Nos programas de melhoramento de cana-de-açúcar todos os anos são instalados experimentos com o objetivo de avaliar genótipos que podem eventualmente ser recomendados para o plantio, ou mesmo como genitores. Este objetivo é atingido com o emprego de experimentos em diferentes locais, durante diferentes colheitas. Além disso, frequentemente há grande desbalanceamento, pois nem todos os genótipos são avaliados em todos os experimentos. O emprego de abordagens tradicionais como análise de variância conjunta (ANAVA) é inviável devido à condição de desbalanceamento e ao fato de as pressuposições não modelarem adequadamente o relacionamento entre as observações. O emprego de modelos misto utilizando a metodologia REML/BLUP é uma alternativa para análise desses experimentos em cana-deaçúcar, permitindo também incorporar a informação de parentesco entre os indivíduos. Nesse contexto, foram analisados 44 experimentos (locais) de cana-de-açúcar do programa de melhoramento da cana-de-açúcar do Instituto Agronômico de Campinas (IAC), com 74 genótipos (clones e variedades) e com até 5 colheitas. O delineamento foi o de blocos ao acaso com 2 a 6 repetições. O caráter analisado foi TPH (Tonelada de pol por hectare). Foram testados 40 modelos, os 20 primeiros foram avaliadas diferentes estrutura de VCOV para locais e colheitas, e os 20 seguintes, além das matrizes de VCOV, foi incorporada a matriz de parentesco genético, A. De acordo com AIC, verificou-se que o Modelo 11, o qual assume as matrizes FA1, AR1 e ID, para locais, colheitas e genótipos, respectivamente, foi o melhor, e portanto, o mais eficiente para seleção de genótipos superiores. Quando comparado ao modelo tradicional (médias dos experimentos), houve mudanças no ranqueamento dos genótipos. Há correlação entre o modelo tradicional e o Modelo 11 (_ = 0, 63, p-valor < 0, 001). A opção de utilizar modelo misto sem ajustar as matrizes de VCOV (Modelo 1) é relativamente melhor do que usar o Modelo Tradicional. Isto foi evidenciado pela correlação mais alta entre os modelos 1 e 11 (_ = 0, 87 com p-valor < 0, 001). Acredita-se que o emprego do Modelo 11 junto com experiência do melhorista poderá aumentar a eficiência de seleção em programas de melhoramento de cana-de-açúcar. / In breeding programs of sugarcane every year experiments are installed to evaluate the performance of genotypes, in order to select superior varieties and genitors. The use of ordinary approaches such as joint analysis of variance (ANOVA) is unfeasible due to unbalancing and assumptions that do not reflect the standard of relationship of the observations. The use of mixed models using the method REML/BLUP is an alternative. It also allows the incorporation of information from kinship between individuals. In this context, we analyzed 44 trials (locations) of sugarcane breeding program of sugarcane (Agronomic Institute Campinas, IAC), with 74 genotypes (varieties and clones), up to 5 harvests. The experimental design was randomized blocks with 2-6 replicates. The character was examined TPH (Tons of pol per hectare). We tested 40 models, the first 20 were evaluated different VCOV structure to locations and harvests, and 20 following addition of matrix VCOV was incorporated genetic relationship matrix, A. Under AIC, it was found that the model 11, which assumes matrices FA1, AR1 and ID for locations, harvests and genotypes, respectively, was the best. There is a moderate correlation between traditional model and model 11 (_ = 0.63, p-value < 0.001), when ranking the genotypes. The option of using mixed model without adjusting matrices VCOV (model 1) is better than using the traditional model. This was suggested by the higher correlation between models 1 and 11 (_ = 0.87 with p-value < 0.001). We believe that the usage of model 11 together with breeders experience can increase the efficiency of selection in sugarcane breeding programs.

This thesis provides a case study on how the wind direction plays an important role in the amount of rainfall, in the village of Somi$acute{o}$. The primary goal is to illustrate how a meta-analysis, together with circular data analytic methods, helps in analyzing certain environmental issues. The existing GLS meta-analysis combines the merits of usual meta-analysis that yields a better precision and also accounts for covariance among coefficients. But, it is quite limited since information about the covariance among coefficients is not utilized. Hence, in my proposed meta-analysis, I take the correlations between adjacent studies into account when employing the GLS meta-analysis. Besides, I also fit a time series linear-circular regression as a comparable model. By comparing the confidence intervals of parameter estimates, covariance matrix, AIC, BIC and p-values, I discuss an improvement on the GLS meta analysis model in its application to forecasting problem in Environmental study.

One major criticism about the traditional mean-variance portfolio optimization is that it tends to magnify the estimation error. A little estimation error can cause the distortion of the whole portfolio. Two popular ways to solve this problem are to use a resampling method or the Black-Litterman method (Bayesian method). The clustering method is a newer way to solve the problem. Clustering means we group the highly correlated stocks first and treat the group as a single stock. After we group the stocks, we will have some clusters of stocks, then we run the traditional mean-variance portfolio optimization for these clusters. The clustering method can improve the stability of the portfolio and reduce the impact of estimation error. In this project, we will explain why it works and we will perform tests to determine if clustering methods do improve the stabilities and performance of the portfolio.

Covariance matrices are ubiquitous in computational sciences, typically describing the correlation of elements of large multivariate spatial data sets. For example, covari- ance matrices are employed in climate/weather modeling for the maximum likelihood estimation to improve prediction, as well as in computational ground-based astronomy to enhance the observed image quality by filtering out noise produced by the adap- tive optics instruments and atmospheric turbulence. The structure of these covariance matrices is dense, symmetric, positive-definite, and often data-sparse, therefore, hier- archically of low-rank. This thesis investigates the performance limit of dense matrix computations (e.g., Cholesky factorization) on covariance matrix problems as the number of unknowns grows, and in the context of the aforementioned applications. We employ recursive formulations of some of the basic linear algebra subroutines (BLAS) to accelerate the covariance matrix computation further, while reducing data traffic across the memory subsystems layers. However, dealing with large data sets (i.e., covariance matrices of billions in size) can rapidly become prohibitive in memory footprint and algorithmic complexity. Most importantly, this thesis investigates the tile low-rank data format (TLR), a new compressed data structure and layout, which is valuable in exploiting data sparsity by approximating the operator. The TLR com- pressed data structure allows approximating the original problem up to user-defined numerical accuracy. This comes at the expense of dealing with tasks with much lower arithmetic intensities than traditional dense computations. In fact, this thesis con- solidates the two trends of dense and data-sparse linear algebra for HPC. Not only does the thesis leverage recursive formulations for dense Cholesky-based matrix al- gorithms, but it also implements a novel TLR-Cholesky factorization using batched linear algebra operations to increase hardware occupancy and reduce the overhead of the API. Performance reported of the dense and TLR-Cholesky shows many-fold speedups against state-of-the-art implementations on various systems equipped with GPUs. Additionally, the TLR implementation gives the user flexibility to select the desired accuracy. This trade-off between performance and accuracy is, currently, a well-established leading trend in the convergence of the third and fourth paradigm, i.e., HPC and Big Data, when moving forward with exascale software roadmap.

Resumo: Esta tese oferece algumas contribuições à área de monitoramento de processos multivariados. Com respeito ao monitoramento do vetor de médias, investigou-se o desempenho dos gráficos de 2 T baseados em componentes principais e também o desempenho dos gráficos de médias utilizados em conjunto, sendo que cada gráfico monitora a média de uma das características de qualidade. Com respeito ao monitoramento da matriz de covariâncias, foi proposta uma nova estatística baseada nas variâncias amostrais (estatística de VMAX). O gráfico de VMAX é mais eficiente do que o gráfico da variância amostral generalizada S , que é o gráfico usual para o monitoramento da matriz de covariâncias. Uma vantagem adicional dessa nova estatística é que o usuário já está bem familiarizado com o cálculo de variâncias amostrais; o mesmo não pode ser dito em relação à variância amostral generalizada S . O desempenho do gráfico de VMAX foi também avaliado quando se utiliza a amostragem dupla, quando se variam os parâmetros do gráfico de controle, quando se adota o esquema de EWMA e quando se aplicam regras especiais de decisão. Investigou-se também o desempenho dos gráficos de controle destinados ao monitoramento simultâneo do vetor de médias e da matriz de covariâncias. / Abstract: This thesis offers some contributions to the field of monitoring multivariate processes. Regarding to the monitoring of the mean vector, we investigated the performance of the 2 T charts based on principal components and also the performance of the mean charts used simultaneously, where each chart is assigned to control one quality characteristic. Regarding to the monitoring of the covariance matrix, we propose a new statistic based on the sample variances (the VMAX statistic). The VMAX chart is more efficient than the generalized variance S chart, which is the usual chart for monitoring the covariance matrix. An additional advantage of this new statistic is that the user is already well familiar with the calculation of sample variances; we can't say the same regarding to the generalized variance S statistic. We also studied the performance of the VMAX chart with double sampling, with adaptive schemes, with the EWMA procedure and also with special run rules. We also investigated the performance of the control charts designed for monitoring the mean vector and the covariance matrix simultaneously. / Orientador: Antonio Fernando Branco Costa / Coorientador: Fernando Augusto Silva Marins / Banca: Messias Borges Silva / Banca: Ubirajara Rocha Ferreira / Banca: Linda Lee Ho / Banca: Roberto da Costa Quinino / Doutor

No description available.

No description available.

High dimensional covariance matrix estimation is considered in the context of empirical asset pricing. In order to see the effects of covariance matrix estimation on asset pricing, parameter estimation, model specification test, and misspecification problems are explored. Along with existing techniques, which is not yet tested in applications, diagonal variance matrix is simulated to evaluate the performances in these problems. We found that modified Stein type estimator outperforms all the other methods in all three cases. In addition, it turned out that heuristic method of diagonal variance matrix works far better than existing methods in Hansen-Jagannathan distance test. High dimensional covariance matrix as a transformation matrix in generalized least squares is also studied. Since the feasible generalized least squares estimator requires ex ante knowledge of the covariance structure, it is not applicable in general cases. We propose fully banding strategy for the new estimation technique. First we look into the sparsity of covariance matrix and the performances of GLS. Then we move onto the discussion of diagonals of covariance matrix and column summation of inverse of covariance matrix to see the effects on GLS estimation. In addition, factor analysis is employed to model the covariance matrix and it turned out that communality truly matters in efficiency of GLS estimation.

This century is surely the century of data (Donoho, 2000). Data analysis has been an emerging activity over the last few decades. High dimensional data is in particular more and more pervasive with the advance of massive data collection system, such as microarrays, satellite imagery, and financial data. However, analysis of high dimensional data is of challenge with the so called curse of dimensionality (Bellman 1961). This research dissertation presents several methodologies in the application of high dimensional data analysis. The first part discusses a joint analysis of multiple microarray gene expressions. Microarray analysis dates back to Golub et al. (1999). It draws much attention after that. One common goal of microarray analysis is to determine which genes are differentially expressed. These genes behave significantly differently between groups of individuals. However, in microarray analysis, there are thousands of genes but few arrays (samples, individuals) and thus relatively low reproducibility remains. It is natural to consider joint analyses that could combine microarrays from different experiments effectively in order to achieve improved accuracy. In particular, we present a model-based approach for better identification of differentially expressed genes by incorporating data from different studies. The model can accommodate in a seamless fashion a wide range of studies including those performed at different platforms, and/or under different but overlapping biological conditions. Model-based inferences can be done in an empirical Bayes fashion. Because of the information sharing among studies, the joint analysis dramatically improves inferences based on individual analysis. Simulation studies and real data examples are presented to demonstrate the effectiveness of the proposed approach under a variety of complications that often arise in practice. The second part is about covariance matrix estimation in high dimensional data. First, we propose a penalised likelihood estimator for high dimensional t-distribution. The student t-distribution is of increasing interest in mathematical finance, education and many other applications. However, the application in t-distribution is limited by the difficulty in the parameter estimation of the covariance matrix for high dimensional data. We show that by imposing LASSO penalty on the Cholesky factors of the covariance matrix, EM algorithm can efficiently compute the estimator and it performs much better than other popular estimators. Secondly, we propose an estimator for high dimensional Gaussian mixture models. Finite Gaussian mixture models are widely used in statistics thanks to its great flexibility. However, parameter estimation for Gaussian mixture models with high dimensionality can be rather challenging because of the huge number of parameters that need to be estimated. For such purposes, we propose a penalized likelihood estimator to specifically address such difficulties. The LASSO penalty we impose on the inverse covariance matrices encourages sparsity on its entries and therefore helps reducing the dimensionality of the problem. We show that the proposed estimator can be efficiently computed via an Expectation-Maximization algorithm. To illustrate the practical merits of the proposed method, we consider its application in model-based clustering and mixture discriminant analysis. Numerical experiments with both simulated and real data show that the new method is a valuable tool in handling high dimensional data. Finally, we present structured estimators for high dimensional Gaussian mixture models. The graphical representation of every cluster in Gaussian mixture models may have the same or similar structure, which is an important feature in many applications, such as image processing, speech recognition and gene network analysis. Failure to consider the sharing structure would deteriorate the estimation accuracy. To address such issues, we propose two structured estimators, hierarchical Lasso estimator and group Lasso estimator. An EM algorithm can be applied to conveniently solve the estimation problem. We show that when clusters share similar structures, the proposed estimator perform much better than the separate Lasso estimator.

<p>Navigation with unmanned aerial vehicles (UAVs) requires good knowledge of the current position and other states. A UAV navigation system often uses GPS and inertial sensors in a state estimation solution. If the GPS signal is lost or corrupted state estimation must still be possible and this is where simultaneous localization and mapping (SLAM) provides a solution. SLAM considers the problem of incrementally building a consistent map of a previously unknown environment and simultaneously localize itself within this map, thus a solution does not require position from the GPS receiver.</p><p>This thesis presents a visual feature based SLAM solution using a low resolution video camera, a low-cost inertial measurement unit (IMU) and a barometric pressure sensor. State estimation in made with a extended information filter (EIF) where sparseness in the information matrix is enforced with an approximation.</p><p>An implementation is evaluated on real flight data and compared to a EKF-SLAM solution. Results show that both solutions provide similar estimates but the EIF is over-confident. The sparse structure is exploited, possibly not fully, making the solution nearly linear in time and storage requirements are linear in the number of features which enables evaluation for a longer period of time.</p>