In this dissertation, the aim is to investigate the empirical relationship between the partial pressure of CO2 (pCO2) and other ocean variables in the Southern Ocean, by using a small percentage of the available data. CO2 is one of the main greenhouse gases that contributes to global warming and climate change. The concentration of anthropogenic CO2 in the atmosphere, however, would have been much higher if some of it was not absorbed by oceanic and terrestrial sinks. The oceans absorb and release CO2 from and to the atmosphere. Large regions in the Southern Ocean are expected to be a CO2 sink. However, the measurements of CO2 concentrations in the ocean are sparse in the Southern Ocean, and accurate values for the sinks and sources cannot be determined. In addition, it is difficult to develop accurate oceanic and ocean-atmosphere models of the Southern Ocean with the sparse observations of CO2 concentrations in this part of the ocean. In this dissertation classical techniques are investigated to determine the empirical relationship between pCO2 and other oceanic variables using in situ measurements. Additionally, sampling techniques are investigated in order to make a judicious selection of a small percentage of the total available data points in order to develop an accurate empirical relationship. Data from the SANAE49 cruise stretching between Antarctica and Cape Town are used in this dissertation. The complete data set contains 6103 data points. The maximum pCO2 value in this stretch is 436.0 μatm, the minimum is 251.2 μatm and the mean is 360.2 μatm. An empirical relationship is investigated between pCO2 and the variables Temperature (T), chlorophyll-a concentration (Chl), Mixed Layer Depth (MLD) and latitude (Lat). The methods are repeated with latitude included and excluded as variable respectively. D-optimal sampling is used to select a small percentage of the available data for determining the empirical relationship. Least squares optimization is used as one method to determine the empirical relationship. For 200 D-optimally sampled points, the pCO2 prediction with the fourth order equation yields a Root Mean Square (RMS) error of 15.39 μatm (on the estimation of pCO2) with latitude excluded as variable and a RMS error of 8.797 μatm with latitude included as variable. Radial basis function (RBF) interpolation is another method that is used to determine the empirical relationship between the variables. The RBF interpolation with 200 D-optimally sampled points yields a RMS error of 9.617 μatm with latitude excluded as variable and a RMS error of 6.716 μatm with latitude included as variable. Optimal scaling is applied to the variables in the RBF interpolation, yielding a RMS error of 9.012 μatm with latitude excluded as variable and a RMS error of 4.065 μatm with latitude included as variable for 200 D-optimally sampled points. / Thesis (MSc (Applied Mathematics))--North-West University, Potchefstroom Campus, 2012

Modern computational science permits the accurate solution of nonlinear partial differential equations (PDEs) on overlapping computational domains, known as an overset approach. The complex grid interconnectivity inherent in the overset method can introduce errors in the solution through “orphan” points, i.e., grid points for which reliable solution donor points cannot be located. For this reason, a variety of data transfer strategies based on scattered data interpolation techniques have been assessed with application to both overset and hybrid methodologies. Scattered data approaches are attractive because they are decoupled from solver type and topology, and may be readily applied within existing methodologies. In addition to standard radial basis function (RBF) interpolation, a novel steered radial basis function (SRBF) interpolation technique has been developed to introduce data adaptivity into the data transfer algorithm. All techniques were assessed by interpolating both continuous and discontinuous analytical test functions. For discontinuous functions, SRBF interpolation was able to maintain solution gradients with the steering technique being the scattered-data analog of a slope limiter. In comparison with linear mappings, the higher-order approaches were able to more accurately preserve flow physics for arbitrary grid configurations. Overset validation test cases included an inviscid convecting vortex, a shock tube, and a turbulent ship airwake. These were studied within unsteady Reynolds-Averaged Navier-Stokes (URANS) simulations to determine quantitative and qualitative improvements when applying RBF interpolation over current methods. The convecting vortex was also analyzed on a grid configuration which contained orphan points under the state-of-the-art overset paradigm. This was successfully solved by the RBF-based algorithm, which effectively eliminated orphans by enabling high-order extrapolation. Order-of-magnitude reductions in error compared to the exact vortex solution were observed. In addition, transient conservation errors that persisted in the original overset methodology were eliminated by the RBF approach. To assess the effect of advanced mapping techniques on the fidelity of a moving grid simulation, RBF interpolation was applied to a hybrid simulation of an isolated wind turbine rotor. The resulting blade pressure distributions were comparable to a rotor simulation with refined near-body grids.

Since the introduction of modern computers, numerical methods for atmospheric simulations have routinely been applied for weather prediction, and in the last fifty years, there has been a steady improvement in the accuracy of forecasts. Accurate numerical models of the atmosphere are also becoming more important as researchers rely on global climate simulations to assess and understand the impact of global warming. The choice of grid in a numerical model is an important design decision and no obvious optimal choice exists for computations in spherical geometry. Despite this disadvantage, grid-based methods are found in all current circulation models. A different approach to the issue of discretizing the surface of the sphere is given by meshless methods, of which radial basis function (RBF) methods are becoming prevalent. In this thesis, RBF methods for simulation of atmospheric flows are explored. Several techniques are introduced to increase the efficiency of the methods. By utilizing a novel algorithm for adaptively placing the node points, accuracy is shown to improve by over one order of magnitude for two relevant test problems. The computational cost can also be reduced by using a local finite difference-like RBF scheme. However, this requires a stabilization mechanism for the hyperbolic problems of interest here. A hyper-viscosity scheme is introduced to address this issue. Another stability issue arising from the ill-conditioning of the RBF basis for almost-flat basis functions is also discussed in the thesis, and two algorithms are proposed for dealing with this stability problem. The algorithms are specifically tailored for the task of creating finite difference weights using RBFs and are expected to overcome the issue of stationary error in local RBF collocation.

Orientador: Leo Pini Magalhães / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-09T21:26:28Z (GMT). No. of bitstreams: 1 Schroeder_GreyceNogueira_M.pdf: 4292885 bytes, checksum: 37f5409f283c62fcb20cf1456b55943f (MD5) Previous issue date: 2007 / Resumo: O morphing de imagens é uma transformação que mapeia uma imagem em outra, alterando tanto a sua forma quanto as suas intensidades. Esta transformação possui diversas aplicações em imagens médicas e especialmente na indústria de entretenimento. Este trabalho objetiva, através do uso de morphing, apresentar um protótipo para a simulação de envelhecimento de faces frontais, utilizando um método chamado Funções de Base Radial (RBF) juntamente com um modelo quantitativo para expressar o processo de envelhecimento no rosto humano. O protótipo trabalha com imagens de pessoas a partir de 20 anos e realiza o envelhecimento até no máximo 70 anos. Além disso, o trabalho apresenta uma revisão bibliográfica sobre as principais técnicas de morphing e sobre o que já foi realizado sobre simulações computacionais de envelhecimento facial por imagens / Abstract: Image morphing is a transformation that maps one image into another, altering both its shape and intensities. These types of transformation have a wide range of applications in medical imaging and, specially, in entertainment industry. This work attempts to present a prototype for aging simulation on frontal face images, using a method of morphing called Radial Basis Functions (RBF) together with a quantitative model for expressing human aging. The prototype works with images from people with 20 years old up and performs the aging up to 70 years. Moreover, the work presents a bibliographical revision on the main techniques of morphing and on the state of the art on computational simulations of image face aging / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

Made available in DSpace on 2014-06-12T15:52:45Z (GMT). No. of bitstreams: 2 arquivo4692_1.pdf: 1118830 bytes, checksum: 96894dd8a22373c59d67d3b286b6c902 (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2003 / Um dos principais obstáculos para o uso em larga escala das Redes Neurais é a dificuldade de definir valores para seus parâmetros ajustáveis. Este trabalho discute como as Redes Neurais de Funções Base Radial (ou simplesmente Redes RBF) podem ter seus parâmetros ajustáveis definidos por algoritmos genéticos (AGs). Para atingir este objetivo, primeiramente é apresentado uma visão abrangente dos problemas envolvidos e as diferentes abordagens utilizadas para otimizar geneticamente as Redes RBF. É também proposto um algoritmo genético para Redes RBF com codificação genética não redundante baseada em métodos de clusterização. Em seguida, este trabalho aborda o problema de encontrar os parâmetros ajustáveis de um algoritmo de aprendizagem via AGs. Este problema é também conhecido como o problema de seleção de modelos. Algumas técnicas de seleção de modelos (e.g., validação cruzada e bootstrap) são usadas como funções objetivo do AG. O AG é modificado para adaptar-se a este problema por meio de heurísticas tais como narvalha de Occam e growing entre outras. Algumas modificações exploram características do AG, como por exemplo, a abilidade para resolver problemas de otimização multiobjetiva e manipular funções objetivo com ruído. Experimentos usando um problema benchmark são realizados e os resultados alcançados, usando o AG proposto, são comparados com aqueles alcançados por outras abordagens. As técnicas propostas são genéricas e podem também ser aplicadas a um largo conjunto de algoritmos de aprendizagem

Orientador: Wilson Castro Ferreira Junior / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Matemática, Estatística e Computação Científica / Made available in DSpace on 2018-08-16T08:15:58Z (GMT). No. of bitstreams: 1 Vilamiu_RaphaelGustavod'Almeida_D.pdf: 17502226 bytes, checksum: af859cc18d3a4c5a74c324d3e6c0865a (MD5) Previous issue date: 2010 / Resumo: Nesta tese, desenvolvemos uma técnica para gerar grafos à partir de conjuntos de séries temporais considerando a correlação entre estas e uma extensão do Método de Decomposição Empírica (EMD) para grafos (GEMD). Tal trabalho se justifica pelo fato de que uma grande gama de sinais formados por conjuntos de séries temporais não possuem uma localização bem definida em nenhum espaço n-dimensional. Desta forma, as relações entre as séries temporais só são satisfatoriamente representadas com o uso de grafos. Contudo, o desenvolvimento do GEMD é dependente do uso de algum método de interpolação em grafos. Tais métodos são escassos e não produzem propriedades satisfatórias para o uso no GEMD. Para esta finalidade, estendemos a interpolação por Funções de Base Radial (RBF) em Grafos (GRBF), onde a norma euclidiana no cálculo da matriz de interpolação por RBF é substituída pela norma induzida pelo grafo. Testes numéricos sugerem que a extensão possui boas propriedades de convergência e uma técnica é desenvolvida para garantir a existência e unicidade da solução. Finalmente, aplicamos o GEMD em um conjunto de dados de incidência de Dengue Hemorrágica na Tailândia. Os modos intrínsecos encontrados desta forma não apresentam nenhuma onda viajante emanando de nenhuma das províncias, contrastando com o resultado utilizando o EMD original [5]. Além disso, os períodos médios dos modos intrínsecos de [5] são claramente distintos dos encontrados pela decomposição por GEMD / Abstract: In this thesis, we developed a technique to generate a graph from a set of temporal series, which are then decomposed trough an extension of the Empirical Mode Decomposition (EMD) on Graphs (GEMD) created by us. This procedure is justified by the fact that a huge amount of signals cannot be properly localized on an n-dimensional space which can only be properly represented by a graph. The development of the GEMD is dependent on some graph interpolation method. Such methods are scarce in the literature and do not have the necessary properties to accomplish the GEMD decomposition. For this goal, we extend the Radial Basis Functions (RBF) interpolation to graphs (GRBF), where the euclidean norm used in the calculation of the RBF interpolation matrix is substituted by a graph induced norm. Numerical tests suggests that GRBF have good convergence properties and we present a technique which guarantees the existence and uniqueness of the solution. We finally apply the GEMD decomposition to a data set of Dengue Hemorrhagic Fever incidence in Thailand. The intrinsic modes found in this way do not show any traveling wave emanating from any of the provinces, contrasting with the results found using the original EMD [5]. Moreover, the mean period for the intrinsic modes in [5] are clearly diverse of those found by GEMD decomposition / Doutorado / Doutor em Matemática Aplicada

This report evaluates possibilities to combine volumetric ultrasound (us) data together with the recent work published on advanced skin rendering techniques. We focus mainly on how to filter us data and localize surfaces within us data. We also evaluate recent skin rendering techniques in order to have a good understanding of what is needed from the us for rendering realistic skin. us data is acquired using sonography and have a low signal-to-noise ratio by nature, this makes it harder to extract surfaces compared to other medical data acquisition methods such as ct and mr. This report present an algorithm which implements a variational classification technique to emphasize surfaces within us and using a rbf network to fit an implicit function to these surfaces. Using this approach presented we have successfully extract smooth meshes from the noisy us data.

Includes bibliographical references / In this thesis, the development of a strongly coupled, partitioned fluid-structure interactions (FSI) solver is outlined. Well established methods are analysed and new methods are proposed to provide robust, accurate and efficient FSI solutions. All the methods introduced and analysed are primarily geared towards the solution of incompressible, transient FSI problems, which facilitate the use of black-box sub-domain field solvers. In the first part of the thesis, radial basis function (RBF) interpolation is introduced for interface information transfer. RBF interpolation requires no grid connectivity information, and therefore presents an elegant means by which to transfer information across a non-matching and non-conforming interface to couple finite element to finite volume based discretisation schemes. The transfer scheme is analysed, with particular emphasis on a comparison between consistent and conservative formulations. The primary aim is to demonstrate that the widely used conservative formulation is a zero order method. Furthermore, while the consistent formulation is not provably conservative, it yields errors well within acceptable levels and converges within the limit of mesh refinement. A newly developed multi-vector update quasi-Newton (MVQN) method for implicit coupling of black-box partitioned solvers is proposed. The new coupling scheme, under certain conditions, can be demonstrated to provide near Newton-like convergence behaviour. The superior convergence properties and robust nature of the MVQN method are shown in comparison to other well-known quasi-Newton coupling schemes, including the least squares reduced order modelling (IBQN-LS) scheme, the classical rank-1 update Broyden's method, and fixed point iterations with dynamic relaxation. Partitioned, incompressible FSI, based on Dirichlet-Neumann domain decomposition solution schemes, cannot be applied to problems where the fluid domain is fully enclosed. A simple example often provided in the literature is that of balloon inflation with a prescribed inflow velocity. In this context, artificial compressibility (AC) will be shown to be a useful method to relax the incompressibility constraint, by including a source term within the fluid continuity equation. The attractiveness of AC stems from the fact that this source term can readily be added to almost any fluid field solver, including most commercial solvers. AC/FSI is however limited in the range of problems it can effectively be applied to. To this end, the combination of the newly developed MVQN method with AC/FSI is proposed. In so doing, the AC modified fluid field solver can continue to be treated as a black-box solver, while the overall robustness and performance are significantly improved. The study concludes with a demonstration of the modularity offered by partitioned FSI solvers. The analysis of the coupled environment is extended to include steady state FSI, FSI with free surfaces and an FSI problem with solid-body contact.

Metalearning aims to obtain knowledge of the relationship between the mechanism of learning and the concrete contexts in which that mechanisms is applicable. As new mechanisms of learning are continually added to the pool of learning algorithms, the chances of encountering behavior similarity among algorithms are increased. Understanding the relationships among algorithms and the interactions between algorithms and tasks help to narrow down the space of algorithms to search for a given learning task. In addition, this process helps to disclose factors contributing to the similar behavior of different algorithms. We first study general characteristics of learning tasks and their correlation with the performance of algorithms, isolating two metafeatures whose values are fairly distinguishable between easy and hard tasks. We then devise a new metafeature that measures the difficulty of a learning task that is independent of the performance of learning algorithms on it. Building on these preliminary results, we then investigate more formally how we might measure the behavior of algorithms at a ner grained level than a simple dichotomy between easy and hard tasks. We prove that, among all many possible candidates, the Classifi er Output Difference (COD) measure is the only one possessing the properties of a metric necessary for further use in our proposed behavior-based clustering of learning algorithms. Finally, we cluster 21 algorithms based on COD and show the value of the clustering in 1) highlighting interesting behavior similarity among algorithms, which leads us to a thorough comparison of Naive Bayes and Radial Basis Function Network learning, and 2) designing more accurate algorithm selection models, by predicting clusters rather than individual algorithms.

To accurately model or predict future operating conditions of a system in engineering or applied mechanics, it is necessary to understand its fundamental principles. These may be the material parameters, defining dimensional characteristics, or the boundary conditions. However, there are instances when there is little to no prior knowledge of the system properties or conditions, and consequently, the problem cannot be modeled accurately. It is therefore critical to define a method that can identify the desired characteristics of the current system without accumulating extensive computation time. This thesis formulates an inverse approach using proper orthogonal decomposition (POD) with an accompanying radial basis function (RBF) interpolation network. This method is capable of predicting the desired characteristics of a specimen even with little prior knowledge of the system. This thesis first develops a conductive heat transfer problem, and by using the truncated POD – RBF interpolation network, temperature values are predicted given a varying Biot number. Then, a simple bifurcation problem is modeled and solved for velocity profiles while changing the mass flow rate. This bifurcation problem provides the data and foundation for future research into the left ventricular assist device (LVAD) and implementation of POD – RBF. The trained POD – RBF inverse approach defined in this thesis can be implemented in several applications of engineering and mechanics. It provides model reduction, error filtration, regularization and an improvement over previous analysis utilizing computational fluid dynamics (CFD).