravitationalwave signals in the noise of a detector and the question of estimation of their parameters. The examples of deterministic signals are gravitational waves from rotating neutron stars, coalescing compact binaries, and supernova explosions. The case of detection of stochastic gravitationalwave signals in the noise of a detector is reviewed in [5]. A very powerful method to detect a signal in noise that is optimal by several criteria consists of correlating the data with the template that is matched to the expected signal. This matchedfiltering technique is a special case of the maximum likelihood detection method. In this review we describe the theoretical foundation of the method and we show how it can be applied to the case of a very general deterministic gravitationalwave signal buried in a stationary and Gaussian noise.
Early gravitationalwave data analysis was concerned with the detection of bursts originating from supernova explosions [99]. It involved analysis of the coincidences among the detectors [52]. With the growing interest in laser interferometric gravitationalwave detectors that are broadband it was realized that sources other than supernovae can also be detectable [92] and that they can provide a wealth of astrophysical information [85, 59]. For example the analytic form of the gravitationalwave signal from a binary system is known in terms of a few parameters to a good approximation. Consequently one can detect such a signal by correlating the data with the waveform of the signal and maximizing the correlation with respect to the parameters of the waveform. Using this method one can pick up a weak signal from the noise by building a large signaltonoise ratio over a wide bandwidth of the detector [92]. This observation has led to a rapid development of the theory of gravitationalwave data analysis. It became clear that the detectability of sources is determined by optimal signaltonoise ratio, Equation (24), which is the power spectrum of the signal divided by the power spectrum of the noise integrated over the bandwidth of the detector.
An important landmark was a workshop entitled Gravitational Wave Data Analysis held in Dyffryn House and Gardens, St. Nicholas near Cardiff, in July 1987 [86]. The meeting acquainted physicists interested in analyzing gravitationalwave data with the basics of the statistical theory of signal detection and its application to detection of gravitationalwave sources. As a result of subsequent studies the Fisher information matrix was introduced to the theory of the analysis of gravitationalwave data [40, 58]. The diagonal elements of the Fisher matrix give lower bounds on the variances of the estimators of the parameters of the signal and can be used to assess the quality of astrophysical information that can be obtained from detections of gravitationalwave signals [32, 57, 18]. It was also realized that application of matchedfiltering to some sources, notably to continuous sources originating from neutron stars, will require extraordinary large computing resources. This gave a further stimulus to the development of optimal and efficient algorithms and data analysis methods [87].
A very important development was the work by Cutler et al. [31] where it was realized that for the case of coalescing binaries matched filtering was sensitive to very small postNewtonian effects of the waveform. Thus these effects can be detected. This leads to a much better verification of Einstein’s theory of relativity and provides a wealth of astrophysical information that would make a laser interferometric gravitationalwave detector a true astronomical observatory complementary to those utilizing the electromagnetic spectrum. As further developments of the theory methods were introduced to calculate the quality of suboptimal filters [9], to calculate the number of templates to do a search using matchedfiltering [74], to determine the accuracy of templates required [24], and to calculate the false alarm probability and thresholds [50]. An important point is the reduction of the number of parameters that one needs to search for in order to detect a signal. Namely estimators of a certain type of parameters, called extrinsic parameters, can be found in a closed analytic form and consequently eliminated from the search. Thus a computationally intensive search needs only be performed over a reduced set of intrinsic parameters [58, 50, 60].
Techniques reviewed in this paper have been used in the data analysis of prototypes of gravitationalwave detectors [73, 71, 6] and in the data analysis of presently working gravitationalwave detectors [90, 15, 3, 2, 1].
We use units such that the velocity of light c=1.
2 Response of a Detector to a Gravitational Wave
There are two main methods to detect gravitational waves which have been implemented in the currently working instruments. One method is to measure changes induced by gravitational waves on the distances between freely moving test masses using coherent trains of electromagnetic waves. The other method is to measure the deformation of large masses at their resonance frequencies induced by gravitational waves. The first idea is realized in laser interferometric detectors and Doppler tracking experiments [82, 65] whereas the second idea is implemented in resonant mass detectors [13].
Let us consider the response to a plane gravitational wave of a freely falling configuration of masses. It is enough to consider a configuration of three masses shown in Figure 1 to obtain the response for all currently working and planned detectors. Two masses model a Doppler tracking experiment where one mass is the Earth and the other one is a distant spacecraft. Three masses model a groundbased laser interferometer where the masses are suspended from seismically isolated supports or a spaceborne interferometer where the three masses are shielded in satellites driven by dragfree control systems.
In Figure 1 we have introduced the following notation: O denotes the origin of the TT coordinate system related to the passing gravitational wave, x_{a} (a = 1, 2, 3) are 3vectors joining O and the masses, n_{a} and L_{a}(a = 1, 2, 3) are, respectively, 3vectors of unit Euclidean length along the lines joining the masses and the coordinate Euclidean distances between the masses, where a is the label of the opposite mass. Let us also denote by k the unit 3vector directed from the origin O to the source of the gravitational wave. We first assume that the spatial coordinates of the masses do not change in time.
Let ν_{0} be the frequency of the coherent beam used in the detector (laser light in the case of an interferometer and radio waves in the case of Doppler tracking). Let y_{21} be the relative change Δν/ν_{0} of frequency induced by a transverse, traceless, plane gravitational wave on the coherent beam travelling from the mass 2 to the mass 1, and let y_{31} be the relative change of frequency induced on the beam travelling from the mass 3 to the mass 1. The frequency shifts y_{21} and y_{31} are given by [37, 10, 83]
where (here T denotes matrix transposition)
In Equation (3) H is the threedimensional matrix of the spatial metric perturbation produced by the wave in the TT coordinate system. If one chooses spatial TT coordinates such that the wave is travelling in the +z direction, then the matrix H is given by
where h_{+} and h_{×} are the two polarizations of the wave.
Real gravitationalwave detectors do not stay at rest with respect to the TT coordinate system related to the passing gravitational wave, because they also move in the gravitational field of the solar system bodies, as in the case of the LISA spacecraft, or are fixed to the surface of Earth, as in the case of Earthbased laser interferometers or resonant bar detectors. Let us choose the origin O of the TT coordinate system to coincide with the solar system barycenter (SSB). The motion of the detector with respect to the SSB will modulate the gravitationalwave signal registered by the detector. One can show that as far as the velocities of the masses (modelling the detector’s parts) with respect to the SSB are nonrelativistic, which is the case for all existing or planned detectors, the Equations (1) and (2) can still be used, provided the 3vectors x_{a} and n_{a} (a = 1, 2, 3) will be interpreted as made of the timedependent components describing the motion of the masses with respect to the SSB.
It is often convenient to introduce the proper reference frame of the detector with coordinates \(({x^{\hat \alpha}})\). Because the motion of this frame with respect to the SSB is nonrelativistic, we can assume that the transformation between the SSBrelated coordinates (x^{α}) and the detector’s coordinates \(({x^{\hat \alpha}})\) has the form
where the functions \(x_{\hat O}^{\hat i}(t)\) describe the motion of the origin Ô of the proper reference frame with respect to the SSB, and the functions \(O_j^{\hat i}(t)\) account for the different orientations of the spatial axes of the two reference frames. One can compute some of the quantities entering Equations (1) and (2) in the detector’s coordinates rather than in the TT coordinates. For instance, the matrix Ĥ of the waveinduced spatial metric perturbation in the detector’s coordinates is related to the matrix of the spatial metric perturbation produced by the wave in the TT coordinate system through equation
where the matrix O has elements \({O^{\hat i}}_j\). If the transformation matrix O is orthogonal, then O^{−1} = O^{T}, and Equation (6) simplifies to
See [23, 42, 50, 60] for more details.
For a standard Michelson, equalarm interferometric configuration Δν is given in terms of oneway frequency changes y_{21} and y_{31} (see Equations (1) and (2) with L_{2} = L_{3} = L, where we assume that the mass 1 corresponds to the corner station of the interferometer) by the expression [93]
In the longwavelength approximation Equation (8) reduces to
The difference of the phase fluctuations ΔΦ(t) measured, say, by a photo detector, is related to the corresponding relative frequency fluctuations Δν by
By virtue of Equation (9) the phase change can be written as
where the function h,
is the response of the interferometer to a gravitational wave in the longwavelength approximation. In this approximation the response of a laser interferometer is usually derived from the equation of geodesic deviation (then the response is defined as the difference between the relative waveinduced changes of the proper lengths of the two arms, i.e., h(t):= ΔL(t)/L). There are important cases where the longwavelength approximation is not valid. These include the spaceborne LISA detector for gravitationalwave frequencies larger than a few mHz and satellite Doppler tracking measurements.
In the case of a bar detector the longwavelength approximation is very accurate and the detector’s response is defined as h_{B}(t):= ΔL(t)/L, where ΔL is the waveinduced change of the proper length L of the bar. The response h_{B} is given by
where n is the unit vector along the symmetry axis of the bar.
In most cases of interest the response of the detector to a gravitational wave can be written as a linear combination of four constant amplitudes a^{(k)},
where the four functions h^{(k)} depend on a set of parameters ξ^{μ} but are independent of the parameters a^{(k)}. The parameters a^{(k)} are called extrinsic parameters whereas the parameters ξ^{μ} are called intrinsic. In the longwavelength approximation the functions h^{(k)} are given by
where Φ(t; ξ^{μ}) is the phase modulation of the signal and u(t;ξ^{μ}), v(t; ξ^{μ}) are slowly varying amplitude modulations.
Equation (14) is a model of the response of the spacebased detector LISA to gravitational waves from a binary system [60], whereas Equation (15) is a model of the response of a groundbased detector to a continuous source of gravitational waves like a rotating neutron star [50]. The gravitationalwave signal from spinning neutron stars may consist of several components of the form (14). For short observation times over which the amplitude modulation functions are nearly constant, the response can be approximated by
where A_{0} and Φ_{0} are constant amplitude and initial phase, respectively, and g(t; ξ^{μ}) is a slowly varying function of time. Equation (16) is a good model for a response of a detector to the gravitational wave from a coalescing binary system [92, 22]. We would like to stress that not all deterministic gravitationalwave signals may be cast into the general form (14).
3 Statistical Theory of Signal Detection
The gravitationalwave signal will be buried in the noise of the detector and the data from the detector will be a random process. Consequently the problem of extracting the signal from the noise is a statistical one. The basic idea behind the signal detection is that the presence of the signal changes the statistical characteristics of the data x, in particular its probability distribution. When the signal is absent the data have probability density function (pdf) p_{0}(x), and when the signal is present the pdf is p_{1}(x).
A full exposition of the statistical theory of signal detection that is outlined here can be found in the monographs [102, 56, 98, 96, 66, 44, 77]. A general introduction to stochastic processes is given in [100]. Advanced treatment of the subject can be found in [64, 101].
The problem of detecting the signal in noise can be posed as a statistical hypothesis testing problem. The null hypothesis H_{0} is that the signal is absent from the data and the alternative hypothesis H_{1} is that the signal is present. A hypothesis test (or decision rule) δ is a partition of the observation set into two sets, \({\mathcal R}\) and its complement \(R^\prime\). If data are in \({\mathcal R}\) we accept the null hypothesis, otherwise we reject it. There are two kinds of errors that we can make. A type I error is choosing hypothesis H_{1} when H_{0} is true and a type II error is choosing H_{0} when H_{1} is true. In signal detection theory the probability of a type I error is called the false alarm probability, whereas the probability of a type II error is called the false dismissal probability. 1 − (false dismissal probability) is the probability of detection of the signal. In hypothesis testing the probability of a type I error is called the significance of the test, whereas 1 − (probability of type II error) is called the power of the test.
The problem is to find a test that is in some way optimal. There are several approaches to find such a test. The subject is covered in detail in many books on statistics, for example see references [54, 41, 62].
3.1 Bayesian approach
In the Bayesian approach we assign costs to our decisions; in particular we introduce positive numbers C_{ij}, i, j = 0, 1, where C_{ij} is the cost incurred by choosing hypothesis H_{i} when hypothesis H_{j} is true. We define the conditional risk R of a decision rule δ for each hypothesis as
where P_{j} is the probability distribution of the data when hypothesis H_{j} is true. Next we assign probabilities π_{0} and π_{1} = 1 − π_{0} to the occurrences of hypothesis H_{0} and H_{1}, respectively. These probabilities are called a priori probabilities or priors. We define the Bayes risk as the overall average cost incurred by the decision rule δ:
Finally we define the Bayes rule as the rule that minimizes the Bayes risk r(δ).
3.2 Minimax approach
Very often in practice we do not have the control over or access to the mechanism generating the state of nature and we are not able to assign priors to various hypotheses. In such a case one criterion is to seek a decision rule that minimizes, over all δ, the maximum of the conditional risks, R_{0}(δ) and R_{1}(δ). A decision rule that fulfills that criterion is called minimax rule.
3.3 NeymanPearson approach
In many problems of practical interest the imposition of a specific cost structure on the decisions made is not possible or desirable. The NeymanPearson approach involves a tradeoff between the two types of errors that one can make in choosing a particular hypothesis. The NeymanPearson design criterion is to maximize the power of the test (probability of detection) subject to a chosen significance of the test (false alarm probability).
3.4 Likelihood ratio test
It is remarkable that all three very different approaches — Bayesian, minimax, and NeymanPearson — lead to the same test called the likelihood ratio test [34]. The likelihood ratio Λ is the ratio of the pdf when the signal is present to the pdf when it is absent:
We accept the hypothesis H_{1} if Λ > k, where k is the threshold that is calculated from the costs C_{ij}, priors π_{i}, or the significance of the test depending on what approach is being used.
3.4.1 Gaussian case — The matched filter
Let h be the gravitationalwave signal and let n be the detector noise. For convenience we assume that the signal h is a continuous function of time t and that the noise n is a continuous random process. Results for the discrete time data that we have in practice can then be obtained by a suitable sampling of the continuousintime expressions. Assuming that the noise is additive the data x can be written as
In addition, if the noise is a zeromean, stationary, and Gaussian random process, the log likelihood function is given by
where the scalar product (·  ·) is defined by
In Equation (22) ℜ denotes the real part of a complex expression, the tilde denotes the Fourier transform, the asterisk is complex conjugation, and \({\tilde S}\) is the onesided spectral density of the noise in the detector, which is defined through equation
where E denotes the expectation value.
From the expression (21) we see immediately that the likelihood ratio test consists of correlating the data x with the signal h that is present in the noise and comparing the correlation to a threshold. Such a correlation is called the matched filter. The matched filter is a linear operation on the data.
An important quantity is the optimal signaltonoise ratio ρ defined by
We see in the following that ρ determines the probability of detection of the signal. The higher the signaltonoise ratio the higher the probability of detection.
An interesting property of the matched filter is that it maximizes the signaltonoise ratio over all linear filters [34]. This property is independent of the probability distribution of the noise.
4 Parameter Estimation
Very often we know the waveform of the signal that we are searching for in the data in terms of a finite number of unknown parameters. We would like to find optimal procedures of estimating these parameters. An estimator of a parameter θ is a function \(\hat \theta (x)\) that assigns to each data the “best” guess of the true value of θ. Note that because \(\hat \theta (x)\) depends on the random data it is a random variable. Ideally we would like our estimator to be (i) unbiased, i.e., its expectation value to be equal to the true value of the parameter, and (ii) of minimum variance. Such estimators are rare and in general difficult to find. As in the signal detection there are several approaches to the parameter estimation problem. The subject is exposed in detail in reference [63]. See also [103] for a concise account.
4.1 Bayesian estimation
We assign a cost function C(θ′, θ) of estimating the true value of θ as θ′. We then associate with an estimator \({\hat \theta}\) a conditional risk or cost averaged over all realizations of data x for each value of the parameter θ:
where X is the set of observations and p(x, θ) is the joint probability distribution of data x and parameter θ. We further assume that there is a certain a priori probability distribution π(θ) of the parameter θ. We then define the Bayes estimator as the estimator that minimizes the average risk defined as
where E is the expectation value with respect to an a priori distribution π, and Θ is the set of observations of the parameter θ. It is not difficult to show that for a commonly used cost function
the Bayesian estimator is the conditional mean of the parameter θ given data x, i.e.,
where p(θx) is the conditional probability density of parameter θ given the data x.
4.2 Maximum a posteriori probability estimation
Suppose that in a given estimation problem we are not able to assign a particular cost function C(θ′, θ). Then a natural choice is a uniform cost function equal to 0 over a certain interval I_{θ} of the parameter θ. From Bayes theorem [20] we have
where p(x) is the probability distribution of data x. Then from Equation (26) one can deduce that for each data x the Bayes estimate is any value of θ that maximizes the conditional probability p(θx). The density p(θx) is also called the a posteriori probability density of parameter θ and the estimator that maximizes p(θx) is called the maximum a posteriori (MAP) estimator. It is denoted by \({{\hat \theta}_{{\rm{MAP}}}}\). We find that the MAP estimators are solutions of the following equation
which is called the MAP equation.
4.3 Maximum likelihood estimation
Often we do not know the a priori probability density of a given parameter and we simply assign to it a uniform probability. In such a case maximization of the a posteriori probability is equivalent to maximization of the probability density p(x, θ) treated as a function of θ. We call the function l(θ, x):= p(x, θ) the likelihood function and the value of the parameter θ that maximizes l(θ, x) the maximum likelihood (ML) estimator. Instead of the function l we can use the function Λ(θ, x) = l(θ, x)/p(x) (assuming that p(x) > 0). Λ is then equivalent to the likelihood ratio [see Equation (19)] when the parameters of the signal are known. Then the ML estimators are obtained by solving the equation
which is called the ML equation.
4.3.1 Gaussian case
For the general gravitationalwave signal defined in Equation (14) the log likelihood function is given by
where the components of the column vector N and the matrix M are given by
with x(t) = n(t) + h(t), and where n(t) is a zeromean Gaussian random process. The ML equations for the extrinsic parameters a can be solved explicitly and their ML estimators â are given by
Substituting â into log Λ we obtain a function
that we call the \({\mathcal F}\)statistic. The \({\mathcal F}\)statistic depends (nonlinearly) only on the intrinsic parameters \({\xi ^\mu}\).
Thus the procedure to detect the signal and estimate its parameters consists of two parts. The first part is to find the (local) maxima of the \({\mathcal F}\)statistic in the intrinsic parameter space. The ML estimators of the intrinsic parameters are those for which the \({\mathcal F}\)statistic attains a maximum. The second part is to calculate the estimators of the extrinsic parameters from the analytic formula (34), where the matrix M and the correlations N are calculated for the intrinsic parameters equal to their ML estimators obtained from the first part of the analysis. We call this procedure the maximum likelihood detection. See Section 4.8 for a discussion of the algorithms to find the (local) maxima of the \({\mathcal F}\)statistic.
4.4 Fisher information
It is important to know how good our estimators are. We would like our estimator to have as small variance as possible. There is a useful lower bound on variances of the parameter estimators called CramerRao bound. Let us first introduce the Fisher information matrix Γ with the components defined by
The CramèrRao bound states that for unbiased estimators the covariance matrix of the estimators C ≥ Γ^{−1}. (The inequality A ≥ B for matrices means that the matrix A−B is nonnegative definite.)
A very important property of the ML estimators is that asymptotically (i.e., for a signaltonoise ratio tending to infinity) they are (i) unbiased, and (ii) they have a Gaussian distribution with covariance matrix equal to the inverse of the Fisher information matrix.
4.4.1 Gaussian case
In the case of Gaussian noise the components of the Fisher matrix are given by
For the case of the general gravitationalwave signal defined in Equation (14) the set of the signal parameters θ splits naturally into extrinsic and intrinsic parameters: θ = (a^{(k)}, ξ^{μ}). Then the Fisher matrix can be written in terms of block matrices for these two sets of parameters as
where the top left block corresponds to the extrinsic parameters, the bottom right block corresponds to the intrinsic parameters, the superscript T denotes here transposition over the extrinsic parameter indices, and the dot stands for the matrix multiplication with respect to these parameters. Matrix M is given by Equation (33), and the matrices F and S are defined as follows:
The covariance matrix C, which approximates the expected covariances of the ML parameter estimators, is defined as Γ^{−1}. Using the standard formula for the inverse of a block matrix [67] we have
where
We call \({{\bar \Gamma}^{\mu \nu}}\) (the Schur complement of M) the projected Fisher matrix (onto the space of intrinsic parameters). Because the projected Fisher matrix is the inverse of the intrinsicparameter submatrix of the covariance matrix C, it expresses the information available about the intrinsic parameters that takes into account the correlations with the extrinsic parameters. Note that \({{\bar \Gamma}^{\mu \nu}}\) is still a function of the putative extrinsic parameters. We next define the normalized projected Fisher matrix
where \(\rho = \sqrt {{{\rm{a}}^{\rm{T}}} \cdot {\rm{M}} \cdot {\rm{a}}}\) is the signaltonoise ratio. From the Rayleigh principle [67] follows that the minimum value of the component \(\bar \Gamma _n^{\mu \nu}\) is given by the smallest eigenvalue (taken with respect to the extrinsic parameters) of the matrix ((S − F^{T} · M^{−1} · F) · M^{−1})^{μν}. Similarly, the maximum value of the component \(\bar \Gamma _n^{\mu \nu}\) is given by the largest eigenvalue of that matrix. Because the trace of a matrix is equal to the sum of its eigenvalues, the matrix
where the trace is taken over the extrinsicparameter indices, expresses the information available about the intrinsic parameters, averaged over the possible values of the extrinsic parameters. Note that the factor 1/4 is specific to the case of four extrinsic parameters. We call \({{\tilde \Gamma}^{\mu \nu}}\) the reduced Fisher matrix. This matrix is a function of the intrinsic parameters alone. We see that the reduced Fisher matrix plays a key role in the signal processing theory that we review here. It is used in the calculation of the threshold for statistically significant detection and in the formula for the number of templates needed to do a given search.
For the case of the signal
the normalized projected Fisher matrix \({{\bar \Gamma}_n}\) is independent of the extrinsic parameters A_{0} and Φ_{0}, and it is equal to the reduced matrix \({\tilde \Gamma}\) [74]. The components of \({\tilde \Gamma}\) are given by
where \(\Gamma _0^{ij}\) is the Fisher matrix for the signal g(t; ξ^{μ}) cos(Φ(t; ξ^{μ}) − Φ_{0}).
Fisher matrix has been extensively used to assess the accuracy of estimation of astrophysically interesting parameters of gravitationalwave signals. First calculations of Fisher matrix concerned gravitationalwave signals from inspiralling binaries in quadrupole approximation [40, 58] and from quasinormal modes of Kerr black hole [38]. Cutler and Flanagan [32] initiated the study of the implications of higher PN order phasing formula as applied to the parameter estimation of inspiralling binaries. They used the 1.5PN phasing formula to investigate the problem of parameter estimation, both for spinning and nonspinning binaries, and examined the effect of the spinorbit coupling on the estimation of parameters. The effect of the 2PN phasing formula was analyzed independently by Poisson and Will [76] and Królak, Kokkotas and Schäfer [57]. In both of these works the focus was to understand the new spinspin coupling term appearing at the 2PN order when the spins were aligned perpendicular to the orbital plane. Compared to [57], [76] also included a priori information about the magnitude of the spin parameters, which then leads to a reduction in the rms errors in the estimation of mass parameters. The case of 3.5PN phasing formula was studied in detail by Arun et al. [12]. Inclusion of 3.5PN effects leads to an improved estimate of the binary parameters. Improvements are relatively smaller for lighter binaries.
Various authors have investigated the accuracy with which LISA detector can determine binary parameters including spin effects. Cutler [30] determined LISA’s angular resolution and evaluated the errors of the binary masses and distance considering spins aligned or antialigned with the orbital angular momentum. Hughes [46] investigated the accuracy with which the redshift can be estimated (if the cosmological parameters are derived independently), and considered the blackhole ringdown phase in addition to the inspiralling signal. Seto [89] included the effect of finite armlength (going beyond the long wavelength approximation) and found that the accuracy of the distance determination and angular resolution improve. This happens because the response of the instrument when the armlength is finite depends strongly on the location of the source, which is tightly correlated with the distance and the direction of the orbital angular momentum. Vecchio [97] provided the first estimate of parameters for precessing binaries when only one of the two supermassive black holes carries spin. He showed that modulational effects decorrelate the binary parameters to some extent, resulting in a better estimation of the parameters compared to the case when spins are aligned or antialigned with orbital angular momentum. Hughes and Menou [47] studied a class of binaries, which they called “golden binaries,” for which the inspiral and ringdown phases could be observed with good enough precision to carry out valuable tests of strongfield gravity. Berti, Buonanno and Will [21] have shown that inclusion of nonprecessing spinorbit and spinspin terms in the gravitationalwave phasing generally reduces the accuracy with which the parameters of the binary can be estimated. This is not surprising, since the parameters are highly correlated, and adding parameters effectively dilutes the available information.
4.5 False alarm and detection probabilities — Gaussian case
4.5.1 Statistical properties of the \({\mathcal F}\)statistic
We first present the false alarm and detection pdfs when the intrinsic parameters of the signal are known. In this case the statistic \({\mathcal F}\) is a quadratic form of the random variables that are correlations of the data. As we assume that the noise in the data is Gaussian and the correlations are linear functions of the data, \({\mathcal F}\) is a quadratic form of the Gaussian random variables. Consequently \({\mathcal F}\)statistic has a distribution related to the χ^{2} distribution. One can show (see Section III B in [49]) that for the signal given by Equation (14), \(2{\mathcal F}\) has a χ^{2} distribution with 4 degrees of freedom when the signal is absent and noncentral χ^{2} distribution with 4 degrees of freedom and noncentrality parameter equal to signaltonoise ratio (hh) when the signal is present.
As a result the pdfs p_{0} and p_{1} of \({\mathcal F}\) when the intrinsic parameters are known and when respectively the signal is absent and present are given by
where n is the number of degrees of freedom of χ^{2} distributions and I_{n/2−1} is the modified Bessel function of the first kind and order n/2−1. The false alarm probability P_{F} is the probability that \({\mathcal F}\) exceeds a certain threshold \({{\mathcal F}_0}\) when there is no signal. In our case we have
The probability of detection P_{D} is the probability that \({\mathcal F}\) exceeds the threshold \({{\mathcal F}_0}\) when the signaltonoise ratio is equal to ρ:
The integral in the above formula can be expressed in terms of the generalized Marcum Qfunction [94, 44], Q(α, β) = P_{D}(α, β^{2}/2). We see that when the noise in the detector is Gaussian and the intrinsic parameters are known, the probability of detection of the signal depends on a single quantity: the optimal signaltonoise ratio ρ.
4.5.2 False alarm probability
Next we return to the case when the intrinsic parameters ξ are not known. Then the statistic \({\mathcal F}(\xi)\) given by Equation (35) is a certain generalized multiparameter random process called the random field (see Adler’s monograph [4] for a comprehensive discussion of random fields). If the vector ξ has one component the random field is simply a random process. For random fields we can define the autocovariance function \({\mathcal C}\) just in the same way as we define such a function for a random process:
where ξ and ξ′ are two values of the intrinsic parameter set, and E_{0} is the expectation value when the signal is absent. One can show that for the signal (14) the autocovariance function \({\mathcal C}\) is given
where
We have \({\mathcal C}(\xi, \xi) = 1\).
One can estimate the false alarm probability in the following way [50]. The autocovariance function \({\mathcal C}\) tends to zero as the displacement Δξ = ξ′ − ξ increases (it is maximal for Δξ = 0). Thus we can divide the parameter space into elementary cells such that in each cell the autocovariance function \({\mathcal C}\) is appreciably different from zero. The realizations of the random field within a cell will be correlated (dependent), whereas realizations of the random field within each cell and outside the cell are almost uncorrelated (independent). Thus the number of cells covering the parameter space gives an estimate of the number of independent realizations of the random field. The correlation hypersurface is a closed surface defined by the requirement that at the boundary of the hypersurface the correlation \({\mathcal C}\) equals half of its maximum value. The elementary cell is defined by the equation
for ξ at cell center and ξ′ on cell boundary. To estimate the number of cells we perform the Taylor expansion of the autocorrelation function up to the secondorder terms:
As \({\mathcal C}\) attains its maximum value when ξ−ξ′ = 0, we have
Let us introduce the symmetric matrix
then the approximate equation for the elementary cell is given by
It is interesting to find a relation between the matrix G and the Fisher matrix. One can show (see [60], Appendix B) that the matrix G is precisely equal to the reduced Fisher matrix \({\tilde \Gamma}\) given by Equation (43).
Let K be the number of the intrinsic parameters. If the components of the matrix G are constant (independent of the values of the parameters of the signal) the above equation is an equation for a hyperellipse. The Kdimensional Euclidean volume V_{cell} of the elementary cell defined by Equation (57) equals
where Γ denotes the Gamma function. We estimate the number N_{c} of elementary cells by dividing the total Euclidean volume V of the Kdimensional parameter space by the volume V_{cell} of the elementary cell, i.e. we have
The components of the matrix G are constant for the signal h(t; A_{0}, Φ_{0}, ξ^{μ}) = A_{0}cos (Φ(t; Φ^{μ}) − Φ_{0}) when the phase Φ(t; ξ^{μ}) is a linear function of the intrinsic parameters ξ^{μ}.
To estimate the number of cells in the case when the components of the matrix G are not constant, i.e. when they depend on the values of the parameters, we write Equation (59) as
This procedure can be thought of as interpreting the matrix G as the metric on the parameter space. This interpretation appeared for the first time in the context of gravitationalwave data analysis in the work by Owen [74], where an analogous integral formula was proposed for the number of templates needed to perform a search for gravitationalwave signals from coalescing binaries.
The concept of number of cells was introduced in [50] and it is a generalization of the idea of an effective number of samples introduced in [36] for the case of a coalescing binary signal.
We approximate the probability distribution of \({\mathcal F}(\xi)\) in each cell by the probability \({p_0}({\mathcal F})\) when the parameters are known [in our case by probability given by Equation (46)]. The values of the statistic \({\mathcal F}\) in each cell can be considered as independent random variables. The probability that \({\mathcal F}\) does not exceed the threshold \({{\mathcal F}_0}\) in a given cell is \(1  {P_{\rm{F}}}({{\mathcal F}_0})\), where \({P_{\rm{F}}}({{\mathcal F}_0})\) is given by Equation (48). Consequently the probability that \({\mathcal F}\) does not exceed the threshold \({{\mathcal F}_0}\) in all the N_{c} cells is \({[1  {P_{\rm{F}}}({{\mathcal F}_0})]^{{N_{\rm{c}}}}}\). The probability \(P_F^T\) that \({\mathcal F}\) exceeds \({{\mathcal F}_0}\) in one or more cell is thus given by
This by definition is the false alarm probability when the phase parameters are unknown. The number of false alarms N_{F} is given by
A different approach to the calculation of the number of false alarms using the Euler characteristic of level crossings of a random field is described in [49].
It was shown (see [29]) that for any finite \({{\mathcal F}_0}\) and N_{c}, Equation (61) provides an upper bound for the false alarm probability. Also in [29] a tighter upper bound for the false alarm probability was derived by modifying a formula obtained by Mohanty [68]. The formula amounts essentially to introducing a suitable coefficient multiplying the number of cells N_{c}.
4.5.3 Detection probability
When the signal is present a precise calculation of the pdf of \({\mathcal F}\) is very difficult because the presence of the signal makes the data random process x(t) nonstationary. As a first approximation we can estimate the probability of detection of the signal when the parameters are unknown by the probability of detection when the parameters of the signal are known [given by Equation (49)]. This approximation assumes that when the signal is present the true values of the phase parameters fall within the cell where \({\mathcal F}\) has a maximum. This approximation will be the better the higher the signaltonoise ratio ρ is.
4.6 Number of templates
To search for gravitationalwave signals we evaluate the \({\mathcal F}\)statistic on a grid in parameter space. The grid has to be sufficiently fine such that the loss of signals is minimized. In order to estimate the number of points of the grid, or in other words the number of templates that we need to search for a signal, the natural quantity to study is the expectation value of the \({\mathcal F}\)statistic when the signal is present. We have
The components of the matrix Q are given in Equation (52). Let us rewrite the expectation value (63) in the following form,
where ρ is the signaltonoise ratio. Let us also define the normalized correlation function
From the Rayleigh principle [67] it follows that the minimum of the normalized correlation function is equal to the smallest eigenvalue of the normalized matrix Q^{T}·M ′^{−1}·Q·M^{−1}, whereas the maximum is given by its largest eigenvalue. We define the reduced correlation function as
As the trace of a matrix equals the sum of its eigenvalues, the reduced correlation function \({\mathcal C}\) is equal to the average of the eigenvalues of the normalized correlation function \({{\mathcal C}_{\rm{n}}}\). In this sense we can think of the reduced correlation function as an “average” of the normalized correlation function. The advantage of the reduced correlation function is that it depends only on the intrinsic parameters ξ, and thus it is suitable for studying the number of grid points on which the \({\mathcal F}\)statistic needs to be evaluated. We also note that the normalized correlation function \({\mathcal C}\) precisely coincides with the autocovariance function \({\mathcal C}\) of the \({\mathcal F}\)statistic given by Equation (51).
Like in the calculation of the number of cells in order to estimate the number of templates we perform a Taylor expansion of \({\mathcal C}\) up to second order terms around the true values of the parameters, and we obtain an equation analogous to Equation (57),
where G is given by Equation (56). By arguments identical to those in deriving the formula for the number of cells we arrive at the following formula for the number of templates:
When C_{0} = 1/2 the above formula coincides with the formula for the number N_{c} of cells, Equation (60). Here we would like to place the templates sufficiently closely so that the loss of signals is minimized. Thus 1 − C_{0} needs to be chosen sufficiently small. The formula (68) for the number of templates assumes that the templates are placed in the centers of hyperspheres and that the hyperspheres fill the parameter space without holes. In order to have a tiling of the parameter space without holes we can place the templates in the centers of hypercubes which are inscribed in the hyperspheres. Then the formula for the number of templates reads
For the case of the signal given by Equation (16) our formula for number of templates is equivalent to the original formula derived by Owen [74]. Owen [74] has also introduced a geometric approach to the problem of template placement involving the identification of the Fisher matrix with a metric on the parameter space. An early study of the template placement for the case of coalescing binaries can be found in [84, 35, 19]. Applications of the geometric approach of Owen to the case of spinning neutron stars and supernova bursts are given in [24, 11].
The problem of how to cover the parameter space with the smallest possible number of templates, such that no point in the parameter space lies further away from a grid point than a certain distance, is known in mathematical literature as the covering problern [28]. The maximum distance of any point to the next grid point is called the covering radius R. An important class of coverings are lattice coverings. We define a lattice in Kdimensional Euclidean space ℝ^{>K} to be the set of points including 0 such that if u and v are lattice points, then also u + v and u−v are lattice points. The basic building block of a lattice is called the fundamental region. A lattice covering is a covering of ℝ^{>K} by spheres of covering radius R, where the centers of the spheres form a lattice. The most important quantity of a covering is its thickness Θ defined as
In the case of a twodimensional Euclidean space the best covering is the hexagonal covering and its thickness ≃ 1.21. For dimensions higher than 2 the best covering is not known. We know however the best lattice covering for dimensions K ≤ 23. These are socalled \(A_K^\ast\) lattices which have a thickness \({\Theta _{A_K^\ast}}\) equal to
where V_{K} is the volume of the Kdimensional sphere of unit radius.
For the case of spinning neutron stars a 3dimensional grid was constructed consisting of prisms with hexagonal bases [16]. This grid has a thickness around 1.84 which is much better than the cubic grid which has thickness of approximately 2.72. It is worse than the best lattice covering which has the thickness around 1.46. The advantage of an \(A_K^\ast\) lattice over the hypercubic lattice grows exponentially with the number of dimensions.
4.7 Suboptimal filtering
To extract signals from the noise one very often uses filters that are not optimal. We may have to choose an approximate, suboptimal filter because we do not know the exact form of the signal (this is almost always the case in practice) or in order to reduce the computational cost and to simplify the analysis. The most natural and simplest way to proceed is to use as our statistic the \({\mathcal F}\)statistic where the filters \(h_k^\prime(t;\zeta)\) are the approximate ones instead of the optimal ones matched to the signal. In general the functions \(h_k^\prime(t;\zeta)\) will be different from the functions h_{k}(t; ξ) used in optimal filtering, and also the set of parameters ζ will be different from the set of parameters ξ in optimal filters. We call this procedure the suboptimal filtering and we denote the suboptimal statistic by \({{\mathcal F}_{\rm{s}}}\).
We need a measure of how well a given suboptimal filter performs. To find such a measure we calculate the expectation value of the suboptimal statistic. We get
where
Let us rewrite the expectation value \({\rm{E}}\left[ {{{\mathcal F}_{\rm{s}}}} \right]\) in the following form,
where ρ is the optimal signaltonoise ratio. The expectation value \({\rm{E}}\left[ {{{\mathcal F}_{\rm{s}}}} \right]\) reaches its maximum equal to 2 + ρ^{2}/2 when the filter is perfectly matched to the signal. A natural measure of the performance of a suboptimal filter is the quantity FF defined by
We call the quantity FF the generalized fitting factor.
In the case of a signal given by
the generalized fitting factor defined above reduces to the fitting factor introduced by Apostolatos [9]:
The fitting factor is the ratio of the maximal signaltonoise ratio that can be achieved with suboptimal filtering to the signaltonoise ratio obtained when we use a perfectly matched, optimal filter. We note that for the signal given by Equation (76), FF is independent of the value of the amplitude A_{0}. For the general signal with 4 constant amplitudes it follows from the Rayleigh principle that the fitting factor is the maximum of the largest eigenvalue of the matrix Q^{T} · M′^{−1} · Q · M^{−1} over the intrinsic parameters of the signal.
For the case of a signal of the form
where Φ_{0} is a constant phase, the maximum over Φ_{0} in Equation (77) can be obtained analytically. Moreover, assuming that over the bandwidth of the signal the spectral density of the noise is constant and that over the observation time cos(Φ(t; ξ)) oscillates rapidly, the fitting factor is approximately given by
In designing suboptimal filters one faces the issue of how small a fitting factor one can accept. A popular rule of thumb is accepting FF = 0. 97. Assuming that the amplitude of the signal and consequently the signaltonoise ratio decreases inversely proportional to the distance from the source this corresponds to 10% loss of the signals that would be detected by a matched filter.
Proposals for good suboptimal (search) templates for the case of coalescing binaries are given in [26, 91] and for the case spinning neutron stars in [49, 16].
4.8 Algorithms to calculate the \({\mathcal F}\)statistic
4.8.1 The twostep procedure
In order to detect signals we search for threshold crossings of the \({\mathcal F}\)statistic over the intrinsic parameter space. Once we have a threshold crossing we need to find the precise location of the maximum of \({\mathcal F}\) in order to estimate accurately the parameters of the signal. A satisfactory procedure is the twostep procedure. The first step is a coarse search where we evaluate \({\mathcal F}\) on a coarse grid in parameter space and locate threshold crossings. The second step, called fine search, is a refinement around the region of parameter space where the maximum identified by the coarse search is located.
There are two methods to perform the fine search. One is to refine the grid around the threshold crossing found by the coarse search [70, 68, 91, 88], and the other is to use an optimization routine to find the maximum of \({\mathcal F}\) [49, 60]. As initial value to the optimization routine we input the values of the parameters found by the coarse search. There are many maximization algorithms available. One useful method is the NelderMead algorithm [61] which does not require computation of the derivatives of the function being maximized.
4.8.2 Evaluation of the \({\mathcal F}\)statistic
Usually the grid in parameter space is very large and it is important to calculate the optimum statistic as efficiently as possible. In special cases the \({\mathcal F}\)statistic given by Equation (35) can be further simplified. For example, in the case of coalescing binaries F can be expressed in terms of convolutions that depend on the difference between the timeofarrival (TOA) of the signal and the TOA parameter of the filter. Such convolutions can be efficiently computed using Fast Fourier Transforms (FFTs). For continuous sources, like gravitational waves from rotating neutron stars observed by groundbased detectors [49] or gravitational waves form stellar mass binaries observed by spaceborne detectors [60], the detection statistic \({\mathcal F}\) involves integrals of the general form
where \({{\tilde \xi}^\mu}\) are the intrinsic parameters excluding the frequency parameter ω, m is the amplitude modulation function, and ωΦ_{mod} the phase modulation function. The amplitude modulation function is slowly varying comparing to the exponential terms in the integral (80). We see that the integral (80) can be interpreted as a Fourier transform (and computed efficiently with an FFT), if Φ_{mod} = 0 and if m does not depend on the frequency ω. In the longwavelength approximation the amplitude function m does not depend on the frequency. In this case Equation (80) can be converted to a Fourier transform by introducing a new time variable t_{b} [87],
Thus in order to compute the integral (80), for each set of the intrinsic parameters \({{\tilde \xi}^\mu}\) we multiply the data by the amplitude modulation function m, resample according to Equation (81), and perform the FFT. In the case of LISA detector data when the amplitude modulation m depends on frequency we can divide the data into several bandpassed data sets, choosing the bandwidth for each set sufficiently small so that the change of m exp(iω_{mod}) is small over the band. In the integral (80) we can then use as the value of the frequency in the amplitude and phase modulation function the maximum frequency of the band of the signal (see [60] for details).
4.8.3 Comparison with the CramèrRao bound
In order to test the performance of the maximization method of the \({\mathcal F}\) statistic it is useful to perform Monte Carlo simulations of the parameter estimation and compare the variances of the estimators with the variances calculated from the Fisher matrix. Such simulations were performed for various gravitationalwave signals [55, 19, 49]. In these simulations we observe that above a certain signaltonoise ratio, that we call the threshold signaltonoise ratio, the results of the Monte Carlo simulations agree very well with the calculations of the rms errors from the inverse of the Fisher matrix. However, below the threshold signaltonoise ratio they differ by a large factor. This threshold effect is wellknown in signal processing [96]. There exist more refined theoretical bounds on the rms errors that explain this effect, and they were also studied in the context of the gravitationalwave signal from a coalescing binary [72]. Use of the Fisher matrix in the assessment of the parameter estimators has been critically examined in [95] where a criterion has been established for signaltonoise ratio above which the inverse of the Fisher matrix approximates well covariance of the estimators of the parameters.
Here we present a simple model that explains the deviations from the covariance matrix and reproduces well the results of the Monte Carlo simulations. The model makes use of the concept of the elementary cell of the parameter space that we introduced in Section 4.5.2. The calculation given below is a generalization of the calculation of the rms error for the case of a monochromatic signal given by Rife and Boorstyn [81].
When the values of parameters of the template that correspond to the maximum of the functional \({\mathcal F}\) fall within the cell in the parameter space where the signal is present, the rms error is satisfactorily approximated by the inverse of the Fisher matrix. However, sometimes as a result of noise the global maximum is in the cell where there is no signal. We then say that an outlier has occurred. In the simplest case we can assume that the probability density of the values of the outliers is uniform over the search interval of a parameter, and then the rms error is given by
where Δ is the length of the search interval for a given parameter. The probability that an outlier occurs will be the higher the lower the signaltonoise ratio is. Let q be the probability that an outlier occurs. Then the total variance σ^{2} of the estimator of a parameter is the weighted sum of the two errors
where σ_{CR} is the rms errors calculated from the covariance matrix for a given parameter. One can show [49] that the probability q can be approximated by the following formula:
where p_{0} and p_{1} are the probability density functions of false alarm and detection given by Equations (46) and (47), respectively, and where Nc is the number of cells in the parameter space. Equation (84) is in good but not perfect agreement with the rms errors obtained from the Monte Carlo simulations (see [49]). There are clearly also other reasons for deviations from the CramèrRao bound. One important effect (see [72]) is that the functional F has many local subsidiary maxima close to the global one. Thus for a low signaltonoise the noise may promote the subsidiary maximum to a global one.
4.9 Upper limits
Detection of a signal is signified by a large value of the \({\mathcal F}\)statistic that is unlikely to arise from the noiseonly distribution. If instead the value of \({\mathcal F}\) is consistent with pure noise with high probability we can place an upper limit on the strength of the signal. One way of doing this is to take the loudest event obtained in the search and solve the equation
for signaltonoise ratio ρ_{UL}, where P_{D} is the detection probability given by Equation (49), \({\mathcal F_{\rm{L}}}\) is the value of the \({\mathcal F}\)statistic corresponding to the loudest event, and β is a chosen confidence [15, 1]. Then ρ_{UL} is the desired upper limit with confidence β.
When gravitationalwave data do not conform to a Gaussian probability density assumed in Equation (49), a more accurate upper limit can be obtained by injecting the signals into the detector’s data and thereby estimating the probability of detection P_{D} [3].
4.10 Network of detectors
Several gravitationalwave detectors can observe gravitational waves from the same source. For example a network of bar detectors can observe a gravitationalwave burst from the same supernova explosion, or a network of laser interferometers can detect the inspiral of the same compact binary system. The spaceborne LISA detector can be considered as a network of three detectors that can make three independent measurements of the same gravitationalwave signal. Simultaneous observations are also possible among different types of detectors, for example a search for supernova bursts can be performed simultaneously by bar and laser detectors [17].
We consider the general case of a network of detectors. Let h be the signal vector and let n be the noise vector of the network of detectors, i.e., the vector component h_{k} is the response of the gravitationalwave signal in the kth detector with noise n_{k}. Let us also assume that each n_{k} has zero mean. Assuming that the noise in all detectors is additive the data vector x can be written as
In addition if the noise is a stationary, Gaussian, and continuous random process the log likelihood function is given by
In Equation (87) the scalar product (· ·) is defined by
where \({{\rm{\tilde S}}}\) is the onesided cross spectral density matrix of the noises of the detector network which is defined by (here E denotes the expectation value)
The analysis is greatly simplified if the cross spectrum matrix S is diagonal. This means that the noises in various detectors are uncorrelated. This is the case when the detectors of the network are in widely separated locations like for example the two LIGO detectors. However, this assumption is not always satisfied. An important case is the LISA detector where the noises of the three independent responses are correlated. Nevertheless for the case of LISA one can find a set of three combinations for which the noises are uncorrelated [78, 80]. When the cross spectrum matrix is diagonal the optimum \({\mathcal F}\)statistic is just the sum of \({\mathcal F}\)statistics in each detector.
A derivation of the likelihood function for an arbitrary network of detectors can be found in [39], and applications of optimal filtering for the special cases of observations of coalescing binaries by networks of groundbased detectors are given in [48, 32, 75], for the case of stellar mass binaries observed by LISA spaceborne detector in [60], and for the case of spinning neutron stars observed by groundbased interferometers in [33]. The reduced Fisher matrix (Equation 43) for the case of a network of interferometers observing spinning neutron stars has been derived and studied in [79].
A least square fit solution for the estimation of the sky location of a source of gravitational waves by a network of detectors for the case of a broad band burst was obtained in [43].
There is also another important method for analyzing the data from a network of detectors — the search for coincidences of events among detectors. This analysis is particularly important when we search for supernova bursts the waveforms of which are not very well known. Such signals can be easily mimicked by nonGaussian behavior of the detector noise. The idea is to filter the data optimally in each of the detector and obtain candidate events. Then one compares parameters of candidate events, like for example times of arrivals of the bursts, among the detectors in the network. This method is widely used in the search for supernovae by networks of bar detectors [14].
4.11 Nonstationary, nonGaussian, and nonlinear data
Equations (34) and (35) provide maximum likelihood estimators only when the noise in which the signal is buried is Gaussian. There are general theorems in statistics indicating that the Gaussian noise is ubiquitous. One is the central limit theorem which states that the mean of any set of variates with any distribution having a finite mean and variance tends to the normal distribution. The other comes from the information theory and says that the probability distribution of a random variable with a given mean and variance which has the maximum entropy (minimum information) is the Gaussian distribution. Nevertheless, analysis of the data from gravitationalwave detectors shows that the noise in the detector may be nonGaussian (see, e.g., Figure 6 in [13]). The noise in the detector may also be a nonlinear and a nonstationary random process.
The maximum likelihood method does not require that the noise in the detector is Gaussian or stationary. However, in order to derive the optimum statistic and calculate the Fisher matrix we need to know the statistical properties of the data. The probability distribution of the data may be complicated, and the derivation of the optimum statistic, the calculation of the Fisher matrix components and the false alarm probabilities may be impractical. There is however one important result that we have already mentioned. The matchedfilter which is optimal for the Gaussian case is also a linear filter that gives maximum signaltonoise ratio no matter what is the distribution of the data. Monte Carlo simulations performed by Finn [39] for the case of a network of detectors indicate that the performance of matchedfiltering (i.e., the maximum likelihood method for Gaussian noise) is satisfactory for the case of nonGaussian and stationary noise.
Allen et al. [7, 8] derived an optimal (in the NeymanPearson sense, for weak signals) signal processing strategy when the detector noise is nonGaussian and exhibits tail terms. This strategy is robust, meaning that it is close to optimal for Gaussian noise but far less sensitive than conventional methods to the excess large events that form the tail of the distribution. This strategy is based on an locally optimal test ([53]) that amounts to comparing a first nonzero derivative Λ_{n},
of the likelihood ratio with respect to the amplitude of the signal with a threshold instead of the likelihood ratio itself.
In the remaining part of this section we review some statistical tests and methods to detect nonGaussianity, nonstationarity, and nonlinearity in the data. A classical test for a sequence of data to be Gaussian is the KolmogorovSmirnov test [27]. It calculates the maximum distance between the cumulative distribution of the data and that of a normal distribution, and assesses the significance of the distance. A similar test is the Lillifors test [27], but it adjusts for the fact that the parameters of the normal distribution are estimated from the data rather than specified in advance. Another test is the JarqueBera test [51] which determines whether sample skewness and kurtosis are unusually different from their Gaussian values.
Let x_{k} and u_{l} be two discrete in time random processes (−∞ < k, l < ∞) and let u_{l} be independent and identically distributed (i.i.d.). We call the process x_{k} linear if it can be represented by
where a_{l} are constant coefficients. If u_{l} is Gaussian (nonGaussian), we say that x_{l} is linear Gaussian (nonGaussian). In order to test for linearity and Gaussianity we examine the thirdorder cumulants of the data. The thirdorder cumulant C_{kl} of a zero mean stationary process is defined by
The bispectrum S_{2}(f_{1}, f_{2}) is the twodimensional Fourier transform of C_{kl}. The bicoherence is defined as
where S(f) is the spectral density of the process x_{k}. If the process is Gaussian then its bispectrum and consequently its bicoherence is zero. One can easily show that if the process is linear then its bicoherence is constant. Thus if the bispectrum is not zero, then the process is nonGaussian; if the bicoherence is not constant then the process is also nonlinear. Consequently we have the following hypothesis testing problems:

1.
H_{1}: The bispectrum of x_{k} is nonzero.

2.
H_{0}: The bispectrum of x_{k} is zero.
If Hypothesis 1 holds, we can test for linearity, that is, we have a second hypothesis testing problem:

3.
\({\rm{H}}_1^\prime\): The bicoherence of x_{k} is not constant.

4.
\({H''_1}\): The bicoherence of x_{k} is a constant.
If Hypothesis 4 holds, the process is linear.
Using the above tests we can detect nonGaussianity and, if the process is nonGaussian, nonlinearity of the process. The distribution of the test statistic B(f_{1}, f_{2}), Equation (93), can be calculated in terms of χ^{2} distributions. For more details see [45].
It is not difficult to examine nonstationarity of the data. One can divide the data into short segments and for each segment calculate the mean, standard deviation and estimate the spectrum. One can then investigate the variation of these quantities from one segment of the data to the other. This simple analysis can be useful in identifying and eliminating bad data. Another quantity to examine is the autocorrelation function of the data. For a stationary process the autocorrelation function should decay to zero. A test to detect certain nonstationarities used for analysis of econometric time series is the DickeyFuller test [25]. It models the data by an autoregressive process and it tests whether values of the parameters of the process deviate from those allowed by a stationary model. A robust test for detection nonstationarity in data from gravitationalwave detectors has been developed by Mohanty [69]. The test involves applying Student’s ttest to Fourier coefficients of segments of the data.
References
Abbott, B. et al. (LIGO Scientific Collaboration), “Analysis of LIGO data for gravitational waves from binary neutron stars”, Phys. Rev. D, 69, 122001, 1–16, (2004). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0308069.
Abbott, B. et al. (LIGO Scientific Collaboration), “First upper limits from LIGO on gravitational wave bursts”, Phys. Rev. D, 69, 102001, 1–21, (2004). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0312056.
Abbott, B. et al. (LIGO Scientific Collaboration), “Setting upper limits on the strength of periodic gravitational waves from PSR J1939+2134 using the first science data from the GEO 600 and LIGO detectors”, Phys. Rev. D, 69, 082004, 1–16, (2004). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0308050.
Adler, R.J., The Geometry of Random Fields, (Wiley, Chichester, U.K.; New York, U.S.A., 1981).
Allen, B., “The stochastic gravitywave background: Sources and detection”, in Marck, J.A., and Lasota, J.P., eds., Relativistic gravitation and gravitational radiation, Proceedings of the Les Houches School of Physics, held in Les Houches, Haute Savoie, 26 September–6 October, 1995, Cambridge Contemporary Astrophysics, (Cambridge University Press, Cambridge, U.K., 1997). Related online version (cited on 8 January 2005): http://arXiv.org/abs/astroph/9604033.
Allen, B., Blackburn, J.K., Brady, P.R., Creighton, J.D., Creighton, T., Droz, S., Gillespie, A.D., Hughes, S.A., Kawamura, S., Lyons, T.T., Mason, J.E., Owen, B.J., Raab, F.J., Regehr, M.W., Sathyaprakash, B.S., Savage Jr, R.L., Whitcomb, S., and Wiseman, A.G., “Observational Limit on Gravitational Waves from Binary Neutron Stars in the Galaxy”, Phys. Rev. Lett., 83, 1498–1501, (1999). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9903108.
Allen, B., Creighton, D.E., Flanagan, E.E., and Romano, J.D., “Robust statistics for deterministic and stochastic gravitational waves in nonGaussian noise I: Frequentist analyses”, Phys. Rev. D, 65, 122002, (2002). Related online version (cited on 10 June 2007): http://arxiv.org/abs/grqc/0105100. (document)
Allen, B., Creighton, D.E., Flanagan, É.É., and Romano, J.D., “Robust statistics for deterministic and stochastic gravitational waves in nonGaussian noise I: Bayesian analyses”, Phys. Rev. D, 67, 122002, (2003). Related online version (cited on 10 June 2007): http://arxiv.org/abs/grqc/0205015. (document)
Apostolatos, T.A., “Search templates for gravitational waves from precessing, inspiraling binaries”, Phys. Rev. D, 52, 605–620, (1995).
Armstrong, J.W., Estabrook, F.B., and Tinto, M., “TimeDelay Interferometry for SpaceBased Gravitational Wave Searches”, Astrophys. J., 527, 814–826, (1999).
Arnaud, N., Barsuglia, M., Bizouard, M., Brisson, V., Cavalier, F., Davier, M., Hello, P., Kreckelbergh, S., and Porter, E.K., “Coincidence and coherent data analysis methods for gravitational wave bursts in a network of interferometric detectors”, Phys. Rev. D, 68, 102001, 1–18, (2003). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0307100.
Arun, K.G., Iyer, B.R., Sathyaprakash, B.S., and Sundararajan, P.A., “Parameter estimation of inspiralling compact binaries using 3.5 postNewtonian gravitational wave phasing: The nonspinning case”, Phys. Rev. D, 71, 084008, 1–16, (2005). Related online version (cited on 23 July 2007): http://arXiv.org/abs/grqc/0411146v4. Erratum: Phys. Rev. D, 72, 069903, (2005). (document)
Astone, P., “Longterm operation of the Rome “Explorer” cryogenic gravitational wave detector”, Phys. Rev. D, 47, 362–375, (1993).
Astone, P., Babusci, D., Baggio, L., Bassan, M., Blair, D.G., Bonaldi, M., Bonifazi, P., Busby, D., Carelli, P., Cerdonio, M., Coccia, E., Conti, L., Cosmelli, C., D’Antonio, S., Fafone, V., Falferi, P., Fortini, P., Frasca, S., Giordano, G., Hamilton, W.O., Heng, I.S., Ivanov, E.N., Johnson, W.W., Marini, A., Mauceli, E., McHugh, M.P., Mezzena, R., Minenkov, Y., Modena, I., Modestino, G., Moleti, A., Ortolan, A., Pallottino, G.V., Pizzella, G., Prodi, G.A., Quintieri, L., Rocchi, A., Rocco, E., Ronga, F., Salemi, F., Santostasi, G., Taffarello, L., Terenzi, R., Tobar, M.E., Torrioli, G., Vedovato, G., Vinante, A., Visco, M., Vitale, S., and Zendri, J.P., “Methods and results of the IGEC search for burst gravitational waves in the years 1997–2000”, Phys. Rev. D, 68, 022001, 1–33, (2003).
Astone, P., Babusci, D., Bassan, M., Borkowski, K.M., Coccia, E., D’Antonio, S., Fafone, V., Giordano, G., Jaranowski, P., Królak, A., Marini, A., Minenkov, Y., Modena, I., Modestino, G., Moleti, A., Pallottino, G.V., Pietka, M., Pizzella, G., Quintieri, L., Rocchi, A., Ronga, F., Terenzi, R., and Visco, M., “Allsky upper limit for gravitational radiation from spinning neutron stars”, Class. Quantum Grav., 20, S665–S676, (2003). Paper from the 7th Gravitational Wave Data Analysis Workshop, Kyoto, Japan, 17–19 December 2002.
Astone, P., Borkowski, K.M., Jaranowski, P., and Krolak, A., “Data Analysis of gravitationalwave signals from spinning neutron stars. IV. An allsky search”, Phys. Rev. D, 65, 042003, 1–18, (2002).
Astone, P., Lobo, J.A., and Schutz, B.F., “Coincidence experiments between interferometric and resonant bar detectors of gravitational waves”, Class. Quantum Grav., 11, 2093–2112, (1994).
Balasubramanian, R., and Dhurandhar, S.V., “Estimation of parameters of gravitational waves from coalescing binaries”, Phys. Rev. D, 57, 3408–3422, (1998).
Balasubramanian, R., Sathyaprakash, B.S., and Dhurandhar, S.V., “Gravitational waves from coalescing binaries: detection strategies and Monte Carlo estimation of parameters”, Phys. Rev. D, 53, 3033–3055, (1996). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9508011. Erratum: Phys. Rev. D, 54, 1860, (1996).
Bayes, T., “An essay towards solving a problem in doctrine of chances”, Philos. Trans. R. Soc. London, 53, 293–315, (1763).
Berti, E., Buonanno, A., and Will, C.M., “Estimating spinning binary parameters and testing alternative theories of gravity with LISA”, Phys. Rev. D, 71, 084025, 1–24, (2005). Related online version (cited on 23 July 2007): http://arXiv.org/abs/grqc/0411129v2. (document)
Blanchet, L., “Gravitational Radiation from PostNewtonian Sources and Inspiralling Compact Binaries”, Living Rev. Relativity, 9, lrr20064, (2006). URL (cited on 10 June 2007): http://www.livingreviews.org/lrr20064.
Bonazzola, S., and Gourgoulhon, E., “Gravitational waves from pulsars: Emission by the magnetic field induced distortion”, Astron. Astrophys., 312, 675–690, (1996). Related online version (cited on 8 January 2005): http://arXiv.org/abs/astroph/9602107.
Brady, P.R., Creighton, T., Cutler, C., and Schutz, B.F., “Searching for periodic sources with LIGO”, Phys. Rev. D, 57, 2101–2116, (1998). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9702050.
Brooks, C., Introductory econometrics for finance, (Cambridge University Press, Cambridge, U.K.; New York, U.S.A., 2002).
Buonanno, A., Chen, Y., and Vallisneri, M., “Detection template families for gravitational waves from the final stages of binaryblackhole inspirals: Nonspinning case”, Phys. Rev. D, 67, 024016, 1–50, (2003). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0205122.
Conover, W.J., Practical Nonparametric Statistics, (Wiley, New York, U.S.A., 1998), 3rd edition.
Conway, J.H., and Sloane, N.J.A., Sphere Packings, Lattices and Groups, vol. 290 of Grundlehren der mathematischen Wissenschaften, (Springer, New York, U.S.A., 1999), 3rd edition.
Croce, R.P., Demma, T., Longo, M., Marano, S., Matta, V., Pierro, V., and Pinto, I.M., “Correlator bank detection of gravitational wave chirps—Falsealarm probability, template density, and thresholds: Behind and beyond the minimalmatch issue”, Phys. Rev. D, 70, 122001, 1–19, (2004). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0405023.
Cutler, C., “Angular resolution of the LISA gravitational wave detector”, Phys. Rev. D, 57, 7089–7102, (1998). Related online version (cited on 23 July 2007): http://arXiv.org/abs/grqc/9703068v1. (document)
Cutler, C., Apostolatos, T.A., Bildsten, L., Finn, L.S., Flanagan, É.É., Kennefick, D., Marković, D.M., Ori, A., Poisson, E., and Sussman, G.J., “The Last Three Minutes: Issues in GravitationalWave Measurements of Coalescing Compact Binaries”, Phys. Rev. Lett., 70, 2984–2987, (1993). Related online version (cited on 8 January 2005): http://arXiv.org/abs/astroph/9208005.
Cutler, C., and Flanagan, É.É., “Gravitational waves from merging compact binaries: How accurately can one extract the binary’s parameters from the inspiral waveform?”, Phys. Rev. D, 49, 2658–2697, (1994). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9402014.
Cutler, C., and Schutz, B.F., “The generalized Fstatistic: multiple detectors and multiple GW pulsars”, Phys. Rev. D, 72, 063006, (2005). Related online version (cited on 10 June 2007): http://arxiv.org/abs/grqc/0504011. (document)
Davis, M.H.A., “A Review of Statistical Theory of Signal Detection”, in Schutz, B.F., ed., Gravitational Wave Data Analysis, Proceedings of the NATO Advanced Research Workshop, held at Dyffryn House, St. Nichols, Cardiff, Wales, 6–9 July 1987, vol. 253 of NATO ASI Series C, 73–94, (Kluwer, Dordrecht, Netherlands; Boston, U.S.A., 1989).
Dhurandhar, S.V., and Sathyaprakash, B.S., “Choice of filters for the detection of gravitational waves from coalescing binaries. II. Detection in colored noise”, Phys. Rev. D, 49, 1707–1722, (1994).
Dhurandhar, S.V., and Schutz, B.F., “Filtering coalescing binary signals: Issues concerning narrow banding, thresholds, and optimal sampling”, Phys. Rev. D, 50, 2390–2405, (1994).
Estabrook, F.B., and Wahlquist, H.D., “Response of Doppler spacecraft tracking to gravitational radiation”, Gen. Relativ. Gravit., 6, 439–447, (1975).
Finn, L.S., “Detection, measurement, and gravitational radiation”, Phys. Rev. D, 46, 5236–5249, (1992). Related online version (cited on 23 July 2007): http://arXiv.org/abs/grqc/9209010v1. (document)
Finn, L.S., “Aperture synthesis for gravitationalwave data analysis: Deterministic sources”, Phys. Rev. D, 63, 102001, 1–18, (2001). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0010033.
Finn, L.S., and Chernoff, D.F., “Observing binary inspiral in gravitational radiation: One interferometer”, Phys. Rev. D, 47, 2198–2219, (1993). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9301003.
Fisz, M., Probability Theory and Mathematical Statistics, (Wiley, New York, U.S.A., 1963).
Giampieri, G., “On the antenna pattern of an orbiting interferometer”, Mon. Not. R. Astron. Soc., 289, 185–195, (1997).
Gürsel, Y., and Tinto, M., “Nearly optimal solution to the inverse problem for gravitationalwave bursts”, Phys. Rev. D, 40, 3884–3938, (1989).
Helstrom, C.W., Statistical Theory of Signal Detection, vol. 9 of International Series of Monographs in Electronics and Instrumentation, (Pergamon Press, Oxford, U.K.; New York, U.S.A., 1968), 2nd edition.
Hinich, M.J., “Testing for Gaussianity and linearity of a stationary time series”, J. Time Series Anal., 3, 169–176, (1982).
Hughes, S.A., “Untangling the merger history of massive black holes with LISA”, Mon. Not. R. Astron. Soc., 331, 805–816, (2002). Related online version (cited on 23 July 2007): http://arXiv.org/abs/astroph/0108483v1. (document)
Hughes, S.A., and Menou, K., “Golden binary gravitationalwave sources: Robust probes of strongfield gravity”, Astrophys. J., 623, 689–699, (2005). Related online version (cited on 23 July 2007): http://arXiv.org/abs/astroph/0410148v2. (document)
Jaranowski, P., and Królak, A., “Optimal solution to the inverse problem for the gravitational wave signal of a coalescing binary”, Phys. Rev. D, 49, 1723–1739, (1994).
Jaranowski, P., and Królak, A., “Data analysis of gravitationalwave signals from spinning neutron stars. III. Detection statistics and computational requirements”, Phys. Rev. D, 61, 062001, 1–32, (2000).
Jaranowski, P., Królak, A., and Schutz, B.F., “Data Analysis of gravitationalwave signals from spinning neutron stars: The signal and its detection”, Phys. Rev. D, 58, 063001, 1–24, (1998).
Judge, G.G., Hill, R.C., Griffiths, W.E., Lutkepohl, H., and Lee, T.C., The Theory and Practice of Econometrics, (Wiley, New York, U.S.A., 1980).
Kafka, P., “Optimal Detection of Signals through Linear Devices with Thermal Noise Sources and Application to the MunichFrascati WeberType Gravitational Wave Detectors”, in De Sabbata, V., and Weber, J., eds., Topics in Theoretical and Experimental Gravitation Physics, Proceedings of the International School of Cosmology and Gravitation held in Erice, Trapani, Sicily, March 13–25, 1975, vol. 27 of NATO ASI Series B, 161, (Plenum Press, New York, U.S.A., 1977).
Kassam, S.A., Signal Detection in NonGaussian Noise, (Springer, New York, U.S.A., 1988). (document)
Kendall, M., and Stuart, A., The Advanced Theory of Statistics. Vol. 2: Inference and Relationship, number 2, (C. Griffin, London, 1979).
Kokkotas, K.D., Królak, A., and Tsegas, G., “Statistical analysis of the estimators of the parameters of the gravitationalwave signal from a coalescing binary”, Class. Quantum Grav., 11, 1901–1918, (1994).
Kotelnikov, V.A., The theory of optimum noise immunity, (McGrawHill, New York, U.S.A., 1959).
Królak, A., Kokkotas, K.D., and Schäfer, G., “Estimation of the postNewtonian parameters in the gravitationalwave emission of a coalescing binary”, Phys. Rev. D, 52, 2089–2111, (1995). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9503013.
Królak, A., Lobo, J.A., and Meers, B.J., “Estimation of the parameters of the gravitationalwave signal of a coalescing binary system”, Phys. Rev. D, 48, 3451–3462, (1993).
Królak, A., and Schutz, B.F., “Coalescing binaries — probe to the Universe”, Gen. Relativ. Gravit., 19, 1163–1171, (1987).
Królak, A., Tinto, M., and Vallisneri, M., “Optimal filtering of the LISA data”, Phys. Rev. D, 70, 022003, 1–24, (2004). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0401108.
Lagarias, J.C., Reeds, J.A., Wright, M.H., and Wright, P.E., “Convergence properties of the NelderMead simplex method in low dimensions”, SIAM J. Optimiz., 9, 112–147, (1998).
Lehmann, E.L., Testing Statistical Hypothesis, (Wiley, New York, U.S.A., 1959).
Lehmann, E.L., Theory of Point Estimation, (Wiley, New York, U.S.A., 1983).
Liptser, R.S., and Shiryaev, A.N., Statistics of Random Processes, 2 vols., Applications of Mathematics, (Springer, New York, U.S.A., 1977).
LISA: Prephase A report, December 1998, MPQ 223, (MaxPlanckInstitut für Quantenoptik, Garching, Germany, 1998).
McDonough, R.N., and Whalen, A.D., Detection of signals in noise, (Academic Press, San Diego, U.S.A., 1995), 2nd edition
Meyer, C., Matrix Analysis and Applied Linear Algebra, (SIAM, Philadelphia, U.S.A., 2000).
Mohanty, S.D., “Hierarchical search strategy for the detection of gravitational waves from coalescing binaries: Extension to postNewtonian waveforms”, Phys. Rev. D, 57, 630–658, (1998). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9703081.
Mohanty, S.D., “A robust test for detecting nonstationarity in data from gravitational wave detectors”, Phys. Rev. D, 61, 122002, 1–12, (2000). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9910027.
Mohanty, S.D., and Dhurandhar, S.V., “Hierarchical search strategy for the detection of gravitational waves from coalescing binaries”, Phys. Rev. D, 54, 7108–7128, (1996).
Nicholson, D., Dickson, C.A., Watkins, W.J., Schutz, B.F., Shuttleworth, J., Jones, G.S., Robertson, D.I., MacKenzie, N.L., Strain, K.A., Meers, B.J., Newton, G.P., Ward, H., Cantley, C.A., Robertson, N.A., Hough, J., Danzmann, K., Niebauer, T.M., Rüdiger, A., Schilling, R., Schnupp, L., and Winkler, W., “Results of the first coincident observations by two laserinterferometric gravitational wave detectors”, Phys. Lett. A, 218, 175–180, (1996).
Nicholson, D., and Vecchio, A., “Bayesian bounds on parameter estimation accuracy for compact coalescing binary gravitational wave signals”, Phys. Rev. D, 57, 4588–4599, (1998). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9705064.
Niebauer, T.M., Rüdiger, A., Schilling, R., Schnupp, L., Winkler, W., and Danzmann, K., “Pulsar search using data compression with the Garching gravitational wave detector”, Phys. Rev. D, 47, 3106–3123, (1993).
Owen, B.J., “Search templates for gravitational waves from inspiraling binaries: Choice of template spacing”, Phys. Rev. D, 53, 6749–6761, (1996). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/9511032.
Pai, A., Dhurandhar, S.V., and Bose, S., “Dataanalysis strategy for detecting gravitationalwave signals from inspiraling compact binaries with a network of laserinterferometric detectors”, Phys. Rev. D, 64, 042004, 1–30, (2001). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0009078.
Poisson, E., and Will, C.M., “Gravitational waves from inspiralling compact binaries: Parameter estimation using secondpostNewtonian waveforms”, Phys. Rev. D, 52, 848–855, (1995). Related online version (cited on 23 July 2007): http://arXiv.org/abs/grqc/9502040v1. (document)
Poor, H.V., An Introduction to Signal Detection and Estimation, (Springer, New York, U.S.A., 1994), 2nd edition
Prince, T.A., Tinto, M., Larson, S.L., and Armstrong, J.W., “LISA optimal sensitivity”, Phys. Rev. D, 66, 122002, 1–7, (2002).
Prix, R., “The search for continuous gravitational waves: metric of the multidetector Fstatistic”, Phys. Rev. D, 75, 023004, (2007). Related online version (cited on 10 June 2007): http://arxiv.org/abs/grqc/0606088. (document)
Rajesh Nayak, K., Pai, A., Dhurandhar, S.V., and Vinet, J.Y., “Improving the sensitivity of LISA”, Class. Quantum Grav., 20, 1217–1231, (2003).
Rife, D.C., and Boorstyn, R.R., “Single tone parameter estimation from discretetime observations”, IEEE Trans. Inform. Theory, 20, 591–598, (1974).
Rowan, S., and Hough, J., “Gravitational Wave Detection by Interferometry (Ground and Space)”, Living Rev. Relativity, 3, lrr20003, (2000). URL (cited on 8 January 2005): http://www.livingreviews.org/lrr20003.
Rubbo, L.J., Cornish, N.J., and Poujade, O., “Forward modeling of spaceborne gravitational wave detectors”, Phys. Rev. D, 69, 082003, 1–14, (2004). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0311069.
Sathyaprakash, B.S., and Dhurandhar, S.V., “Choice of filters for the detection of gravitational waves from coalescing binaries”, Phys. Rev. D, 44, 3819–3834, (1991).
Schutz, B.F., “Determining the nature of the Hubble constant”, Nature, 323, 310–311, (1986).
Schutz, B.F., ed., Gravitational Wave Data Analysis, Proceedings of the NATO Advanced Research Workshop held at Dyffryn House, St. Nichols, Cardiff, Wales, 6–9 July 1987, vol. 253 of NATO ASI Series C, (Kluwer, Dordrecht, Netherlands; Boston, U.S.A., 1989).
Schutz, B.F., “Data processing, analysis and storage for interferometric antennas”, in Blair, D.G., ed., The Detection of Gravitational Waves, 406–452, (Cambridge University Press, Cambridge, U.K.; New York, U.S.A., 1991).
Sengupta, S.A., Dhurandhar, S.V., and Lazzarini, A., “Faster implementation of the hierarchical search algorithm for detection of gravitational waves from inspiraling compact binaries”, Phys. Rev. D, 67, 082004, 1–14, (2003).
Seto, N., “Effects of finite armlength of LISA on analysis of gravitational waves from massiveblackholes binaries”, Phys. Rev. D, 66, 122001, 1–7, (2002). Related online version (cited on 23 July 2007): http://arXiv.org/abs/grqc/0210028v1. (document)
Tagoshi, H., Kanda, N., Tanaka, T., Tatsumi, D., Telada, S., Ando, M., Arai, K., Araya, A., Asada, H., Barton, M.A., Fujimoto, M.K., Fukushima, M., Futamase, T., Heinzel, G., Horikoshi, G., Ishizuka, H., Kamikubota, N., Kawabe, K., Kawamura, S., Kawashima, N., Kojima, Y., Kozai, Y., Kuroda, K., Matsuda, N., Matsumura, S., Miki, S., Mio, N., Miyakawa, O., Miyama, S.M., Miyoki, S., Mizuno, E., Moriwaki, S., Musha, M., Nagano, S., Nakagawa, K., Nakamura, T., Nakao, K., Numata, K., Ogawa, Y., Ohashi, M., Ohishi, N., Okutomi, A., Oohara, K., Otsuka, S., Saito, Y., Sasaki, M., Sato, S., Sekiya, A., Shibata, M., Shirakata, K., Somiya, K., Suzuki, T., Takahashi, R., Takamori, A., Taniguchi, S., Tochikubo, K., Tomaru, T., Tsubono, K., Tsuda, N., Uchiyama, T., Ueda, A., Ueda, K., Waseda, K., Watanabe, Y., Yakura, H., Yamamoto, K., and Yamazaki, T. (The TAMA Collaboration), “First search for gravitational waves from inspiraling compact binaries using TAMA300 data”, Phys. Rev. D, 63, 062001, 1–5, (2001). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0012010.
Tanaka, T., and Tagoshi, H., “Use of new coordinates for the template space in hierarchical search for gravitational waves from inspiraling binaries”, Phys. Rev. D, 62, 082001, 1–8, (2000). Related online version (cited on 8 January 2005): http://arXiv.org/abs/grqc/0001090.
Thorne, K.S., “Gravitational radiation”, in Hawking, S.W., and Israel, W., eds., Three Hundred Years of Gravitation, 330–458, (Cambridge University Press, Cambridge, U.K.; New York, U.S.A., 1987).
Tinto, M., and Armstrong, J.W., “Cancellation of laser noise in an unequalarm interferometer detector of gravitational radiation”, Phys. Rev. D, 59, 102003, 1–11, (1999).
Table of Q Functions, RAND Research Memorandum, M339, (U.S. Air Force, Rand Corporation, Santa Monica, U.S.A., 1950).
Vallisneri, M., “Use and Abuse of the Fisher Information Matrix in the Assessment of GravitationalWave ParameterEstimation Prospects”, (2007). URL (cited on 23 July 2007): http://arXiv.org/abs/grqc/0703086v1. (document)
Van Trees, H.L., Detection, Estimation and Modulation Theory. Part 1: Detection, Estimation, and Linear Modulation Theory, number 1, (Wiley, New York, U.S.A., 1968).
Vecchio, A., “LISA observations of rapidly spinning massive black hole binary systems”, Phys. Rev. D, 70, 042001, 1–17, (2004). Related online version (cited on 23 July 2007): http://arXiv.org/abs/astroph/0304051v1. (document)
Wainstein, L.A., and Zubakov, V.D., Extraction of signals from noise, (PrenticeHall, Englewood Cliffs, U.S.A., 1962).
Weber, J., “Evidence for Discovery of Gravitational Radiation”, Phys. Rev. Lett., 22, 1320–1324, (1969).
Wong, E., Introduction to Random Processes, (Springer, New York, U.S.A., 1983).
Wong, E., and Hajek, B., Stochastic Processes in Engineering Systems, (Springer, New York, U.S.A., 1985).
Woodward, P.M., Probability and information theory with applications to radar, (Pergamon Press, London, U.K., 1953).
Zieliński, R., “Theory of parameter estimation”, in Królak, A., ed., Mathematics of Gravitation. Part II: Gravitational Wave Detection, Proceedings of the Workshop on Mathematical Aspects of Theories of Gravitation, held in Warsaw, February 29–March 30, 1996, vol. 41 (II) of Banach Center Publications, 209–220, (Institute of Mathematics, Polish Academy of Sciences, Warsaw, Poland, 1997).
Acknowledgments
One of us (AK) acknowledges support from the National Research Council under the Resident Research Associateship program at the Jet Propulsion Laboratory, California Institute of Technology and from MaxPlanckInstitut für Gravitationsphysik. This research was in part funded by the Polish KBN Grant No. 1 P03B 029 27.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Jaranowski, P., Królak, A. GravitationalWave Data Analysis. Formalism and Sample Applications: The Gaussian Case. Living Rev. Relativ. 8, 3 (2005). https://doi.org/10.12942/lrr20053
Accepted:
Published:
DOI: https://doi.org/10.12942/lrr20053
Keywords
 Gravitational Waves
 Gravitational Wave Data Analysis
 Fisher Matrix
 Gravitational Wave Signal
 Bicoherence
Article history
Latest
GravitationalWave Data Analysis. Formalism and Sample Applications: The Gaussian Case Published:
 09 March 2012
 Accepted:
 14 February 2012
DOI: https://doi.org/10.12942/lrr20124
Original
GravitationalWave Data Analysis. Formalism and Sample Applications: The Gaussian Case Published:
 21 March 2005
 Accepted:
 03 March 2005
DOI: https://doi.org/10.12942/lrr20053