An FETI-preconditioned conjuerate gradient method for large-scale stochastic finite element problems

In the spectral stochastic finite element method for analyzing an uncertain system. the uncertainty is represented by a set of random variables, and a quantity of Interest such as the system response is considered as a function of these random variables Consequently, the underlying Galerkin projection yields a block system of deterministic equations where the blocks are sparse but coupled. The solution of this algebraic system of equations becomes rapidly challenging when the size of the physical system and/or the level of uncertainty is increased This paper addresses this challenge by presenting a preconditioned conjugate gradient method for such block systems where the preconditioning step is based on the dual-primal finite element tearing and interconnecting method equipped with a Krylov subspace reusage technique for accelerating the iterative solution of systems with multiple and repeated right-hand sides. Preliminary performance results on a Linux Cluster suggest that the proposed Solution method is numerically scalable and demonstrate its potential for making the uncertainty quantification Of realistic systems tractable.

Elliptical summary randomisation for sensor-based human activity recognition

Many conventional statistical machine learning al- gorithms generalise poorly if distribution bias ex- ists in the datasets. For example, distribution bias arises in the context of domain generalisation, where knowledge acquired from multiple source domains need to be used in a previously unseen target domains. We propose Elliptical Summary Randomisation (ESRand), an efficient domain generalisation approach that comprises of a randomised kernel and elliptical data summarisation. ESRand learns a domain interdependent projection to a la- tent subspace that minimises the existing biases to the data while maintaining the functional relationship between domains. In the latent subspace, ellipsoidal summaries replace the samples to enhance the generalisation by further removing bias and noise in the data. Moreover, the summarisation enables large-scale data processing by significantly reducing the size of the data. Through comprehensive analysis, we show that our subspace-based approach outperforms state-of-the-art results on several activity recognition benchmark datasets, while keeping the computational complexity significantly low.

A model based factorization approach for dense 3D recovery from monocular video

Feature track matrix factorization based methods have been attractive solutions to the Structure-front-motion (Sfnl) problem. Group motion of the feature points is analyzed to get the 3D information. It is well known that the factorization formulations give rise to rank deficient system of equations. Even when enough constraints exist, the extracted models are sparse due the unavailability of pixel level tracks. Pixel level tracking of 3D surfaces is a difficult problem, particularly when the surface has very little texture as in a human face. Only sparsely located feature points can be tracked and tracking error arc inevitable along rotating lose texture surfaces. However, the 3D models of an object class lie in a subspace of the set of all possible 3D models. We propose a novel solution to the Structure-from-motion problem which utilizes the high-resolution 3D obtained from range scanner to compute a basis for this desired subspace. Adding subspace constraints during factorization also facilitates removal of tracking noise which causes distortions outside the subspace. We demonstrate the effectiveness of our formulation by extracting dense 3D structure of a human face and comparing it with a well known Structure-front-motion algorithm due to Brand.

Machine recognition of online handwritten Devanagari characters

In this paper, we describe a system for the automatic recognition of isolated handwritten Devanagari characters obtained by linearizing consonant conjuncts. Owing to the large number of characters and resulting demands on data acquisition, we use structural recognition techniques to reduce some characters to others. The residual characters are then classified using the subspace method. Finally the results of structural recognition and feature-based matching are mapped to give final output. The proposed system Ifs evaluated for the writer dependent scenario.

Narrowband Detection of Acoustic Source in Shallow Ocean using Vector Sensor Array

The problem of narrowband CFAR (constant false alarm rate) detection of an acoustic source at an unknown location in a range-independent shallow ocean is considered. If a target is present, the received signal vector at an array of N sensors belongs to an M-dimensional subspace if N exceeds the number of propagating modes M in the ocean. A subspace detection method which utilises the knowledge of the signal subspace to enhance the detector performance is presented in thisMpaper. It is shown that, for a given number of sensors N, the performance of a detector using a vector sensor array is significantly better than that using a scalar sensor array. If a target is detected, the detector using a vector sensor array also provides a concurrent coarse estimate of the bearing of the target.

Quenching across quantum critical points: Role of topological patterns

We introduce a one-dimensional version of the Kitaev model consisting of spins on a two-legged ladder and characterized by Z(2) invariants on the plaquettes of the ladder. We map the model to a fermionic system and identify the topological sectors associated with different Z2 patterns in terms of fermion occupation numbers. Within these different sectors, we investigate the effect of a linear quench across a quantum critical point. We study the dominant behavior of the system by employing a Landau-Zener-type analysis of the effective Hamiltonian in the low-energy subspace for which the effective quenching can sometimes be non-linear. We show that the quenching leads to a residual energy which scales as a power of the quenching rate, and that the power depends on the topological sectors and their symmetry properties in a non-trivial way. This behavior is consistent with the general theory of quantum quenching, but with the correlation length exponent nu being different in different sectors. Copyright (C) EPLA, 2010

Decorrelation of broad band signals by toeplitz-block-toeplitz structuring

We propose F-norm of the cross-correlation part of the array covariance matrix as a measure of correlation between the impinging signals and study the performance of different decorrelation methods in the broadband case using this measure. We first show that dimensionality of the composite signal subspace, defined as the number of significant eigenvectors of the source sample covariance matrix, collapses in the presence of multipath and the spatial smoothing recovers this dimensionality. Using an upper bound on the proposed measure, we then study the decorrelation of the broadband signals with spatial smoothing and the effect of spacing and directions of the sources on the rate of decorrelation with progressive smoothing. Next, we introduce a weighted smoothing method based on Toeplitz-block-Toeplitz (TBT) structuring of the data covariance matrix which decorrelates the signals much faster than the spatial smoothing. Computer simulations are included to demonstrate the performance of the two methods.

Finite data performance of MUSIC and minimum norm methods

In the direction of arrival (DOA) estimation problem, we encounter both finite data and insufficient knowledge of array characterization. It is therefore important to study how subspace-based methods perform in such conditions. We analyze the finite data performance of the multiple signal classification (MUSIC) and minimum norm (min. norm) methods in the presence of sensor gain and phase errors, and derive expressions for the mean square error (MSE) in the DOA estimates. These expressions are first derived assuming an arbitrary array and then simplified for the special case of an uniform linear array with isotropic sensors. When they are further simplified for the case of finite data only and sensor errors only, they reduce to the recent results given in [9-12]. Computer simulations are used to verify the closeness between the predicted and simulated values of the MSE.

Cascades of Selective Pulses on Connected Single-Quantum Transitions Leading to the Selective Excitation of Multiple-Quantum Coherences

A symmetric cascade of selective pulses applied on connected transitions leads to the excitation of a selected multiple-quantum coherence by a well-defined angle. This cascade selectively operates on the subspace of the multiple-quantum coherence and acts as a generator of rotation selectively on the multiple-quantum subspace. Single-transition operator algebra has been used to explain these experiments. Experiments have been performed on two- and three-spin systems. It is shown that such experiments can be utilized to measure the relaxation times of selected multiple-quantum coherences or of a specifically prepared initial longitudinal state of the spin system.

Some Algorithms for Eigensubspace Estimation

An important tool in signal processing is the use of eigenvalue and singular value decompositions for extracting information from time-series/sensor array data. These tools are used in the so-called subspace methods that underlie solutions to the harmonic retrieval problem in time series and the directions-of-arrival (DOA) estimation problem in array processing. The subspace methods require the knowledge of eigenvectors of the underlying covariance matrix to estimate the parameters of interest. Eigenstructure estimation in signal processing has two important classes: (i) estimating the eigenstructure of the given covariance matrix and (ii) updating the eigenstructure estimates given the current estimate and new data. In this paper, we survey some algorithms for both these classes useful for harmonic retrieval and DOA estimation problems. We begin by surveying key results in the literature and then describe, in some detail, energy function minimization approaches that underlie a class of feedback neural networks. Our approaches estimate some or all of the eigenvectors corresponding to the repeated minimum eigenvalue and also multiple orthogonal eigenvectors corresponding to the ordered eigenvalues of the covariance matrix. Our presentation includes some supporting analysis and simulation results. We may point out here that eigensubspace estimation is a vast area and all aspects of this cannot be fully covered in a single paper. (C) 1995 Academic Press, Inc.

Source localization in a partially known shallow?water channel

The source localization in shallow water is beset with problems arising from the presence of a large number of correlated multipaths. Nevertheless, given a complete knowledge of the water channel it is definitely possible to localize a source. A complete knowledge of the channel, however, is rarely available under most practical conditions. A new approach is proposed wherein the bottom reflection coefficients are not required; hence the bottom conditions need not be known. Further, because of the use of signal subspace for localization, the proposed approach is robust against the background noise (-20 dB) and channel depth uncertainty (10 lambda). All these nice features of the proposed approach are possible only when the array size is large (>40 sensors). (C) 1995 Acoustical Society of America.

Direction of arrival estimation in the presence of distributed noise sources: Cumulant based approach

In the past few years there have been attempts to develop subspace methods for DoA (direction of arrival) estimation using a fourth?order cumulant which is known to de?emphasize Gaussian background noise. To gauge the relative performance of the cumulant MUSIC (MUltiple SIgnal Classification) (c?MUSIC) and the standard MUSIC, based on the covariance function, an extensive numerical study has been carried out, where a narrow?band signal source has been considered and Gaussian noise sources, which produce a spatially correlated background noise, have been distributed. These simulations indicate that, even though the cumulant approach is capable of de?emphasizing the Gaussian noise, both bias and variance of the DoA estimates are higher than those for MUSIC. To achieve comparable results the cumulant approach requires much larger data, three to ten times that for MUSIC, depending upon the number of sources and how close they are. This is attributed to the fact that in the estimation of the cumulant, an average of a product of four random variables is needed to make an evaluation. Therefore, compared to those in the evaluation of the covariance function, there are more cross terms which do not go to zero unless the data length is very large. It is felt that these cross terms contribute to the large bias and variance observed in c?MUSIC. However, the ability to de?emphasize Gaussian noise, white or colored, is of great significance since the standard MUSIC fails when there is colored background noise. Through simulation it is shown that c?MUSIC does yield good results, but only at the cost of more data.

Proof of a conjecture of Frankl and Furedi

We give a simple linear algebraic proof of the following conjecture of Frankl and Furedi [7, 9, 13]. (Frankl-Furedi Conjecture) if F is a hypergraph on X = {1, 2, 3,..., n} such that 1 less than or equal to /E boolean AND F/ less than or equal to k For All E, F is an element of F, E not equal F, then /F/ less than or equal to (i=0)Sigma(k) ((i) (n-1)). We generalise a method of Palisse and our proof-technique can be viewed as a variant of the technique used by Tverberg to prove a result of Graham and Pollak [10, 11, 14]. Our proof-technique is easily described. First, we derive an identity satisfied by a hypergraph F using its intersection properties. From this identity, we obtain a set of homogeneous linear equations. We then show that this defines the zero subspace of R-/F/. Finally, the desired bound on /F/ is obtained from the bound on the number of linearly independent equations. This proof-technique can also be used to prove a more general theorem (Theorem 2). We conclude by indicating how this technique can be generalised to uniform hypergraphs by proving the uniform Ray-Chaudhuri-Wilson theorem. (C) 1997 Academic Press.

Model exact low-lying states and spin dynamics in ferric wheels: Fe-6 to Fe-12

Using an efficient numerical scheme that exploits spatial symmetries and spin parity, we have obtained the exact low-lying eigenstates of exchange Hamiltonians for ferric wheels up to Fe-12. The largest calculation involves the Fe-12 ring which spans a Hilbert space dimension of about 145x10(6) for the M-S=0 subspace. Our calculated gaps from the singlet ground state to the excited triplet state agree well with the experimentally measured values. Study of the static structure factor shows that the ground state is spontaneously dimerized for ferric wheels. The spin states of ferric wheels can be viewed as quantized states of a rigid rotor with the gap between the ground and first excited states defining the inverse of the moment of inertia. We have studied the quantum dynamics of Fe-10 as a representative of ferric wheels. We use the low-lying states of Fe-10 to solve exactly the time-dependent Schrodinger equation and find the magnetization of the molecule in the presence of an alternating magnetic field at zero temperature. We observe a nontrivial oscillation of the magnetization which is dependent on the amplitude of the ac field. We have also studied the torque response of Fe-12 as a function of a magnetic field, which clearly shows spin-state crossover.

Two Dimensional Principal Component Analysis for Online Tamil Character Recognition

This paper presents a new application of two dimensional Principal Component Analysis (2DPCA) to the problem of online character recognition in Tamil Script. A novel set of features employing polynomial fits and quartiles in combination with conventional features are derived for each sample point of the Tamil character obtained after smoothing and resampling. These are stacked to form a matrix, using which a covariance matrix is constructed. A subset of the eigenvectors of the covariance matrix is employed to get the features in the reduced sub space. Each character is modeled as a separate subspace and a modified form of the Mahalanobis distance is derived to classify a given test character. Results indicate that the recognition accuracy using the 2DPCA scheme shows an approximate 3% improvement over the conventional PCA technique.