A reduced load approximation accounting for link interactions in a loss network

This paper is concerned with evaluating the performance of loss networks. Accurate determination of loss network performance can assist in the design and dimen- sioning of telecommunications networks. However, exact determination can be difficult and generally cannot be done in reasonable time. For these reasons there is much interest in developing fast and accurate approximations. We develop a reduced load approximation that improves on the famous Erlang fixed point approximation (EFPA) in a variety of circumstances. We illustrate our results with reference to a range of networks for which the EFPA may be expected to perform badly.

A reduced load approximation accounting for link interactions in a loss network

This paper is concerned with evaluating the performance of loss networks. Accurate determination of loss network performance can assist in the design and dimensioning of telecommunications networks. However, exact determination can be difficult and generally cannot be done in reasonable time. For these reasons there is much interest in developing fast and accurate approximations. We develop a reduced load approximation which improves on the famous Erlang fixed point approximation (EFPA) in a variety of circumstances. We illustrate our results with reference to a range of networks for which the EFPA may be expected to perform badly.

Reduced basis approximations for electromagnetic applications

Otto-von-Guericke-Universität Magdeburg, Fakultät für Mathematik, Dissertation, 2016

Certified Rapid Solution of Parametrized Linear Elliptic Equations: Application to Parameter Estimation

We present a technique for the rapid and reliable evaluation of linear-functional output of elliptic partial differential equations with affine parameter dependence. The essential components are (i) rapidly uniformly convergent reduced-basis approximations — Galerkin projection onto a space WN spanned by solutions of the governing partial differential equation at N (optimally) selected points in parameter space; (ii) a posteriori error estimation — relaxations of the residual equation that provide inexpensive yet sharp and rigorous bounds for the error in the outputs; and (iii) offline/online computational procedures — stratagems that exploit affine parameter dependence to de-couple the generation and projection stages of the approximation process. The operation count for the online stage — in which, given a new parameter value, we calculate the output and associated error bound — depends only on N (typically small) and the parametric complexity of the problem. The method is thus ideally suited to the many-query and real-time contexts. In this paper, based on the technique we develop a robust inverse computational method for very fast solution of inverse problems characterized by parametrized partial differential equations. The essential ideas are in three-fold: first, we apply the technique to the forward problem for the rapid certified evaluation of PDE input-output relations and associated rigorous error bounds; second, we incorporate the reduced-basis approximation and error bounds into the inverse problem formulation; and third, rather than regularize the goodness-of-fit objective, we may instead identify all (or almost all, in the probabilistic sense) system configurations consistent with the available experimental data — well-posedness is reflected in a bounded "possibility region" that furthermore shrinks as the experimental error is decreased.

Logarithme d'harmoniques sphériques pour le rendu d'ombres douces de champs de hauteurs et de maillages

Les ombres sont un élément important pour la compréhension d'une scène. Grâce à elles, il est possible de résoudre des situations autrement ambigües, notamment concernant les mouvements, ou encore les positions relatives des objets de la scène. Il y a principalement deux types d'ombres: des ombres dures, aux limites très nettes, qui résultent souvent de lumières ponctuelles ou directionnelles; et des ombres douces, plus floues, qui contribuent à l'atmosphère et à la qualité visuelle de la scène. Les ombres douces résultent de grandes sources de lumière, comme des cartes environnementales, et sont difficiles à échantillonner efficacement en temps réel. Lorsque l'interactivité est prioritaire sur la qualité, des méthodes d'approximation peuvent être utilisées pour améliorer le rendu d'une scène à moindre coût en temps de calcul. Nous calculons interactivement les ombres douces résultant de sources de lumière environnementales, pour des scènes composées d'objets en mouvement et d'un champ de hauteurs dynamique. Notre méthode enrichit la méthode d'exponentiation des harmoniques sphériques, jusque là limitée aux bloqueurs sphériques, pour pouvoir traiter des champs de hauteurs. Nous ajoutons également une représentation pour les BRDFs diffuses et glossy. Nous pouvons ainsi combiner les visibilités et BRDFs dans un même espace, afin de calculer efficacement les ombres douces et les réflexions de scènes complexes. Un algorithme hybride, qui associe les visibilités en espace écran et en espace objet, permet de découpler la complexité des ombres de la complexité de la scène.

The reduced modulus in the analysis of sharp instrumented indentation tests

In the analysis of instrumented indentation data, it is common practice to incorporate the combined moduli of the indenter (E-i) and the specimen (E) in the so-called reduced modulus (E-r) to account for indenter deformation. Although indenter systems with rigid or elastic tips are considered as equivalent if E-r is the same, the validity of this practice has been questioned over the years. The present work uses systematic finite element simulations to examine the role of the elastic deformation of the indenter tip in instrumented indentation measurements and the validity of the concept of the reduced modulus in conical and pyramidal (Berkovich) indentations. It is found that the apical angle increases as a result of the indenter deformation, which influences in the analysis of the results. Based upon the inaccuracies introduced by the reduced modulus approximation in the analysis of the unloading segment of instrumented indentation applied load (P)-penetration depth (delta) curves, a detailed examination is then conducted on the role of indenter deformation upon the dimensionless functions describing the loading stages of such curves. Consequences of the present results in the extraction of the uniaxial stress-strain characteristics of the indented material through such dimensional analyses are finally illustrated. It is found that large overestimations in the assessment of the strain hardening behavior result by neglecting tip compliance. Guidelines are given in the paper to reduce such overestimations.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.

Contribución al desarrollo de técnicas CAD para el diseño de antenas impresas y dispositivos pasivos de microondas basadas en el método de elementos finitos

El objetivo principal de esta tesis es el desarrollo de herramientas numéricas basadas en técnicas de onda completa para el diseño asistido por ordenador (Computer-Aided Design,‘CAD’) de dispositivos de microondas. En este contexto, se desarrolla una herramienta numérica basada en el método de los elementos finitos para el diseño y análisis de antenas impresas mediante algoritmos de optimización. Esta técnica consiste en dividir el análisis de una antena en dos partes. Una parte de análisis 3D que se realiza sólo una vez en cada punto de frecuencia de la banda de funcionamiento donde se sustituye una superficie que contiene la metalización del parche por puertas artificiales. En una segunda parte se inserta entre las puertas artificiales en la estructura 3D la superficie soportando una metalización y se procede un análisis 2D para caracterizar el comportamiento de la antena. La técnica propuesta en esta tesis se puede implementar en un algoritmo de optimización para definir el perfil de la antena que permite conseguir los objetivos del diseño. Se valida experimentalmente dicha técnica empleándola en el diseño de antenas impresas de banda ancha para diferentes aplicaciones mediante la optimización del perfil de los parches. También, se desarrolla en esta tesis un procedimiento basado en el método de descomposición de dominio y el método de los elementos finitos para el diseño de dispositivos pasivos de microonda. Se utiliza este procedimiento en particular para el diseño y sintonía de filtros de microondas. En la primera etapa de su aplicación se divide la estructura que se quiere analizar en subdominios aplicando el método de descomposición de dominio, este proceso permite analizar cada segmento por separado utilizando el método de análisis adecuado dado que suele haber subdominios que se pueden analizar mediante métodos analíticos por lo que el tiempo de análisis es más reducido. Se utilizan métodos numéricos para analizar los subdominios que no se pueden analizar mediante métodos analíticos. En esta tesis, se utiliza el método de los elementos finitos para llevar a cabo el análisis. Además de la descomposición de dominio, se aplica un proceso de barrido en frecuencia para reducir los tiempos del análisis. Como método de orden reducido se utiliza la técnica de bases reducidas. Se ha utilizado este procedimiento para diseñar y sintonizar varios ejemplos de filtros con el fin de comprobar la validez de dicho procedimiento. Los resultados obtenidos demuestran la utilidad de este procedimiento y confirman su rigurosidad, precisión y eficiencia en el diseño de filtros de microondas. ABSTRACT The main objective of this thesis is the development of numerical tools based on full-wave techniques for computer-aided design ‘CAD’ of microwave devices. In this context, a numerical technique based on the finite element method ‘FEM’ for the design and analysis of printed antennas using optimization algorithms has been developed. The proposed technique consists in dividing the analysis of the antenna in two stages. In the first stage, the regions of the antenna which do not need to be modified during the CAD process are initially characterized only once from their corresponding matrix transfer function (Generalized Admittance matrix, ‘GAM’). The regions which will be modified are defined as artificial ports, precisely the regions which will contain the conducting surfaces of the printed antenna. In a second stage, the contour shape of the conducting surfaces of the printed antenna is iteratively modified in order to achieve a desired electromagnetic performance of the antenna. In this way, a new GAM of the radiating device which takes into account each printed antenna shape is computed after each iteration. The proposed technique can be implemented with a genetic algorithm to achieve the design objectives. This technique is validated experimentally and applied to the design of wideband printed antennas for different applications by optimizing the shape of the radiating device. In addition, a procedure based on the domain decomposition method and the finite element method has been developed for the design of microwave passive devices. In particular, this procedure can be applied to the design and tune of microwave filters. In the first stage of its implementation, the structure to be analyzed is divided into subdomains using the domain decomposition method; this process allows each subdomains can be analyzed separately using suitable analysis method, since there is usually subdomains that can be analyzed by analytical methods so that the time of analysis is reduced. For analyzing the subdomains that cannot be analyzed by analytical methods, we use the numerical methods. In this thesis, the FEM is used to carry out the analysis. Furthermore the decomposition of the domain, a frequency sweep process is applied to reduce analysis times. The reduced order model as the reduced basis technique is used in this procedure. This procedure is applied to the design and tune of several examples of microwave filters in order to check its validity. The obtained results allow concluding the usefulness of this procedure and confirming their thoroughness, accuracy and efficiency for the design of microwave filters.

Projected sequential Gaussian processes: flexible interpolation for large data sets

Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential framework for inference in such projected processes is presented, where the observations are considered one at a time. We introduce a C++ library for carrying out such projected, sequential estimation which adds several novel features. In particular we have incorporated the ability to use a generic observation operator, or sensor model, to permit data fusion. We can also cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the variogram parameters is based on maximum likelihood estimation. We illustrate the projected sequential method in application to synthetic and real data sets. We discuss the software implementation and suggest possible future extensions.

Projected sequential Gaussian processes:a C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed. © 2010 Elsevier Ltd.

A radial basis function network (RBFN) for function approximation

A radial basis function network (RBFN) circuit for function approximation is presented. Simulation and experimental results show that the network has good approximation capabilities. The RBFN was a squared hyperbolic secant with three adjustable parameters amplitude, width and center. To test the network a sinusoidal and sine function,vas approximated.

A new variational radial basis function approximation for inference in multivariate diffusions

In this paper we present a radial basis function based extension to a recently proposed variational algorithm for approximate inference for diffusion processes. Inference, for state and in particular (hyper-) parameters, in diffusion processes is a challenging and crucial task. We show that the new radial basis function approximation based algorithm converges to the original algorithm and has beneficial characteristics when estimating (hyper-)parameters. We validate our new approach on a nonlinear double well potential dynamical system.

Molecular basis for reduced estrone sulfate transport and altered modulator sensitivity of transmembrane helix (TM) 6 and TM17 mutants of multidrug resistance protein 1 (ABCC1)

Multidrug resistance protein 1 (MRP1) confers drug resistance and also mediates cellular efflux of many organic anions. MRP1 also transports glutathione (GSH); furthermore, this tripeptide stimulates transport of several substrates, including estrone 3-sulfate. We have previously shown that mutations of Lys(332) in transmembrane helix (TM) 6 and Trp(1246) in TM17 cause different substrate-selective losses in MRP1 transport activity. Here we have extended our characterization of mutants K332L and W1246C to further define the different roles these two residues play in determining the substrate and inhibitor specificity of MRP1. Thus, we have shown that TM17-Trp(1246) is crucial for conferring drug resistance and for binding and transport of methotrexate, estradiol glucuronide, and estrone 3-sulfate, as well as for binding of the tricyclic isoxazole inhibitor N-[3-(9-chloro-3-methyl-4-oxo-4H-isoxazolo-[4,3-c]quinolin-5-yl)-cyclohexylmethyl]-benzamide (LY465803). In contrast, TM6-Lys(332) is important for enabling GSH and GSH-containing compounds to serve as substrates (e.g., leukotriene C(4)) or modulators (e.g., S-decyl-GSH, GSH disulfide) of MRP1 and, further, for enabling GSH (or S-methyl-GSH) to enhance the transport of estrone 3-sulfate and increase the inhibitory potency of LY465803. On the other hand, both mutants are as sensitive as wild-type MRP1 to the non-GSH-containing inhibitors (E)-3-[[[3-[2-(7-chloro-2-quinolinyl)ethenyl]phenyl][[3-(dimethylamino)-3-oxopropyl]thio]methyl]thio]-propanoic acid (MK571), 1-[2-hydroxy-3-propyl-4-[4-(1H-tetrazol-5-yl)butoxy]phenyl]-ethanone (LY171883), and highly potent 6-[4'-carboxyphenylthio]-5[S]-hydroxy-7[E], 11[Z]14[Z]-eicosatetrenoic acid (BAY u9773). Finally, the differing abilities of the cysteinyl leukotriene derivatives leukotriene C(4), D(4), and F(4) to inhibit estradiol glucuronide transport by wild-type and K332L mutant MRP1 provide further evidence that TM6-Lys(332) is involved in the recognition of the gamma-Glu portion of substrates and modulators containing GSH or GSH-like moieties.

Changes In Motor Behavior During Pregnancy In Rats: The Basis For A Possible Animal Model Of Restless Legs Syndrome.

Pregnant women have a 2-3 fold higher probability of developing restless legs syndrome (RLS - sleep-related movement disorders) than general population. This study aims to evaluate the behavior and locomotion of rats during pregnancy in order to verify if part of these animals exhibit some RLS-like features. We used 14 female 80-day-old Wistar rats that weighed between 200 and 250 g. The rats were distributed into control (CTRL) and pregnant (PN) groups. After a baseline evaluation of their behavior and locomotor activity in an open-field environment, the PN group was inducted into pregnancy, and their behavior and locomotor activity were evaluated on days 3, 10 and 19 of pregnancy and in the post-lactation period in parallel with the CTRL group. The serum iron and transferrin levels in the CTRL and PN groups were analyzed in blood collected after euthanasia by decapitation. There were no significant differences in the total ambulation, grooming events, fecal boli or urine pools between the CTRL and PN groups. However, the PN group exhibited fewer rearing events, increased grooming time and reduced immobilization time than the CTRL group (ANOVA, p<0.05). These results suggest that pregnant rats show behavioral and locomotor alterations similar to those observed in animal models of RLS, demonstrating to be a possible animal model of this sleep disorder.