Global ETD Search

1	Robust A-optimal designs for mixture experiments in Scheffe' models Chou, Chao-Jin 28 July 2003 (has links) A mixture experiment is an experiments in which the q-ingredients are nonnegative and subject to the simplex restriction on the (q-1)-dimentional probability simplex. In this work , we investigate the robust A-optimal designs for mixture experiments with uncertainty on the linear, quadratic models considered by Scheffe' (1958). In Chan (2000), a review on the optimal designs including A-optimal designs are presented for each of the Scheffe's linear and quadratic models. We will use these results to find the robust A-optimal design for the linear and quadratic models under some robust A-criteria. It is shown with the two types of robust A-criteria defined here, there exists a convex combination of the individual A-optimal designs for linear and quadratic models respectively to be robust A-optimal. In the end, we compare efficiencies of these optimal designs with respect to different A-criteria. equivalence theorem invariant symmetric block matrices robust design. Convex combination
2	Algorithms for a Partially Regularized Least Squares Problem Skoglund, Ingegerd January 2007 (has links) <p>Vid analys av vattenprover tagna från t.ex. ett vattendrag betäms halten av olika ämnen. Dessa halter är ofta beroende av vattenföringen. Det är av intresse att ta reda på om observerade förändringar i halterna beror på naturliga variationer eller är orsakade av andra faktorer. För att undersöka detta har föreslagits en statistisk tidsseriemodell som innehåller okända parametrar. Modellen anpassas till uppmätta data vilket leder till ett underbestämt ekvationssystem. I avhandlingen studeras bl.a. olika sätt att säkerställa en unik och rimlig lösning. Grundidén är att införa vissa tilläggsvillkor på de sökta parametrarna. I den studerade modellen kan man t.ex. kräva att vissa parametrar inte varierar kraftigt med tiden men tillåter årstidsvariationer. Det görs genom att dessa parametrar i modellen regulariseras.</p><p>Detta ger upphov till ett minsta kvadratproblem med en eller två regulariseringsparametrar. I och med att inte alla ingående parametrar regulariseras får vi dessutom ett partiellt regulariserat minsta kvadratproblem. I allmänhet känner man inte värden på regulariseringsparametrarna utan problemet kan behöva lösas med flera olika värden på dessa för att få en rimlig lösning. I avhandlingen studeras hur detta problem kan lösas numeriskt med i huvudsak två olika metoder, en iterativ och en direkt metod. Dessutom studeras några sätt att bestämma lämpliga värden på regulariseringsparametrarna.</p><p>I en iterativ lösningsmetod förbättras stegvis en given begynnelseapproximation tills ett lämpligt valt stoppkriterium blir uppfyllt. Vi använder här konjugerade gradientmetoden med speciellt konstruerade prekonditionerare. Antalet iterationer som krävs för att lösa problemet utan prekonditionering och med prekonditionering jämförs både teoretiskt och praktiskt. Metoden undersöks här endast med samma värde på de två regulariseringsparametrarna.</p><p>I den direkta metoden används QR-faktorisering för att lösa minsta kvadratproblemet. Idén är att först utföra de beräkningar som kan göras oberoende av regulariseringsparametrarna samtidigt som hänsyn tas till problemets speciella struktur.</p><p>För att bestämma värden på regulariseringsparametrarna generaliseras Reinsch’s etod till fallet med två parametrar. Även generaliserad korsvalidering och en mindre beräkningstung Monte Carlo-metod undersöks.</p> / <p>Statistical analysis of data from rivers deals with time series which are dependent, e.g., on climatic and seasonal factors. For example, it is a well-known fact that the load of substances in rivers can be strongly dependent on the runoff. It is of interest to find out whether observed changes in riverine loads are due only to natural variation or caused by other factors. Semi-parametric models have been proposed for estimation of time-varying linear relationships between runoff and riverine loads of substances. The aim of this work is to study some numerical methods for solving the linear least squares problem which arises.</p><p>The model gives a linear system of the form <em>A</em><em>1x1</em><em> + A</em><em>2x2</em><em> + n = b</em><em>1</em>. The vector <em>n</em> consists of identically distributed random variables all with mean zero. The unknowns, <em>x,</em> are split into two groups, <em>x</em><em>1</em><em> </em>and <em>x</em><em>2</em><em>.</em> In this model, usually there are more unknowns than observations and the resulting linear system is most often consistent having an infinite number of solutions. Hence some constraint on the parameter vector x is needed. One possibility is to avoid rapid variation in, e.g., the parameters<em> x</em><em>2</em><em>.</em> This can be accomplished by regularizing using a matrix <em>A</em><em>3</em>, which is a discretization of some norm. The problem is formulated</p><p>as a partially regularized least squares problem with one or two regularization parameters. The parameter <em>x</em><em>2</em> has here a two-dimensional structure. By using two different regularization parameters it is possible to regularize separately in each dimension.</p><p>We first study (for the case of one parameter only) the conjugate gradient method for solution of the problem. To improve rate of convergence blockpreconditioners of Schur complement type are suggested, analyzed and tested. Also a direct solution method based on QR decomposition is studied. The idea is to first perform operations independent of the values of the regularization parameters. Here we utilize the special block-structure of the problem. We further discuss the choice of regularization parameters and generalize in particular Reinsch’s method to the case with two parameters. Finally the cross-validation technique is treated. Here also a Monte Carlo method is used by which an approximation to the generalized cross-validation function can be computed efficiently.</p> Least squares Regularization Block-matrices Conjugate gradient QR-factorization Cross-validation Numerical analysis Numerisk analys
3	Algorithms for a Partially Regularized Least Squares Problem Skoglund, Ingegerd January 2007 (has links) Vid analys av vattenprover tagna från t.ex. ett vattendrag betäms halten av olika ämnen. Dessa halter är ofta beroende av vattenföringen. Det är av intresse att ta reda på om observerade förändringar i halterna beror på naturliga variationer eller är orsakade av andra faktorer. För att undersöka detta har föreslagits en statistisk tidsseriemodell som innehåller okända parametrar. Modellen anpassas till uppmätta data vilket leder till ett underbestämt ekvationssystem. I avhandlingen studeras bl.a. olika sätt att säkerställa en unik och rimlig lösning. Grundidén är att införa vissa tilläggsvillkor på de sökta parametrarna. I den studerade modellen kan man t.ex. kräva att vissa parametrar inte varierar kraftigt med tiden men tillåter årstidsvariationer. Det görs genom att dessa parametrar i modellen regulariseras. Detta ger upphov till ett minsta kvadratproblem med en eller två regulariseringsparametrar. I och med att inte alla ingående parametrar regulariseras får vi dessutom ett partiellt regulariserat minsta kvadratproblem. I allmänhet känner man inte värden på regulariseringsparametrarna utan problemet kan behöva lösas med flera olika värden på dessa för att få en rimlig lösning. I avhandlingen studeras hur detta problem kan lösas numeriskt med i huvudsak två olika metoder, en iterativ och en direkt metod. Dessutom studeras några sätt att bestämma lämpliga värden på regulariseringsparametrarna. I en iterativ lösningsmetod förbättras stegvis en given begynnelseapproximation tills ett lämpligt valt stoppkriterium blir uppfyllt. Vi använder här konjugerade gradientmetoden med speciellt konstruerade prekonditionerare. Antalet iterationer som krävs för att lösa problemet utan prekonditionering och med prekonditionering jämförs både teoretiskt och praktiskt. Metoden undersöks här endast med samma värde på de två regulariseringsparametrarna. I den direkta metoden används QR-faktorisering för att lösa minsta kvadratproblemet. Idén är att först utföra de beräkningar som kan göras oberoende av regulariseringsparametrarna samtidigt som hänsyn tas till problemets speciella struktur. För att bestämma värden på regulariseringsparametrarna generaliseras Reinsch’s etod till fallet med två parametrar. Även generaliserad korsvalidering och en mindre beräkningstung Monte Carlo-metod undersöks. / Statistical analysis of data from rivers deals with time series which are dependent, e.g., on climatic and seasonal factors. For example, it is a well-known fact that the load of substances in rivers can be strongly dependent on the runoff. It is of interest to find out whether observed changes in riverine loads are due only to natural variation or caused by other factors. Semi-parametric models have been proposed for estimation of time-varying linear relationships between runoff and riverine loads of substances. The aim of this work is to study some numerical methods for solving the linear least squares problem which arises. The model gives a linear system of the form A1x1 + A2x2 + n = b1. The vector n consists of identically distributed random variables all with mean zero. The unknowns, x, are split into two groups, x1 and x2. In this model, usually there are more unknowns than observations and the resulting linear system is most often consistent having an infinite number of solutions. Hence some constraint on the parameter vector x is needed. One possibility is to avoid rapid variation in, e.g., the parameters x2. This can be accomplished by regularizing using a matrix A3, which is a discretization of some norm. The problem is formulated as a partially regularized least squares problem with one or two regularization parameters. The parameter x2 has here a two-dimensional structure. By using two different regularization parameters it is possible to regularize separately in each dimension. We first study (for the case of one parameter only) the conjugate gradient method for solution of the problem. To improve rate of convergence blockpreconditioners of Schur complement type are suggested, analyzed and tested. Also a direct solution method based on QR decomposition is studied. The idea is to first perform operations independent of the values of the regularization parameters. Here we utilize the special block-structure of the problem. We further discuss the choice of regularization parameters and generalize in particular Reinsch’s method to the case with two parameters. Finally the cross-validation technique is treated. Here also a Monte Carlo method is used by which an approximation to the generalized cross-validation function can be computed efficiently. Least squares Regularization Block-matrices Conjugate gradient QR-factorization Cross-validation Numerical analysis Numerisk analys
4	Robust D-optimal designs for mixture experiments in Scheffe models Hsu, Hsiang-Ling 10 July 2003 (has links) A mixture experiment is an experiment in which the q-ingredients {xi,i=1,...,q} are nonnegative and subject to the simplex restriction sum_{i=1}^q x_i=1 on the (q-1)-dimensional probability simplex S^{q-1}. In this work, we investigate the robust D-optimal designs for mixture experiments with consideration on uncertainties in the Scheffe's linear, quadratic and cubic model without 3-way effects. The D-optimal designs for each of the Scheffe's models are used to find the robust D-optimal designs. With uncertianties on the Scheffe's linear and quadratic models, the optimal convex combination of the two model's D-optimal designs can be proved to be a robust D-optimal design. For the case of the Scheffe's linear and cubic model without 3-way effects, we have some numerical results about the robust D-optimal designs, as well as that for Scheffe's linear, quadratic and cubic model without 3-way effects. Ultimately, we discuss the efficiency of a maxmin type criterion D_r under given the robust D-optimal designs for the Scheffe's linear and quadratic models. invariant symmetric block matrices Complete class Dr-efficiency equivalence theorem convex combination
5	Optimal Designs for Log Contrast Models in Experiments with Mixtures Huang, Miao-kuan 05 February 2009 (has links) A mixture experiment is an experiment in which the k ingredients are nonnegative and subject to the simplex restriction £Ux_i=1 on the (k-1)-dimensional probability simplex S^{k-1}. This dissertation discusses optimal designs for linear and quadratic log contrast models for experiments with mixtures suggested by Aitchison and Bacon-Shone (1984), where the experimental domain is restricted further as in Chan (1992). In this study, firstly, an essentially complete class of designs under the Kiefer ordering for linear log contrast models with mixture experiments is presented. Based on the completeness result, £X_p-optimal designs for all p, -¡Û<p≤1 including D- and A-optimal are obtained, where the eigenvalues of the design moment matrix are used. By using the approach presented here, we gain insight on how these £X_p-optimal designs behave. Following that, the exact N-point D-optimal designs for linear log contrast models with three and four ingredients are further investigated. The results show that for k=3 and N=3p+q ,1 ≤q≤2, there is an exact N-point D-optimal design supported at the points of S_1 (S_2) with equal weight n/N, 0≤n≤p , and puts the remaining weight (N-3n)/N uniformly on the points of S_2 (S_1) as evenly as possible, where S_1 and S_2 are sets of the supports of the approximate D-optimal designs. When k=4 and N=6p+q , 1 ≤q≤5, an exact N-point design which distributes the weights as evenly as possible among the supports of the approximate D-optimal design is proved to be exact D-optimal. Thirdly, the approximate D_s-optimal designs for discriminating between linear and quadratic log contrast models for experiments with mixtures are derived. It is shown that for a symmetric subspace of the finite dimensional simplex, there is a D_s-optimal design with the nice structure that puts a weight 1/(2^{k-1}) on the centroid of this subspace and the remaining weight is uniformly distributed on the vertices of the experimental domain. Moreover, the D_s-efficiency of the D-optimal design for quadratic model and the design given by Aitchison and Bacon-Shone (1984) are also discussed Finally, we show that an essentially complete class of designs under the Kiefer ordering for the quadratic log contrast model is the set of all designs in the boundary of T or origin of T . Based on the completeness result, numerical £X_p -optimal designs for some p, -¡Û<p≤1 are obtained. Equivalence theorem Geometric-arithmetic means inequality Invariant symmetric block matrices Kiefer ordering Lagrange interpolation polynomial Loewner ordering
6	On Numerical Solution Methods for Block-Structured Discrete Systems Boyanova, Petia January 2012 (has links) The development, analysis, and implementation of efficient methods to solve algebraic systems of equations are main research directions in the field of numerical simulation and are the focus of this thesis. Due to their lesser demands for computer resources, iterative solution methods are the choice to make, when very large scale simulations have to be performed. To improve their efficiency, iterative methods are combined with proper techniques to accelerate convergence. A general technique to do this is to use a so-called preconditioner. Constructing and analysing various preconditioning methods has been an active field of research already for decades. Special attention is devoted to the class of the so-called optimal order preconditioners, that possess both optimal convergence rate and optimal computational complexity. The preconditioning techniques, proposed and studied in this thesis, utilise the block structure of the underlying matrices, and lead to methods that are of optimal order. In the first part of the thesis, we construct an Algebraic MultiLevel Iteration (AMLI) method for systems arising from discretizations of parabolic problems, using Crouzeix-Raviart finite elements. The developed AMLI method is based on an approximated block factorization of the original system matrix, where the partitioning is associated with a sequence of nested discretization meshes. In the second part of the thesis we develop solution methods for the numerical simulation of multiphase flow problems, modelled by the Cahn-Hilliard (C-H) equation. We consider the discrete C-H problem, obtained via finite element discretization in space and implicit schemes in time. We propose techniques to precondition the Jacobian of the discrete nonlinear system, based on its natural two-by-two block structure. The preconditioners are used in the framework of inexact Newton methods. We develop two nonlinear solution algorithms for the Cahn-Hilliard problem. Both lead to efficient optimal order methods. One of the main advantages of the proposed methods is that they are implemented using available software toolboxes for both sequential and distributed execution. The theoretical analysis of the solution methods presented in this thesis is combined with numerical studies that confirm their efficiency. Preconditioning techniques Finite element method Two-by-two block matrices Optimal order methods AMLI method Cahn-Hilliard equation Multiphase flow Inexact Newton method
7	Typical representations for GL_n(F) / Représentations typiques pour GL_n(F) Nadimpalli, Santosh VRN 16 June 2015 (has links) Dans cette thèse, nous classifions représentations typiques pour certaines composants Bernstein. Suite aux travaux de Henniart dans le cas de GL_2(F) et Paskunas pour les composants cuspidales, nous classifions représentations typiques pour les composants de niveau zéro pour GL_n(F) pour n> 2, composants de série principale, composants avec Levi sous-groupe de la forme (n, 1) pour n>1 et certains composants avec sous-groupe de Levi de la forme (2,2). Chacun des composants ci-dessus est traité dans un chapitre distinct. La classification utilise la théorie des types développés par Bushnell-Kutzko d'une manière significative. Nous allons donner la classification en termes de types de Bushnell-Kutzko. / In this thesis we classify typical representations for certain non-cuspidal Bernstein components. Following the work of Henniart in the case of GL_2(F) and Paskunas for the cuspidal components, we classify typical representations for of level-zero components for GL_n(F) for n>2, principal series components, components with Levi subgroup of the form (n, 1) for n>1 and certain components with Levi subgroup of the form (2,2). Each of the above component is treated in a separate chapter. The classification uses the theory of types developed by Bushnell-Kutzko in a significant way. We will give the classification in terms of Bushnell-Kutzko types for a given inertial class. Représentations typiques Types de Bushnell-Kutzko Représentations lisse Type simple Opérateurs d’entrelacement Matrices par bloc Sous-groupes paraboliques Typical representations Bushnell-Kutzko types Smooth representations Simple type Intertwining operators Block matrices Parabolic sub-groups
8	Generalizovana dijagonalna dominacija za blok matrice i mogućnosti njene primene / Generalized diagonal dominance for block matrices and possibilites of its application Doroslovački Ksenija 06 May 2014 (has links) <p>Ova doktorska disertacija izučava matrice zapisane u blok formi. Ona<br />sistematizuje postojeća i predstavlja nova tvrđenja o osobinama takvih matrica,<br />koja se baziraju na ideji generalizovane dijagonalne dominacije. Poznati<br />rezultati u tačkastom slučaju dobra su osnova za blok generalizacije, koje su<br />izvedene na dva različita načina, prvi zbog svoje jednostavnije primenljivosti,<br />a drugi zbog obuhvatanja šire klase matrica na koju se rezultati odnose.</p> / <p>This thesis is related to matrices written in their block form. It systematizes known and<br />represents new knowledge about properties of such matrices, which is based on the idea<br />of generalized diagonal dominance. Known results in the point case serve as a good basis<br />for block generalization, which is done in two different ways, the first one because of its<br />simple usability, and the other for capturing wider class of matrices which are treated.</p>
9	Methodenentwicklung zur Simulation von Strömungen mit freier Oberfläche unter dem Einfluss elektromagnetischer Wechselfelder Beckstein, Pascal 16 February 2018 (has links) (PDF) Im Bereich der industriellen Metallurgie und Kristallzüchtung treten bei zahlreichen Anwendungen, wo magnetische Wechselfelder zur induktiven Beeinflussung von leitfähigen Werkstoffen eingesetzt werden, auch Strömungen mit freier Oberfläche auf. Das Anwendungsspektrum reicht dabei vom einfachen Aufschmelzen eines Metalls in einem offenen Tiegel bis hin zur vollständigen Levitation. Auch der sogenannte RGS-Prozess, ein substratbasiertes Kristallisationsverfahren zur Herstellung siliziumbasierter Dünnschichtmaterialien, ist dafür ein Beispiel. Um bei solchen Prozessen die Interaktion von Magnetfeld und Strömung zu untersuchen, ist die numerische Simulationen ein wertvolles Hilfsmittel. Für beliebige dreidimensionale Probleme werden entsprechende Berechnungen bisher durch eine externe Kopplung kommerzieller Programme realisiert, die für Magnetfeld und Strömung jeweils unterschiedliche numerische Techniken nutzen. Diese Vorgehensweise ist jedoch im Allgemeinen mit unnötigem Rechenaufwand verbunden. In dieser Arbeit wird ein neu entwickelter Methodenapparat auf Basis der FVM vorgestellt, mit welchem sich diese Art von Berechnungen effizient durchführen lassen. Mit der Implementierung dieser Methoden in foam-extend, einer erweiterten Version der quelloffenen Software OpenFOAM, ist daraus ein leistungsfähiges Werkzeug in Form einer freien Simulationsplattform entstanden, welches sich durch einen modularen Aufbau leicht erweitern lässt. Mit dieser Plattform wurden in foam-extend auch erstmalig dreidimensionale Induktionsprozesse im Frequenzraum gelöst. Metallurgie Kristallzüchtung Elektroprozesstechnik elektromagnetisches Prozessieren Optimierung Simulation RGS Ribbon-Growth on Substrate freie Oberfläche Interface-Tracking Induktion metallurgische MHD MHD Magnetodynamik Hydrodynamik Thermodynamik Magnetohydrodynamik OpenFOAM foam-extend Methodenentwicklung Softwareplattform Numerische Methoden Strömung mit freier Oberfläche Metalle Halbmetalle Silizium Wafer 3D METIS FVM Finite-Volumen-Methode elektromagnetische Wechselfelder zeitharmonische Anregung Frequenzraum Blockmatrizen diskontinuierliche Materialeigenschaften magnetische Permeabilität elektrische Leitfähigkeit eingebettete Diskretisierung GFM Ghost-Fluid-Methode Mehrfeldproblem Multi-Physics Biot-Savart Biot-Savart-Gesetz Biot-Savart-Induktoren überlagerte Gitter ALE Gitterbewegung Kopplung FAM Finite-Flächen-Methode RMF rotierendes Magnetfeld TMF wanderndes Magnetfeld Validierung Maxwell Maxwell-Gleichungen Navier-Stokes Navier-Stokes-Gleichung inkompressibel Schmelze Lorentz-Kraft Induktor Konduktor Induktionsspulen Wirbelströme Wirbelstromproblem PISO Oberflächenspannung semi-gekoppelter Lösungsalgorithmus Diskretisierung Skineffekt AC quasistatische Approximation Magneto-Quasistatik metallurgy crystal growth electromagnetic processing optimization Simulation RGS Ribbon-Growth on Substrate free-surface interface-tracking Induction metallurgical MHD MHD magnetodynamics hydrodynamics thermodynamics magnetohydrodynamics OpenFOAM foam-extend method development software platform numerical methods free-surface flow metals metalloids silicon wafer 3D METIS FVM Finite-Volumen-Method oscillating magnetic fields time-harmonic excitation frequency domain block-matrices discontinuous material properties magnetic permeability electrical conductivity embedded discretization GFM Ghost-Fluid-Method multi-physics Biot-Savart Biot-Savart-inductor overlapped meshes ALE mesh movement coupling FAM Finite-Area-Method RMF rotating magnetic field TMF traveling magnetic field validation Maxwell Maxwell-equations Navier-Stokes Navier-Stokes-equation incompressible melt melt flow Lorentz-force inductor conductor induction coil eddy-currents eddy-current problem PISO surface-tension semi-coupled solution discretization skin effect AC quasi-static approximation ddc:620 rvk:UF 4600
10	Methodenentwicklung zur Simulation von Strömungen mit freier Oberfläche unter dem Einfluss elektromagnetischer Wechselfelder Beckstein, Pascal 08 January 2018 (has links) Im Bereich der industriellen Metallurgie und Kristallzüchtung treten bei zahlreichen Anwendungen, wo magnetische Wechselfelder zur induktiven Beeinflussung von leitfähigen Werkstoffen eingesetzt werden, auch Strömungen mit freier Oberfläche auf. Das Anwendungsspektrum reicht dabei vom einfachen Aufschmelzen eines Metalls in einem offenen Tiegel bis hin zur vollständigen Levitation. Auch der sogenannte RGS-Prozess, ein substratbasiertes Kristallisationsverfahren zur Herstellung siliziumbasierter Dünnschichtmaterialien, ist dafür ein Beispiel. Um bei solchen Prozessen die Interaktion von Magnetfeld und Strömung zu untersuchen, ist die numerische Simulationen ein wertvolles Hilfsmittel. Für beliebige dreidimensionale Probleme werden entsprechende Berechnungen bisher durch eine externe Kopplung kommerzieller Programme realisiert, die für Magnetfeld und Strömung jeweils unterschiedliche numerische Techniken nutzen. Diese Vorgehensweise ist jedoch im Allgemeinen mit unnötigem Rechenaufwand verbunden. In dieser Arbeit wird ein neu entwickelter Methodenapparat auf Basis der FVM vorgestellt, mit welchem sich diese Art von Berechnungen effizient durchführen lassen. Mit der Implementierung dieser Methoden in foam-extend, einer erweiterten Version der quelloffenen Software OpenFOAM, ist daraus ein leistungsfähiges Werkzeug in Form einer freien Simulationsplattform entstanden, welches sich durch einen modularen Aufbau leicht erweitern lässt. Mit dieser Plattform wurden in foam-extend auch erstmalig dreidimensionale Induktionsprozesse im Frequenzraum gelöst. info:eu-repo/classification/ddc/620 ddc:620

Search results