Minimum variance

Adaptive spectral estimation techniques are known to provide good spectral resolution and contrast even when the observation window(OW) is very short. In this paper two adaptive techniques are tested and compared to the averaged periodogram (Welch) for blood velocity estimation. The Blood Power spectral Capon (BPC) method is based on a standard minimum variance technique adapted to account for both averaging over slowtime and depth. The Blood Amplitude and Phase Estimation technique (BAPES) is based on finding a set of matched filters (one for each velocity component of interest) and filtering the blood process over slow-time and averaging over depth to find the power spectral density estimate. In this paper, the two adaptive methods are explained, and performance is assessed in controlled steady flow experiments and in-vivo measurements. The three methods were tested on a circulating flow rig with a blood mimicking fluid flowing in the tube. The scanning section is submerged in water to allow ultrasound data acquisition. Data was recorded using a BK8804 linear array transducer and the RASMUS ultrasound scanner. The controlled experiments showed that the OW could be significantly reduced when applying the adaptive methods without compromising spectral resolution or contrast. The in-vivo data was acquired using a BK8812 transducer. OWs of 128, 64, 32 and 16 slow-time samples were tested. Spectrograms with duration of 2 seconds were generated. Welch's method required 128 samples to give a reasonable spectrogram, whereas the BPC only required 32 samples before the SNR became a limiting factor. The BAPES managed to display the spectrogram with sufficient quality at 16 slow-time samples.

Due to the dramatic increase of data volumes in different applications, it is becoming infeasible to keep these data in one centralized machine. It is becoming more and more natural to deal with distributed databases and networks. That is why distributed data mining techniques have been introduced. One of the most important data mining problems is data clustering. While many clustering algorithms exist for centralized databases, there is a lack of efficient algorithms for distributed databases. In this paper, an efficient algorithm is proposed for clustering distributed databases. The proposed methodology employs an iterative optimization technique to achieve better clustering objective. The experimental results reported in this paper show the superiority of the proposed technique over a recently proposed algorithm based on a distributed version of the well known K-Means algorithm (Datta et al. 2009) [1].

We measure the large-scale real-space power spectrum Pk using luminous red galaxies (LRGs) in the Sloan Digital Sky Survey (SDSS) and use this measurement to sharpen constraints on cosmological parameters from the Wilkinson Microwave Anisotropy Probe (WMAP). We employ a matrix-based power spectrum estimation method using Pseudo-Karhunen-Loe `ve eigenmodes, producing uncorrelated minimum-variance measurements in 20 k-bands of both the clustering power and its anisotropy due to

We measure the large-scale real-space power spectrum Pk using luminous red galaxies (LRGs) in the Sloan Digital Sky Survey (SDSS) and use this measurement to sharpen constraints on cosmological parameters from the Wilkinson Microwave Anisotropy Probe (WMAP). We employ a matrix-based power spectrum estimation method using Pseudo-Karhunen-Loe `ve eigenmodes, producing uncorrelated minimum-variance measurements in 20 k-bands of both the clustering power and its anisotropy due to

Constructing portfolios based on second-order stochastic dominance (SSD) is theoretically attractive since all risk-averse investors would prefer a dominating portfolio. However, choosing among SSD efficient portfolios is a challenge without an obvious ranking metric. We explore particular choices based on plus , comparing their performance to other SSD-related strategies and to standard portfolio choice approaches. These SSD-related choices outperform portfolios chosen based on their Sharpe ratio, information ratio, or using equal weights. Portfolios based on minimum variance that also match the benchmark's mean return perform on a par with the SSD-related choices.

Discrete minimum variance controllers subject to a constraint on the variance of the manipulated vatiable, that is, constrained minimum variance (CMV) controllers, are shown to exhibit good performance and robustness properties. These C M V controllers are internal model controllers (IMC) for specified disturbance structures, and the Wiener-Hopf design procedure used to obtain them provides a convenient means of obtaining stable model inverses and filters. The higher order filters usually provided by this design approach give both good disturbance prediction and rapid gain roll-off at high frequencies.

Co-channel interference limits the capacity of cellular systems, reduces the throughput of wireless local area networks, and is the major hurdle in deployment of high altitude communication platforms. It is also a problem for systems operating in unlicensed bands such as the 2.4 GHz ISM band and for narrowband systems that have been overlaid with spread spectrum systems. In this work we have developed model-based techniques for the demodulation of cochannel MSK signals. It is shown that MSK signals can be written in the linear model form, hence a minimum variance unbiased (MVU) estimator exists that satisfies the Cramer-Rao lower bound (CRLB) with equality. This framework allows us to derive the best estimators for a single-user and a two-user case. These concepts can also be extended to wideband signals and it is shown that the MVU estimator for Direct Sequence Spread Spectrum signals is in fact a decorrelator-based multiuser detector. However, this simple linear representation does not always exist for continuous phase modulations. Furthermore, these linear estimators require perfect channel state information and phase synchronization at the receiver, which is not always implemented in wireless communication systems. To overcome these shortcomings of the linear estimation techniques, we employed an autoregressive modeling approach. It is well known that the AR model can accurately represent peaks in the spectrum and therefore can be used as a general FM demodulator. It does not require knowledge of the exact signal model or phase synchronization at the receiver. Since it is a non-coherent reception technique, its performance is compared to that of the limiter discriminator. Simulation results have shown that model-based demodulators can give significant gains for certain phase and frequency offsets between the desired signal and an interferer. phase difference between the signal of interest and an interferer 4.7 Performance of model-based demodulator with LMS update 4.8 Performance of model-based demodulator using angle of filter coefficients 4.9 Comparison of various model-based demodulation schemes 4.10 Performance of limiter discriminator for various carrier frequency separations 4.11 Performance of model-based demodulator for various carrier frequency separations 4.12 Comparison of limiter discriminator and model-based demodulator for carrier frequency separation of 1/2T 4.13 Comparison of model-based demodulator and limiter discriminator in Rayleigh fading environment (FdT = 0.002, Eb/No = 20dB)

was produced in a fed-batch bioprocess with L-leucine auxotrophic Corynebacterium glutamicum strains developed by genetic engineering. An efficient supply with nutrients was achieved by applying closed-loop control of glucose as the main carbon source, with a model-based, parameter-adaptive control strategy. This control strategy is based on an extended, semi-continuous Kalman filter for process identification and a minimum variance controller. The lab scale fed-batch process with C. glutamicum SMl and C. glutamicum DR17 pECM3::iluA38 was characterized with respect to biomass, product and by-product accumulation. A differential analysis of growth, specific productivities, and selectivities was performed to characterize the carbon flow over process time. Characterization of L-isoleucine transport steps across the cell membrane resulted in a balance of L-isoleucine transport over process time. Up to an extracellular L-isoleucine concentration of 140 mM the cytosolic L-isoleucine, provided by the biosynthesis, was quantitatively excreted into the medium via the export carrier system. Optimized feeding profiles for, L-leucine and phosphate in correlation with the on-line estimated glucose consumption were achieved up to the pilot scale (300-l stirred tank reactor). The maximum L-isoleucine concentration was 150 mM (21 g 1-l) with a space-time yield of 4.3 mmol 1 -' h -'. With a 98% closed carbon balance the selectivity for isoleucine was 14%, for biomass 13%, and for CO, 68%.

A method for the identification and quantification of cross-reacting analytes using competitive immunochemical assays is described. The method uses information both of antibodies with and without significant inhibition caused by a sample. Maximum concentrations of all possible analytes were estimated for all antibodies not showing a significant inhibition. These maximum concentrations could be used to exclude certain analytes from the further identification process. A minimum variance method was used for the identification of analytes from the data given by antibodies showing significant inhibition. All samples were measured in a parallel affinity sensor array (PASA). The PASA system allows the parallel performance of numerous individual immunochemical assays. Triazines were used as a model substance class. Samples containing either atrazine, terbuthylazine, simazine or deethylatrazine at different concentration levels were generated and analyzed in the PASA system. 11 out of 13 samples were correctly identified, 2 samples could not be identified without doubt, no wrong identification was observed. Samples of atrazine, terbuthylazine and simazine at a concentration level of 0.1 µg/L, the EU maximum admissible concentration for individual pesticides, and of deethylatrazine at 0.3 µg/L could be quantified.

The development of the Keck All sky Precision Adaptive optics (KAPA) project was initiated in September 2018 to upgrade the Keck I adaptive optics (AO) system to enable laser tomography adaptive optics (LTAO) with a four laser guide star (LGS) asterism. The project includes the replacement of the existing LMCT laser with a Toptica laser, the implementation of a new real-time controller (RTC) and wavefront sensor optics and camera, and a new daytime calibration and test platform to provide the required infrastructure for laser tomography. The work presented here describes the new daytime calibration infrastructure to test the performance for the KAPA tomographic algorithms. This paper outlines the hardware infrastructure for daytime calibration and performance assessment of tomographic algorithms. This includes the implementation of an asterism simulator having fiber-coupled light sources simulating four Laser Guide Stars (LGS) and two Natural Guide Stars (NGS) at the AO bench focus, as well as the upgrade of the existing TelSim on the AO bench to simulate focal anisoplanatism and wind driven atmospheric turbulence. A phase screen, that can be adjusted in effective altitude, is used to simulate wind speeds up to 10 m/s for a duration of upto 3 s.

The Best Linear Unbiased Estimator (BLUE) has widely been used in atmospheric and oceanic data assimilation (DA). However, when the errors from data (observations and background forecasts) have non-Gaussian probability density functions (pdfs), the BLUE differs from the absolute Minimum Variance Unbiased Estimator (MVUE), minimizing the mean square a posteriori error. The non-Gaussianity of errors can be due to the inherent statistical skewness and positiveness of some physical observables (e.g. moisture, chemical species) or because of the nonlinearity of the data assimilation models and observation operators acting on Gaussian errors. Non-Gaussianity of assimilated data errors can be justified from a priori hypotheses or inferred from statistical diagnostics of innovations (observation minus background). Following this rationale, we compute measures of innovation non-Gaussianity, namely its skewness and kurtosis, relating it to: a) the non-Gaussianity of the individual error themselves, b) the correlation between nonlinear functions of errors, and c) the heteroscedasticity of errors within diagnostic samples. Those relationships impose bounds for skewness and kurtosis of errors which are critically dependent on the error variances, thus leading to a necessary tuning of error variances in order to accomplish consistency with innovations. We evaluate the sub-optimality of the BLUE as compared to the MVUE, in terms of excess of error variance, under the presence of non-Gaussian errors. The error pdfs are obtained by the maximum entropy method constrained by error moments up to fourth order, from which the Bayesian probability density function and the MVUE are computed. The impact is higher for skewed extreme innovations and grows in average with the skewness of data errors, especially if those skewnesses have the same sign. Application has been performed to the quality-accepted ECMWF innovations of brightness temperatures of a set of High Resolution Infrared Sounder (HIRS) channels. In this context, the MVUE has led in some extreme cases to a potential reduction of 20-60% error variance as compared to the BLUE.

In this paper, we propose brain-like (BL) sensorimotor control system that is assumed to be composed of its unconscious and conscious part. The conscious part of BL sensorimotor control system is emerged when Prefrontal Cortex (PFC) acts with respect to spatiotemporal selective attention mechanism in particular. In our framework, the selective attention can be induced by Baye's rule comprising of PFC-based neural network where both unsupervised and supervised learning schemes are employed. A preliminary experiment is manually carried out to analyse human control mechanism for the stability of RC-helicopter. As a result, we suggest the relationship between the selective attention and the minimum variance theory. BL sensorimotor control system is favorably considered to build subject to suffice the minimum variance theory that is the key for computing spatiotemporal selective attention mechanism, computed by the PFC-based neural network. In this paper, the network is also examined with respect to control the RC-helicopter and the result shows the soundness of BL sensorimotor control system.

The paper deals with a class of quantization problems and algorithms based on partitions of R d which are defined by maximal support planes (MSP) of a convex function. This kind of problems generalizes the well-known minimum variance partition problem of statistical cluster analysis. The extension is a multivariate version of the topic considered by Bock, [4]. It is basically different from those kinds of quantization problems which have been considered by Pollard, [15], and Pärna, [14], and which are called principal point problems by Flury, [7]. As a side result it is shown that some competitive learning problems considered by Kohonen, [10], belong to the class considered in this paper. The paper contains theoretical results like existence of optima, consistency of approximate optima and characterization of local optima as fixpoints of a fixpoint algorithm. A fix point algorithm is proposed and its termination after finite time is proved for empirical distributions. Modifying the underlying convex function gives room for a large variety of new classification procedures, e.g., of robust quantifiers. Several examples are discussed by way of illustration. Numerical studies of the performance of such procedures are the subject of the Doctoral Thesis by Steiner,

The paper deals with a class of quantization problems and algorithms based on partitions of R d which are defined by maximal support planes (MSP) of a convex function. This kind of problems generalizes the well-known minimum variance partition problem of statistical cluster analysis. The extension is a multivariate version of the topic considered by Bock, [4]. It is basically different from those kinds of quantization problems which have been considered by Pollard, [15], and Pärna, [14], and which are called principal point problems by Flury, [7]. As a side result it is shown that some competitive learning problems considered by Kohonen, [10], belong to the class considered in this paper. The paper contains theoretical results like existence of optima, consistency of approximate optima and characterization of local optima as fixpoints of a fixpoint algorithm. A fix point algorithm is proposed and its termination after finite time is proved for empirical distributions. Modifying the underlying convex function gives room for a large variety of new classification procedures, e.g., of robust quantifiers. Several examples are discussed by way of illustration. Numerical studies of the performance of such procedures are the subject of the Doctoral Thesis by Steiner,

Linear minimum variance unbiased state estimation is considered for systems with uncertain parameters in their state space models and sensor failures. The existing results are generalized to the case where each sensor may fail at any sample time independently of the others. For robust performance, stochastic parameter perturbations are included in the system matrix. Also, stochastic perturbations are allowed in the estimator gain to guarantee resilient operation. An illustrative example is included to demonstrate performance improvement over the Kalman filter which does not include sensor failures in its measurement model.

This report gives a survey of the results achieved at the Division of Automatic Control during the year 1972-1973. ' Most of the research has been centered around the Process Control Project. The project contains system identification, computational control, real time computing, adaptive control, system theory, non-technical systems and applications. During the year 19 published papers and 44 technical reports have been written. Two PhD theses, one licentiat thesis and 18 MS-theses have also been completed.

Cluster data are used to study the statistics of a particular type of foreshock wave: quasi‐monochromatic ultra‐low‐frequency (ULF) waves with characteristic periods of 30 s in the spacecraft frame. On the basis of a large set of foreshock observations made in 2001 with spacecraft separations ∼600 km, the intrinsic properties of the waves are found using a cross‐correlation timing analysis. This method, which allows the solar wind rest frame properties of the waves to be calculated with fewer assumptions when compared to previous dual spacecraft analysis, is described in detail. The performance of minimum variance analysis (MVA) as a wave analysis tool is investigated experimentally using this multispacecraft analysis. MVA estimates of propagation direction are shown to diverge from the multispacecraft estimate in the limit of linear wave polarization. Theoretical estimates of MVA error are also compared to the observations. Previously established wave properties derived from ISEE a...

Cluster and ACE magnetic field measurements from the 13 February 2001 are presented. On this day the four Cluster spacecraft made two crossings of the heliospheric current sheet (HCS) while in the solar wind and observed a large‐scale rotation of the magnetic field in the magnetosheath. These features were also observed by ACE upstream of the Earth at the L1 point. Minimum variance analysis (MVA) is used to show the similarity of the HCS orientation at ACE and Cluster, and Cluster multispacecraft analysis of the timing, orientation, and changes across the reversal is used to show that these results are consistent with crossings of the HCS. The structure of the currents in the HCS is also investigated using the Cluster curlometer technique. The field rotation observed by Cluster in the magnetosheath and ACE in the solar wind is identified as a magnetic flux rope, and MVA shows that although the structure and ordering of the flux rope are unchanged by the bow shock, the orientation of...

We present Cluster spacecraft observations of large-amplitude (𝛿|B|∕|B|∼1) fast-mode magnetosonic waves in the heliospheric current sheet (HCS). The crossings of the HCS and the associated heliospheric plasma sheet (HPS) are encountered by Cluster in the near-Earth solar wind and by ACE upstream of Earth. Multiple current layers are detected in correspondence with small-scale discontinuities in the regime of open magnetic field lines within the HCS. Fast magnetosonic waves are observed at one current layer, accompanying the phase-steeped edge of a large-amplitude transverse Alfvén wave. The observed fast-mode waves are in the frequency range 0.01 Hz-0.2 Hz, characterized by a strong correlation between variations of plasma density and magnetic field strength. Analysis of ratio 𝛿E∕𝛿B indicate that the fast-mode wave packet consists of an antisunward propagating component and a larger sunward propagating component in the rest frame of the solar wind. The fast-mode waves are not observed by ACE in the upstream solar wind. The generation of the fast-mode waves may relate to the development of the phase-steeped Alfvén wave and has profound effects on the evolution of the solar wind plasma.

In this paper, we use Cluster data from one magnetopause event on 5 July 2001 to compare predictions from various methods for determination of the velocity, orientation, and thickness of the magnetopause current layer. We employ established as well as new multi-spacecraft techniques, in which time differences between the crossings by the four spacecraft, along with the duration of each crossing, are used to calculate magnetopause speed, normal vector, and width. The timing is based on data from either the Cluster Magnetic Field Experiment (FGM) or the Electric Field Experiment (EFW) instruments. The multi-spacecraft results are compared with those derived from various singlespacecraft techniques, including minimum-variance analysis of the magnetic field and deHoffmann-Teller, as well as Minimum-Faraday-Residue analysis of plasma velocities and magnetic fields measured during the crossings. In order to improve the overall consistency between multi-and single-spacecraft results, we have also explored the use of hybrid techniques, in which timing information from the four spacecraft is combined with certain limited results from single-spacecraft methods, the remaining results being left for consistency checks. The results show good agreement between magnetopause orientations derived from appropriately chosen single-spacecraft techniques and those obtained from multi-spacecraft timing. The agreement between magnetopause speeds derived from single-and multi-spacecraft methods is quantitatively somewhat less good but it is evident that the speed can change substantially from one crossing to the next within an event. The magnetopause thickness varied

Two orbits were selected in January-February 2006 when the separation between the Cluster spacecraft was large and mirror type magnetic field fluctuations were observed by all spacecraft in different regions of the terrestrial magnetosheath. Minimum variance analysis was applied to find the mirror type fluctuations, and the amplitude of the fluctuations was determined individually. Mirror mode structures are moving along the streamlines frozen in the plasma. A model was developed for the calculation of plasma flowtime from the bow shock to the observation point. The growth rate of the field strength perturbations was estimated by comparing the amplitudes of fluctuations observed simultaneously at distant locations ($10,000 km) based on the assumption that dB $ exp(ct). The obtained growth rate values were about an order of magnitude smaller than those provided by linear models and they decreased in the inner regions of the magnetosheath, indicating some saturation in the growth of the waves when proceeding towards the magnetopause. The results of these two case studies suggest that mirror type fluctuations originate from the compression region downstream of the quasi-perpendicular bow shock, and the growth of the fluctuations cannot be described by linear approximations.

In contrast with existing literature that focuses on conditional Value-At-Risk (CVaR) as a portfolio risk measure, we examine here the properties of portfolios built to minimize CVaR. We look into the stability and performance potential of CVaR-optimal portfolios and compare our results with Minimum Variance portfolios. Finally, we run realistic simulations of risk-minimising strategies using CVaR as risk measure and compare their performances to otherwise identical variance minimising strategies.

The importance of covariance modelling has long been recognised in the field of portfolio management and large dimensional multivariate problems are increasingly becoming the focus of research. This paper provides a straightforward and commonsense approach toward investigating whether simpler moving average based correlation forecasting methods have equal predictive accuracy as their more complex multivariate GARCH counterparts for large dimensional problems. We find simpler forecasting techniques do provide equal (and often superior) predictive accuracy in a minimum variance sense. A portfolio allocation problem is used to compare forecasting methods. The global minimum variance portfolio and Model Confidence Set (Hansen, Lunde, and Nason (2003)) are used to compare methods, whilst portfolio weight stability and computational time are also considered.

In the framework of continuous-time finance theory, this paper derives the optimal consumption and portfolio rules for an international investor with constant expenditure shares c, and constant relative risk aversion l-y in a dynamic context. The index of value obtained from the consumition rule is used to obtain real returns on N different currencies in terms of their purchasing power over N goods. The portfolio rule is expressed in terms of the determinants of the purchasing powers, namely exchange rates and prices expressed in the numeraire currency. The optimal portfolio is interpreted as a capital position given by the expenditure shares and hedging zero net-worth ortolios depending on unnticioated inflation and risk aversion. It is shown that the minimum variance portfolio is independent of returns, but depends on expenditure patterns.while the speculative portfolio depends on risk aversion and real return differentials. When the effect of preferences on real return differentials is made explicit, it is shown that the minimum variance portfolio is affected by risk aversion. In that case, the effect of an increase in on the portfolio proportions x will be positive when relative risk aversion is greater than one, as generally presumed. Actual data from eight major countries is used to compute optimal 'ortfolios based on real return differentials for different weighting schemes, degrees of risk aversion and sample periods,when exchange rates and prices are assumed to be Brownian.

We propose a simple procedure for the real‐time estimation of the average rainfall over a catchment area. The rainfall is modeled as a two‐dimensional random field. The average areal rainfall is computed by a linear unbiased minimum variance estimation method (kriging) which requires knowledge of the variogram of the random field. We propose a time‐varying estimator for the variogram which takes into account the influences of both the seasonal variations and the rainfall intensity. Our average areal rainfall estimator has been implemented in practice. We illustrate its application to real data in two river basins in Belgium. Finally, it is shown how the method can be used for the optimal selection of the rain gauge locations in a basin.

Résumé Dans cette communication, on propose d'utiliser le filtre de Kalmana deux étages pour estimer l'état et les défauts de systemes stochastiquesa temps variant eta temps discret. Ces défauts sont générés par un processus stochastique et affectent, par des actions additives, l'équation d'état et l'équation de mesure. Deux cas de figure sont analysés. Le premier suppose que les propriétés statistiques du modele de défaut sont connues, par contre dans le second cas les propriétés statistiques et les conditions ...

Abstract:Direction-of-arrival (DOA) estimation algorithms in array processing applications have been developed under time-invariant wavefronts. In most applications this assumption is not realistic due to the nonhomogeneous propagation medium which can distort the wavefront received by the array. This paper extends the author’s previous work on signal-to-noise ratio (SNR) estimation, in developing a novel approach for estimating the DOA of a single narrow-band amplitude-distorted wavefront received by an arbitrary antenna array. The distorted-amplitude wavefront is assumed to vary according to the first order autoregressive AR(1) model with unknown coefficients. An approximate maximum-likelihood-based (ML-based) approach to estimate the DOA parameter is developed in the high SNR scenario. Compared with the classical ML method that requires computationally prohibitive multi-dimensional search, the proposed approach obtains the DOA estimate by maximizing a new cost function with respe...

This paper examines the asymptotic performance of MUSIC-like algorithms for estimating directions of arrival (DOA) of narrowband complex noncircular sources. Using closed-form expressions of the covariance of the asymptotic distribution of different projection matrices, it provides a unifying framework for investigating the asymptotic performance of arbitrary subspacebased algorithms valid for Gaussian or non-Gaussian and complex circular or noncircular sources. We also derive different robustness properties from the asymptotic covariance of the estimated DOA given by such algorithms. These results are successively applied to four algorithms: to two attractive MUSIC-like algorithms previously introduced in the literature, to an extension of these algorithms, and to an optimally weighted MUSIC algorithm proposed in this paper. Numerical examples illustrate the performance of the studied algorithms compared to the asymptotically minimum variance (AMV) algorithms introduced as benchmarks.

This paper presents a series of user parameter-free iterative Sparse Asymptotic Minimum Variance (SAMV) approaches for array processing applications based on the asymptotically minimum variance (AMV) criterion. With the assumption of abundant snapshots in the direction-of-arrival (DOA) estimation problem, the signal powers and noise variance are jointly estimated by the proposed iterative AMV approach, which is later proved to coincide with the Maximum Likelihood (ML) estimator. We then propose a series of power-based iterative SAMV approaches, which are robust against insufficient snapshots, coherent sources and arbitrary array geometries. Moreover, to overcome the direction grid limitation on the estimation accuracy, the SAMV-Stochastic ML (SAMV-SML) approaches are derived by explicitly minimizing a closed form stochastic ML cost function with respect to one scalar parameter, eliminating the need of any additional grid refinement techniques. To assist the performance evaluation, approximate solutions to the SAMV approaches are also provided at high signal-to-noise ratio (SNR) and low SNR, respectively. Finally, numerical examples are generated to compare the performance of the proposed approaches with existing approaches.

The main goal of this project is to identify new mathematical approaches to best match and proximity searching with particular application to very large data stores. In this dissertation I develop a new, linear time hierarchical clustering algorithm and I validate it in a wide range of cases.

Coding of data, usually upstream of data analysis, has crucial implications for the data analysis results. By modifying the data codingthrough use of less than full precision in data values-we can aid appreciably the effectiveness and efficiency of the hierarchical clustering. In our first application, this is used to lessen the quantity of data to be hierarchically clustered. The approach is a hybrid one, based on hashing and on the Ward minimum variance agglomerative criterion. In our second application, we derive a hierarchical clustering from relationships between sets of observations, rather than the traditional use of relationships between the observations themselves. This second application uses embedding in a Baire space, or longest common prefix ultrametric space. We compare this second approach, which is of O(n log n) complexity, to k-means.

Loran-C is a navigational aid that relies on the ability to make correct estimates of the Times of Arrival (TOAs) of signals received over a noisy radio channel. How good is the performance of Loran-C in this respect? Is it close to optimal? Are there perhaps better estimation techniques available than we currently employ? In this paper we use Estimation Theory to find the optimal limit of the accuracy of TOA estimates, for a given signal-to-noise ratio (SNR). Specifically, we compute the Cramer-Rao Lower Bound (CRLB) which specifies the absolute minimum error variance that can be achieved by an unbiased estimator. From the CRLB we can estimate the best repeatable accuracy that can be attained using a set of TOAs. We compute here the CRLBs for the TOA of a Loran-C signal with both Additive White Gaussian Noise (AWGN) and Additive Coloured Gaussian Noise (ACGN). We then employ the Maximum Likelihood Estimator (MLE) and compare its results with that of CRLB. Simulation results show th...

Most evolutionary processes occur in a spatial context and several spatial analysis techniques have been employed in an exploratory context. However, the existence of autocorrelation can also perturb significance tests when data is analyzed using standard correlation and regression techniques on modeling genetic data as a function of explanatory variables. In this case, more complex models incorporating the effects of autocorrelation must be used. Here we review those models and compared their relative performances in a simple simulation, in which spatial patterns in allele frequencies were generated by a balance between random variation within populations and spatially-structured gene flow. Notwithstanding the somewhat idiosyncratic behavior of the techniques evaluated, it is clear that spatial autocorrelation affects Type I errors and that standard linear regression does not provide minimum variance estimators. Due to its flexibility, we stress that principal coordinate of neighbor matrices (PCNM) and related eigenvector mapping techniques seem to be the best approaches to spatial regression. In general, we hope that our review of commonly used spatial regression techniques in biology and ecology may aid population geneticists towards providing better explanations for population structures dealing with more complex regression problems throughout geographic space.

A thorough and extensive wild germplasm exploration survey was undertaken and 50 high yielding candidate plus trees (CPTs) of Pongamia pinnata (L.) Pierre from different locations from a latitudinal and longitudinal spread between 12°41 0 and 22°E longitude and 77°and 84°40 0 N latitude covering 11 locations in an area spread of 150,000 km 2 were collected for evaluating genetic association and variability in seed and growth characters. There were significant differences observed in seed morphology and oil content as was in plant height, and number of branches in the progeny trial. Plant height and number of branches exhibited much higher values of both phenotypic and genotypic variance than observed in the seed characters. Among seed characters oil content exhibited highest broad sense heritability of more than 93% followed by seed length (90.0%). In contrast seed width showed the second highest genetic advance of 5.64% following the highest genetic advance of 10.15% exhibited by oil content. Hierarchical clustering by Ward's Minimum Variance Cluster Analysis method showed phylogeographic patterns of genetic diversity. K means clustering revealed that trees from different geographic regions were grouped together in a cluster and as were trees from the same geographical area placed in different clusters suggesting that geographical diversity did not go hand in hand with genetic diversity. In addition clustering identified promising accessions with favourable traits for future establishment of orchards.

This is a Working Paper and the aulhor(s) would welcome any comments on the prcsent tcxt. Citations should refer to a Worloi1g Paper of die J11tematio11al Monetary F1111d. The views exprcssed are those of thc author(s) and do not necessarily represent those of the Fund. WP/98/28 INTERNATIONAL MONETARY FUND Monetary and Exchange Affairs Department Dollarization of Financia! Intermediation: Causes and Policy lmplications • • Prepared ~y Alain !ze and l;'.~úardo Levy-Yeyati 1 Authorized far distribution by Tomás Baliño

We consider a probability model where the design-based approach to inference under simple random sampling of a ÿnite population encompasses a simple random permutation super-population model. The model consists of an expanded set of random variables following a random permutation probability distribution that keeps track of both the units' labels and positions in the permutation. In particular, since we keep track of the labels, the model allows us to attack the problem of estimation of a unit's parameter. While some linear combinations of the expanded set of random variables correspond to linear combinations of the unit parameters, other linear combinations correspond to random variables known as random e ects. Using a prediction technique similar to that employed under the model-based approach, we develop optimum estimators of the linear combinations of the unit parameters and optimum predictors of the random e ects. The unbiased minimum variance estimator of the population mean is the sample mean and of a unit parameter is the Horvitz-Thompson estimator if the unit is included in the sample, and zero otherwise. The predictor of the random variable at a given position in the permutation is the realized unit's parameter for positions in the sample, and the sample mean for other positions. For other linear functions, unique minimum variance unbiased estimators may not exist.

Abstract: This study deals with a noise robust distributed speech recognizer for real-world applications by deploying feature parameter compensation technique. To realize this objective, Mel-LP based speech analysis has been used in speech coding on the linear frequency scale by applying a first-order all-pass filter instead of a unit delay. To minimize the mismatch between training and test phases, Cepstral Mean Normalization (CMN) and Blind Equalization (BEQ) have been applied to enhance Mel-LP cepstral coefficient as an ...

As institutional investors have become more aggressive in deploying their capital, fund managers have become more creative with their product offerings. In this paper, we consider a new institutional fund of mutual funds, a portfolio that combines the "best-idea" stocks from two underlying primary funds. The sponsor of this portfolio has chosen to weight all of the stocks equally, even those chosen by both of the underlying funds' managers. However, stocks chosen by both managers may be more likely to outperform. We propose an alternative weighting scheme, where these "confirmed" stocks are weighted more heavily. We show that this overweighting strategy leads to a higher expected portfolio return than does the equally weighted scheme used by the sponsor.

A brief account of the geology of the Isukasis area is given and is biased toward the main theme of the itinerary for the area: What has been established about the protoliths of the early Archean rocks of the area - the Isua supracrustal belt and the Amitsoq gneisses? The area's long and complex tectonometamorphic history of events can be divided into episodes using a combination of dike chronology, isotopic, and petrological studies. The earliest dikes, the ca 3700 Ma Inaluk dikes, intrude the earliest (tonalitic) components of the Amitsoq gneisses but are themselves cut up by the injection of the younger (granitic and pegmatitic) phases of the Amitsoq gneisses of the area. The areas of low late Archean deformation, strongly deformed early Archean mafic rocks have coarse grained metamorphic segregations and are cut by virtually undeformed mid-Archean Tarssartoq (Ameralik) dikes devoid of metamorphic segregations. The shows that the area was affected by regional amphibolite faci...

Due to the dramatic increase of data volumes in different applications, it is becoming infeasible to keep these data in one centralized machine. It is becoming more and more natural to deal with distributed databases and networks. That is why distributed data mining techniques have been introduced. One of the most important data mining problems is data clustering. While many clustering algorithms exist for centralized databases, there is a lack of efficient algorithms for distributed databases. In this paper, an efficient algorithm is proposed for clustering distributed databases. The proposed methodology employs an iterative optimization technique to achieve better clustering objective. The experimental results reported in this paper show the superiority of the proposed technique over a recently proposed algorithm based on a distributed version of the well known K-Means algorithm (Datta et al. 2009) [1].

Co-channel interference limits the capacity of cellular systems, reduces the throughput of wireless local area networks, and is the major hurdle in deployment of high altitude communication platforms. It is also a problem for systems operating in unlicensed bands such as the 2.4 GHz ISM band and for narrowband systems that have been overlaid with spread spectrum systems. v V. Conclusions and Future Work 5.1 Conclusions 79 5.2 Future Work References Appendix A vi List of figures 2.1 Parallel MSK receiver 2.2 In-phase and quadrature components of MSK signal 2.3 Signal constellation of the MSK signals for (a) odd (b) even sampling instants 2.4 Phase trellis for the two MSK signals. One possible path for each of the two signals is high lightened 2.5 Receiver architecture of MSK 2.6 Signal constellation for MAMSK (a) constituent signals have equal amplitude (b) constituent signals have unequal amplitude 2.7 Power spectra of MSK, OMSK/MAMSK and QPSK 15 2.8 Error performance of MSK signal in an AWGN channel with and without interference 2.9 Error performance when the interference is not orthogonal (SIR = 0dB) 17 2.10 Error performance for various SIRs and phase deviations, Eb/No = 10dB 2.11 Error performance for various SIRs and phase deviations, Eb/No = 18dB 2.12 Error performance for different values of frequency separation 2.13 (a) Both tones overlap (b) One tone overlaps 3.1 Correlation receiver 3.2 Signal constellation of a QPSK signal (a) ideal (b) skewed 29 3.3 Comparison of MF and MVU for non-orthogonal carriers. Phase error = 10 degrees 3.4 Comparison of MF and MVU for non-orthogonal carriers. Phase error = 20 degrees 3.5 Checkerboard plots for various phase error values. Lighter shades represent stronger values. White is strongest and black is zero 3.6 Decorrelating detector for the synchronous channel 3.7 Geometrical representation of matched filter for two users 3.8 Geometrical representation of a decorrelator for two users 3.9 MSK bit error rate at SIR = 0dB and Eb/No = 18dB 3.10 MSK bit error rate at SIR = 0dB and phase deviation = 10deg 3.11 MSK bit error rate at Eb/No = 18 dB and SIR = 0,1,2,3 dB 3.12 MSK bit error rate at SIR = 0dB and Eb/No = 12 dB 3.13 MSK bit error rate for SIR = 0dB, frequency offset = 0.5 and phase offset = 30 deg 4.1 Steps involved in Model-based spectral estimation 4.2 Signal flow graph for the AR model 59 4.3 Cramer-Rao lower bound for the AR(1) PSD 63 4.4 Pole-zero plot of a fifth order filter for noise and high noise case 69 4.5 Performance of LD and model-based demodulator for various filter lengths 4.6 Performance of model-based demodulator for various values of the vii phase difference between the signal of interest and an interferer 4.7 Performance of model-based demodulator with LMS update 4.8 Performance of model-based demodulator using angle of filter coefficients 4.9 Comparison of various model-based demodulation schemes 4.10 Performance of limiter discriminator for various carrier frequency separations 4.11 Performance of model-based demodulator for various carrier frequency separations 4.12 Comparison of limiter discriminator and model-based demodulator for carrier frequency separation of 1/2T 4.13 Comparison of model-based demodulator and limiter discriminator in Rayleigh fading environment (FdT = 0.002, Eb/No = 20dB) Chapter II MSK Modulation and Co-Channel Interference Chapter II MSK Modulation and Co-Channel Interference 2.1 Introduction Co-channel interference limits the capacity of cellular systems [Las97], reduces the throughput of wireless local area networks (WLANs) and personal area networks (WPANs) [Amr01], and is also the major hurdle in deployment of high altitude communication platforms [Saf00]. In this chapter we will study the effect of co-channel interference on Minimum Shift Keyed (MSK) signals, as it is the most popular modulation format. GMSK, Gaussian pulse shaped MSK, has been adopted by many cellular standards including GSM (Global System for Mobile Communications), CDPD (Cellular Digital Packet Data) and DECT (Digital European Cordless Telephone). Bluetooth also uses a variant of GMSK called GFSK with a frequency separation less than the theoretical minimum required for orthogonal detection of two tones. We will briefly describe the problem of co-channel interference followed by an introduction to the MSK signal and its mathematical model. We will then consider a special case of co-channel interference when the two signals are orthogonal to each other. We next quantify the interference when the signals are not orthogonal. Finally simulation results are presented for both the cases. 2.2 Co-Channel Interference Co-channel interference is a major problem in current as well as emerging wireless communication systems, e.g., in cellular radio, Bluetooth and wireless local area networks and systems employing an airborne communications node. In cellular radio networks the coverage area is split into cells and the same set of frequencies is used for cells that are separated geographically. These are called co-channel cells. The minimum

We present a comprehensive analysis of the performance of different classes of Gaussian states in the estimation of Gaussian phase-insensitive dissipative channels. In particular, we investigate the optimal estimation of the damping constant and reservoir temperature. We show that, for two-mode squeezed vacuum probe states, the quantum-limited accuracy of both parameters can be achieved simultaneously. Moreover, we show that for both parameters two-mode squeezed vacuum states are more efficient than either coherent, thermal or single-mode squeezed states. This suggests that at high energy regimes two-mode squeezed vacuum states are optimal within the Gaussian setup. This optimality result indicates a stronger form of compatibility for the estimation of the two parameters. Indeed, not only the minimum variance can be achieved at fixed probe states, but also the optimal state is common to both parameters. Additionally, we explore numerically the performance of non-Gaussian states for particular parameter values to find that maximally entangled states within d-dimensional cutoff subspaces [d ≤ 6] perform better than any randomly sampled states with similar energy. However, we also find that states with very similar performance and energy exist with much less entanglement than the maximally entangled ones.

In this work different concepts of efficient solutioos to problems of Stochastic Multiple Objective Prograrnming are analyzed. We centre OUT interest on problems in which sorne of the objective functions depelld on random parameters. The existence of different concepts of efficiency for oue single stochastic problem, 8uch as expected value efficiency, minimum-risk efficiency, etc., raise the question of their "quality". Starting from this idea we establish sorne relationships between the different concepts. OUT study enables liS to determine what type of efficient solutions are obtained by each of these concepts.

A detailed analysis of small period (15900 set) magnetohydrodynamic (MHD) turbulences of the interplanetary magnetic field (IMF) has been made using Pioneer-l 1 high time resolution data (0.75 set) inside a Corotating Interaction Region (CIR) at a heliocentric distance of 2.5 AU in 1973. The metods used are the hodogram analysis, the minimum variance matrix analysis and the coherence analysis. The minimum variance analysis gives evidence of linear polarized wave modes. Coherence analysis has shown that the field fluctuations are dominated by the magnetosonic fast modes with periods 15 set to 15 min. However, it is also shown that some small amplitude Alfven waves are present in the trailing edge of this region with characteristic periods (15-200 set). The observed wave modes are locally generated and possibly attributed to the scattering of Alfven waves energy into random magnetosonic waves.