Дисертації з теми "Méthode de Newton régularisé"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "Méthode de Newton régularisé".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
Nguyen, Van Vu. "Méthode de Newton revisitée pour les équations généralisées." Thesis, Limoges, 2016. http://www.theses.fr/2016LIMO0066.
Повний текст джерелаThis thesis is devoted to present some results in the scope of Newton-type methods applied for inclusion involving set-valued mappings. In the first part, we follow the Kantorovich's and/or Smale's approaches to study the convergence of Josephy-Newton method for generalized equation (GE) in Banach spaces. Such results can be viewed as an extension of the classical Kantorovich's theorem as well as Smale's (alpha, gamma)-theory which were stated for nonlinear equations. The second part develops an algorithm using set-valued differentiation in order to solve GE. We proved that, under some suitable conditions imposed on the input data and the choice of the starting point, the algorithm produces a sequence converging at least linearly to a solution of considering GE. Moreover, by imposing some stronger assumptions related to the approximation of set-valued part, the proposed method converges locally superlinearly. The last part deals with inclusions involving maps defined on Riemannian manifolds whose values belong to an Euclidean space. Using the relationship between the geometric structure of manifolds and the retraction maps, we show that, our scheme converges locally superlinearly to a solution of the initial problem. With some more regularity assumptions on the data involved in the problem, the quadratic convergence (local and semi-local) can be ensured
Abbas, Boushra. "Méthode de Newton régularisée pour les inclusions monotones structurées : étude des dynamiques et algorithmes associés." Thesis, Montpellier, 2015. http://www.theses.fr/2015MONTS250/document.
Повний текст джерелаThis thesis is devoted to finding zeroes of structured maximal monotone operators, by using discrete and continuous dissipative dynamical systems. The solutions are obtained as the limits of trajectories when the time t tends towards infinity.We pay special attention to the dynamics that are obtained by Levenberg-Marquardt regularization of Newton's method. We also revisit the approaches based on some related dynamical systems.In a Hilbert framework, we are interested in finding zeroes of a structured maximal monotone operator M = A + B, where A is a general maximal monotone operator, and B is monotone and locally Lipschitz continuous. We introduce discrete and continuous dynamical systems which are linked to Newton's method. They involve separately B and the resolvents of A, and are designed to splitting methods. Based on the Minty representation of A as a Lipschitz manifold, we show that these dynamics can be formulated as differential systems, which are relevant to the Cauchy-Lipschitz theorem. We focus on the particular case where A is the subdifferential of a convex lower semicontinuous proper function, and B is the gradient of a convex, continuously differentiable function. We study the asymptotic behavior of trajectories. When the regularization parameter does not tend to zero too rapidly, and by using Lyapunov asymptotic analysis, we show the convergence of trajectories. Besides, we show the Lipschitz continuous dependence of the solution with respect to the regularization term.Then we extend our study by considering various classes of dynamical systems which aim at solving inclusions governed by structured monotone operators M = $partialPhi$+ B, where $partialPhi$ is the subdifferential of a convex lower semicontinuous function, and B is a monotone cocoercive operator. By a Lyapunov analysis, we show the convergence properties of the orbits of these systems. The time discretization of these dynamics gives various forward-backward splittingmethods (some new).Finally, we focus on the study of the asymptotic behavior of trajectories of the regularized Newton dynamics, in which we introduce an additional vanishing Tikhonov-like viscosity term.We thus obtain the asymptotic selection of the solution of minimal norm
Ferzly, Joëlle. "Adaptive inexact smoothing Newton method for nonlinear systems with complementarity constraints. Application to a compositional multiphase flow in porous media." Thesis, Sorbonne université, 2022. http://www.theses.fr/2022SORUS376.
Повний текст джерелаWe consider variational inequalities written in the form of partial differential equations with nonlinear complementarity constraints. The discretization of such problems leads to nonlinear non-differentiable discrete systems that can be solved employing an iterative linearization method of semismooth type like, e.g., the Newton-min algorithm. Our goal in this thesis is to conceive a simple smoothing approach that involves approximating the problem as a system of nonlinear smooth (differentiable) equations. In this setting, a direct application of classical Newton-type methods is possible. We construct a posteriori error estimates that lie at the foundation of an adaptive inexact smoothing Newton algorithm for a solution of the considered problems. We first present the strategy in a discrete framework. Then, we develop the method for the model problem of contact between two membranes. Last, an application to a compositional multiphase flow industrial model is introduced. In Chapter 1, we are concerned about nonlinear algebraic systems with complementarity constraints arising from numerical discretizations of PDEs with nonlinear complementarity problems. We produce a smooth approximation of a nonsmooth function, reformulating the complementarity conditions. The ensuing nonlinear system is solved employing the Newton method, together with an iterative linear algebraic solver to approximately solve the linear system. We establish an upper bound on the considered system’s residual and design a posteriori error estimators identifying the smoothing, linearization, and algebraic error components. These ingredients are used to formulate efficient stopping criteria for the nonlinear and algebraic solvers. With the same methodology, an adaptive interior-point method is proposed. We apply our algorithm to the algebraic system of variational inequalities describing the contact between two membranes and a two-phase flow problem. We provide numerical comparison of our approach with a semismooth Newton method, possibly combined with a path-following strategy, and a nonparametric interior-point method. In Chapter 2, in an infinite-dimensional framework, we consider as a model problem the contact problem between two membranes. We employ a finite volume discretization and apply the smoothing approach proposed in Chapter 1 to smooth the non-differentiability in the complementarity constraints. The resolution of the arising nonlinear smooth system is again realized thanks to the Newton method, in combination with an iterative algebraic solver for the solution of the resulting linear system. We design H1-conforming potential reconstructions as well as H(div)-conforming discrete equilibrated flux reconstructions. We prove an upper bound for the total error in the energy norm and conceive discretization, smoothing, linearization, and algebraic estimators reflecting the errors stemming from the finite volume discretization, the smoothing of the non-differentiability, the linearization by the Newton method, and the algebraic solver, respectively. This enables us to establish adaptive stopping criteria to stop the different solvers in the proposed algorithm and design adaptive algorithm steering all these four components. In Chapter 3, we consider a compositional multiphase flow (oil, gas, and water) with phase transitions in a porous media. A finite volume discretization yields a nonlinear non-differentiable algebraic system which we solve employing our inexact smoothing Newton technique. Following the process of Chapter 1, we build a posteriori estimators by bounding the norm of the discrete system’s residual, resulting in adaptive criteria that we incorporate in the employed algorithm. Throughout this thesis, numerical experiments confirm the efficiency of our estimates. In particular, we show that the developed adaptive algorithms considerably reduce the overall number of iterations in comparison with the existing methods
Boussandel, Sahbi. "Méthodes de résolution d'équations algébriques et d'évolution en dimension finie et infinie." Thesis, Metz, 2010. http://www.theses.fr/2010METZ027S/document.
Повний текст джерелаIn this work, we solve algebraic and evolution equations in finite and infinite-dimensional sapces. In the first chapter, we use the Galerkin method to study existence and maximal regularity of solutions of a gradient abstract system with applications to non-linear diffusion equations and to non-degenerate quasilinear parabolic equations with nonlocal coefficients. In the second chapter, we Study local existence, uniqueness and maximal regularity of solutions of the curve shortening flow equation by using the local inverse theorem. Finally, in the third chapter, we solve an algebraic equation between two Banach spaces by using the continuous Newton’s method and we apply this result to solve a non-linear ordinary differential equation with periodic boundary conditions
Sokol, Sergueï. "Approche multi-échelle pour appariement d'images par modèles élastiques." Toulouse, ENSAE, 1997. http://www.theses.fr/1997ESAE0020.
Повний текст джерелаDaridon, Loïc. "Une modélisation des ponts de fibres pour le délaminage des matériaux composites." Metz, 1993. http://docnum.univ-lorraine.fr/public/UPV-M/Theses/1993/Daridon.Loic.SMZ9323.pdf.
Повний текст джерелаIn this work, we model the delamination in the long fiber composite material. More precisely, we study a modelling of the fiber bridging contribution in the dissipative energy during delamination test in mode I, mode II, mixed mode. For this, we introduce a density of damageable spring along the crack, to simulate the fiber bridging. This approach permit us to keep simple propagation criteria. Then, we perform a numerical simulation of different tests, as the D. C. B. , the E. L. S. And the M. M. F. S. With the Newton-Raphson method. The result, we obtain, agree with the experimental observation
Cuvilliez, Sam. "Passage d'un modèle d'endommagement continu régularisé à un modèle de fissuration cohésive dans le cadre de la rupture quasi-fragile." Phd thesis, Ecole Nationale Supérieure des Mines de Paris, 2012. http://pastel.archives-ouvertes.fr/pastel-00817940.
Повний текст джерелаHama, Muzafar. "Matrix eigenvalues : localization through subsets and triangularization with Newton-like iterations." Saint-Etienne, 2009. http://www.theses.fr/2009STET4016.
Повний текст джерелаDans cette thèse on développe deux sujets : d'une part, la notion de pseudo-spectre d'une matrice complexe carrée et, d'autre part, l'application de la méthode de Newton-Kantorovich au calcul des formes triangulaires supérieures semblables d'une matrice carrée complexe. Le travail sur le pseudo-spectre est une compilation de résultats qui réalise une synthèse de cette notion. Quelques contributions originales complètent ce texte. La recherche sur les formes triangulaires repose sur l'application de la méthode de Newton-Kantorovich et sa variante de la Pente xe au calcul d'une forme de Schur par similarité unitaire et d'une forme de Gauss par similarité triangulaire infèrieure a diagonal unite. On propose que les factorisations QR et (L+I)U nécessaires au calcul des formes triangulaires de Schur et de Gauss par les algorithmes de Francis et de Crout respectivement, soient accomplies par ranement itératif en utilisant encore une fois la méthode de Newton-Kantorovich et sa variante de la Pente x
Choquet, Rémi. "Étude de la méthode de Newton-GMRES. Application aux équations de Navier-Stokes compressibles." Rennes 1, 1995. http://www.theses.fr/1995REN10147.
Повний текст джерелаCotte, Romain. "L'enjeu de la différentiation automatique dans les méthodes de Newton d'ordres supérieurs." Mémoire, Université de Sherbrooke, 2010. http://savoirs.usherbrooke.ca/handle/11143/4884.
Повний текст джерелаCuvilliez, Sam. "Passage d’un modèle d’endommagement continu régularisé à un modèle de fissuration cohésive dans le cadre de la rupture quasi-fragile." Thesis, Paris, ENMP, 2012. http://www.theses.fr/2012ENMP0064/document.
Повний текст джерелаThe present work deals with the study and the improvement of regularized (non local) damage models. It aims to study the transition from a continuous damage field distributed on a structure to a discontinuous macroscopic failure model.First, an analytical one-dimensional study is carried out (on a bar submitted to tensile loading) in order to identify a set of interface laws that enable to switch from an inhomogeneous solution obtained with a continuous gradient damage model to a cohesive zone model. This continuous / discontinuous transition is constructed so that the energetic equivalence between both models remains ensured whatever the damage level reached when switching.This strategy is then extended to the bi-dimensional (and tri-dimensional) case of rectilinear (and plane) crack propagation under mode I loading conditions, in a finite element framework. An explicit approach based on a critical damage criterion that allows coupling both continuous and discontinuous approaches is then proposed. Finally, results of several simulations led with this coupled approach are presented
Heyouni, Mohammed. "Méthode de Hessenberg généralisée et applications." Lille 1, 1996. http://www.theses.fr/1996LIL10171.
Повний текст джерелаMartin, Claire. "Adaptation de mailllages structurés par un modèle d'élasticité non linéaire : application aux équations de Navier-Stokes." Bordeaux 1, 1995. http://www.theses.fr/1995BOR10653.
Повний текст джерелаDobranszky, Gabriela. "Systèmes d'aide à l'inversion des modèles stratigraphiques." Nice, 2005. http://www.theses.fr/2005NICE4077.
Повний текст джерелаStratigraphic modeling aims at rebuilding the history of the sedimentary basins by simulating the processes of erosion, transport and deposit of sediments using physical models. The objective is to determine the location of the bedrocks likely to contain the organic matter, the location of the porous rocks that could trap the hydrocarbons during their migration and the location of the impermeable rocks likely to seal the reservoir. The model considered within this thesis is based on a multi-lithological diffusive transport model and applies to large scales of time and space. Due to the complexity of the phenomena and scales considered, none of the model parameters is directly measurable. Therefore it is essential to inverse them. The standard approach, which consists in inversing all the parameters by minimizing a cost function using a gradient method, proved very sensitive to the choice of the parameterization, to the weights given to the various terms of the cost function (bearing on data of very diverse nature) and to the numerical noise. These observations led us to give up this method and to carry out the inversion step by step by decoupling the parameters. This decoupling is not obtained by fixing the parameters but by making several assumptions on the model resulting in a range of reduced but relevant models. In this thesis, we show how these models enable us to inverse all the parameters in a robust and interactive way
Pivoteau, Carine. "Génération aléatoire de structures combinatoires : méthode de Boltzmann effective." Paris 6, 2008. http://www.theses.fr/2008PA066496.
Повний текст джерелаBeringer, Frédéric. "Contributions à la résolution d'équations différentielles non linéaires scalaires par la méthode du polygone de Newton." Grenoble INPG, 2002. http://www.theses.fr/2002INPG0133.
Повний текст джерелаCiccoli, Marie Claude. "Schémas numériques efficaces pour le calcul d'écoulements hypersoniques réactifs." Nice, 1992. http://www.theses.fr/1992NICE4574.
Повний текст джерелаHadji, Sofiane. "Méthodes de résolution pour les fluides incompressibles." Compiègne, 1995. http://www.theses.fr/1995COMPD805.
Повний текст джерелаHoarau-Mantel, Thierry-Vincent. "Contribution à l'étude mathématique et numérique de quelques problèmes en mécanique du contact." Perpignan, 2003. http://www.theses.fr/2003PERP0531.
Повний текст джерелаOwing to their inherent diversity and complexity, contact problems lead to various mathematical models governed by nonlinear partial differential equations. Motivated by the richness of this domain, our aim is to study some contact problems involving elastic and viscoelastic materials in the framework of small and large deformation theory. This thesis is structured in four parts. The first one provides the background in Mathematics and Mechanics needed in the rest of the manuscript. The second part concerns the study of three elastic or visocelastic problems in the small strain theory. For these problems we prove existence, uniqueness and continuous dependence of weak solutions in terms of displacements and stress. The third part deals with the study of a unilateral frictionless contact problem between two viscoelastic bodies, again in the framework of small deformations. We illustrate the theoritical results with numerical simulations in dimension two. The fourth part is completely devoted to the numerical modelling of elastic and viscoelastic contact problems with Coulomb law of dry friction in the frame of large deformation. We present numerical simulations modelling the compression of hexagonal cells in two dimension
Djellali, Assia. "Optimisation technico-économique d'un réseau d'énergie électrique dans un environnement dérégulé." Paris 11, 2003. http://www.theses.fr/2003PA112211.
Повний текст джерелаThe electric utility industry is undergoing a process of liberalization and deregulation. In this context new difficulties are occurring in the field of transmission network management and optimization. In addition to the classical difficulties encountered in a monopolistic context such as the nature of the network constraints, the considerable size of the problem to be solved and the nonlinearity of the network equations, the optimization procedure has to take into account the new constraints, which are related to the deregulation of the electrical energy market. The nature of this problem requires mathematical models, which allow us the optimization of a nonlinear criterion being subject to nonlinear constraints. In this thesis we investigate two different methods in order to determine on the one hand the difficulties related to the resolution of a nonlinear optimization problem and on the other hand the difficulties related to the network operation in a deregulated environment. The first method is the so-called Newton-Lagrange method, which is applied to a simplified 5-buses network in a monopolistic context in order to achieve a technico-economical optimization. The optimization goal is the determination of the optimal power generation of each power producer to ensure the security of the system operation and to minimize the system operation costs. Even though convergence time can be considerable due to the inequality constraints, the method provides satisfactory results and will be used as a basis in the second part. A second optimization tool is developed, which is based on the primal-dual interior point method. It is applied to a 12-buses test network in order to investigate and to resolve the difficulties related to a competitive environment such as congestion and energy lasses management, the control of generation deviations and the impact of the occurrence of new independent power producers in an established network. An important advantage of this method is the capacity to treat the inequality constraints in an easy way. The reliable and robust optimization tool provides very satisfactory results
Wazner, Alain. "Formes canoniques invariantes d'un système linéaire différentiel homogène, polygone de Newton, calcul de la partie exponentielle des solutions formelles." Université Joseph Fourier (Grenoble), 1998. http://www.theses.fr/1998GRE10233.
Повний текст джерелаDaude, Frédéric. "Méthode d'intégration temporelle implicite pour la simulation des grandes échelles : application à la réduction du bruit de cavité." Poitiers, 2007. http://www.theses.fr/2007POIT2256.
Повний текст джерелаA study concerning the analysis of implicit time integration methods for Large-Eddy Simulation of compressible flows is presented. The objective is the increasement of the numerical efficiency of schemes to tackle multiscale problems in aerodynamic like controlled flows. This work is divided into three parts: 1- an analysis of the influence of the convergence residual and numerical parameters on the accuracy of the numerical solution, 2 - the developpement of an efficient strategy based on an new block local optimisation, 3 - the demonstration of the potentiability of the method proposed in the case of a transonic cavity flow controlled by means a spanwise cylinder. The method makes it possible to reduce the computational effort by a factor 10
Ayyad, Youssef. "Analyse variationnelle et numérique de quelques problèmes dynamiques en mécanique de contact." Perpignan, 2008. http://www.theses.fr/2008PERP0936.
Повний текст джерелаThe aim of my thesis is the study of dynamics frictional and frictionless contact problems between a deformable body and a foundation. We present first the mechanical models as well as some preliminaries of functional analysis and partial differential equations. Then we provide the variational analysis of two frictionless contact problems with elasto-visco-plastic materials, the contact being modelled with the normal compliance condition and with the normal damped response condition, respectively. For these problems we obtain existence and uniqueness results of the solution. Next, we analyze numerically hyperelastodynamic frictional contact problems, with the goal to ensure the conservation of energy during the impact. To this end we developed a 2 steps continuation Newton method in which the unilateral contact condition is satisfied in the first step and which takes into account the consistence contact condition in the second step. This study also includes numerical simulations, which allows us to compare the method developed here with two different numerical methods based on the energy conservation
Gratton, Serge. "Outils théoriques d'analyse du calcul à précision finie." Toulouse, INPT, 1998. http://www.theses.fr/1998INPT015H.
Повний текст джерелаRey, Christian. "Développement d'algorithmes parallèles de résolution en calcul non-linéaire de structures hétérogènes : application au cas d'une butée acier élastomère." Cachan, Ecole normale supérieure, 1994. http://www.theses.fr/1994DENS0025.
Повний текст джерелаPaşca, Ioana. "Vérification formelle pour les méthodes numériques." Nice, 2010. http://www.theses.fr/2010NICE4104.
Повний текст джерелаThis thesis deals with the formalization of mathematics in the proof assistant Coq with the purpose of verifying numerical methods. We focus in particular on formalizing concepts involved in solving systems of equations, both linear and non-linear. We analyzed Newton's method which is a numerical method widely used for approximating solutions of equations or systems of equations. The goal was to formalize Kantorovitch's theorem which gives the convergence of Newton’s method to a solution, the speed of the convergence and the local stability of the method. The formal study of this theorem also demanded a formalization of concepts of multivariate analysis. Based on these classic results on Newton's method, we showed that rounding at each step in Newton's method still yields a convergent process with an accurate correlation between the precision of the input ant that of the result. In a joint work with Nicolas Julien, we studied formally computations with Newton's method in a library of exact real arithmetic. For linear systems of equations, we analyzed the case where the associated matrix has interval coefficients. For solving such systems, an important issue is to establish whether the associated matrix is regular. We provide a collection of formally verified criteria for regularity of interval matrices
Gaydu, Michaël. "Développements autour de la résolution variationnelles métriquement régulières." Antilles-Guyane, 2010. http://www.theses.fr/2010AGUY0353.
Повний текст джерелаThis work is devoted to the study of metrically regular variational inclusions, l. E. , inclusions involving a set-valued mapping enjoying some metric regularity properties. First, we present a so-called Tikhonov regularization method for solving such inclusions. We subsequently study the case when the set-valued mapping we deal with is metrically regular, strongly metrically regular, strongly metrically subregular and Lipschitz continuous. We provide several convergence theorems and establish also stability results for the regularization of Tikhonov As an illustration we present an optimization problem, deriving from a simplified model of oligopolistic competition in an electricity spot market, and we show how our results can be of interest for solving such a problem. Finally, we study a Newton-type method for solving parametric generalized equations. We state some Lyusternik-Graves theorem in relation with the method we consider and establish the uniform convergence of our algorithm
Ziani, Mohammed. "Accélération de la convergence des méthodes de type Newton pour la résolution des systèmes non-linéaires." Rennes 1, 2008. ftp://ftp.irisa.fr/techreports/theses/2008/ziani.pdf.
Повний текст джерелаIn this thesis, we propose, on one hand, a new Broyden like method called Broyden autoadaptive limited memory Broyden method. The key point of this method is that only the necessary Broyden directions for the convergence are stored. The method starts with a minimal memory, but when a lack of convergence is detected, the size of the approximation subspace is automatically increased. Unlike classical limited memory methods, its advantage is that it does not require the parameter for the dimension of the approximating subspace. The autoadaptive method reduces efficiently computational time and storage cost. Moreover, under classic assumptions, we prove occurrence of superlinear convergence. On the other hand, we solve two nonlinear partial differential equations which arise in two contexts. The first problem consists in solving nonlinear models in image processing. In that application, the autoadaptive method converges better than the other variants of Newton method. When nonuniform noise is introduced, Newton type methods cannot converge. Actually, nonlinearities of the system are unbalanced. We so apply a nonlinear preconditioner to the problem. We use in particular the nonlinear preconditioner based on the nonlinear additive Schwarz algorithm. The second application concerns the solution of a nonlinear problem modelling the displacement of a pile inserted in a ground
Floc'h, France. "Prédictions de trajectoires d'objets immergés par couplage entre modèle d’écoulement et équations d'Euler-Newton." Brest, 2011. http://www.theses.fr/2011BRES2026.
Повний текст джерелаNumerical instabilities due to fluid inertia appear when solving the free motion of a solid submerged within a heavy fluid such as water. In the present thesis, a numerical scheme is proposed to overcome this problem. Three-dimensional simulations using Computational Fluid Dynamics and Euler-Newton equations use too much computing resources for a reasonable investigation of the general case. It was therefore decided to design and build a two-dimensional hydrodynamic tunnel in order to validate the numerical tool. First, a static two-dimensional tank has been built to verify the feasibility of such an apparatus. It reveals the chaotic aspect of the trajectories of light objects when viscous forces are highly unsteady. It is observed in the hydrodynamic tunnel that an income flow stabilizes the translations. The evolution of die angle is still controlled by the wake. In the case of a parallelepipedic object, presenting sharp corners, boundary layer separations occur and induce instabilities. The prediction of tire angle is then difficult. This method is then used to simulate biomimetic propulsion using a porpoising foil. The hydrodynamic solver is a potential flow code. To understand the influence of each parameter on the performances, all degrees of freedom are fixed. Our results for the thrust loading coefficient are in conformity with tire Theodorsen theory over the whole range of parameters. The parametric study confirms that the Strouhal number is playing the same role for the oscillating wing, the advance parameter is playing for the propeller. The two propulsion devices are found to be comparable and a general guidance for comparison between the two propulsion systems is developed. When a change of pace is required, the variable pitch propeller is more efficient than a variation of the pitch amplitude during the foil motion. Results in free motion demonstrate the robustness of the method
Castellanos, Lopez Clara. "Accélération et régularisation de la méthode d'inversion des formes d'ondes complètes en exploration sismique." Phd thesis, Université Nice Sophia Antipolis, 2014. http://tel.archives-ouvertes.fr/tel-01064412.
Повний текст джерелаSenet, Philippe. "Simulation en régime permanent de procédés de séparation multiétagés de systèmes di-et tri-phasiques." Toulouse, INPT, 1987. http://www.theses.fr/1987INPT042G.
Повний текст джерелаGuo, Chaomei. "Amélioration des propriétés de convergence des algorithmes de simulation des circuits non linéaires microondes." Limoges, 1995. http://www.theses.fr/1995LIMO0024.
Повний текст джерелаAl, Sayed Ali Mouhamad. "Accélération de schémas d'intégration temporelle pour la résolution d'équations différentielles." Brest, 2007. http://www.theses.fr/2007BRES2021.
Повний текст джерелаWhen solving ordinary differential equations or algebraic-differential equations by implicit schemes, one is faced with the difficulty of solving correctly the repeated non linear and linear systems of large size that arise in the implicit schemes. In such a case, Newton-like iteration methods for solving nonlinear systems and iterative methods for linear systems can be used. The Newton-like iteration methods are based upon the idea of using a basic Newton iteration in which Newton equations are solved approximately by an available iterative method. The Newton method converges when the initial guess is close enough to a solution, so a modification is needed to guarantee convergence for arbitrary initial guess. This thesis presents a new approach to compute good initial solutions to the linear systems arising in the implicit schemes, for the Newton method and for the linear systems in the Newton method
Gao, Dong Ming. "Modélisation numérique du remplissage des moules de fonderie par la méthode des éléments finis." Compiègne, 1991. http://www.theses.fr/1991COMPD404.
Повний текст джерелаMachui, Jürgen. "Simulation magnétostatique de têtes magnétiques en 3D par décomposition du domaine." Paris 11, 1988. http://www.theses.fr/1988PA112055.
Повний текст джерелаThis work concerns the simulation of planar magnetic recording heads in the context of its industrial development. Finite elements and reduced potential are used for the 3D calculation of the magnetostatic problem. The particular difficulty of magnetic recording heads lies in the enormous difference in scale between the gap and the whole head. We resolve this difficulty using an iteratif algorithm for domain decomposition for symmetrical heads that converges very rapidly. The non-linear saturation problem can be resolved using the Newton-Raphson method. Our decomposition algorithm is equally efficient for this kind of problem
Gallimard, Laurent. "Contrôle adaptatif des calculs en élastoplasticité et en viscoplasticité." Cachan, Ecole normale supérieure, 1994. http://www.theses.fr/1994DENS0002.
Повний текст джерелаMinazzoli, Olivier. "Étude relativiste du transfert de temps dans le système solaire." Nice, 2009. http://www.theses.fr/2009NICE4090.
Повний текст джерелаThe relativistic vision of time has been tested for decades. However, theoretical considerations springing from the will of unifying fundamental physics suggest that the theory we use to describe space and time, General Relativity, will be no longer accurate enough to describe the flow of time in our future experiments. Especially in time transfer experiments which link remote clocks by laser links. The goal of such experiments can be either the synchronisation of remote clocks or the theoretical study of the time transfer problem and its application to present and future space experiments
Pons, Bernard. "Etude numérique de l'allumage et du développement de la combustion dans un jet de monergol pulvérisé." Poitiers, 1996. http://www.theses.fr/1996POIT2369.
Повний текст джерелаCabuzel-Zèbre, Catherine. "Résolution d'inclusions variationnelles par des méthodes multipoints et des méthodes classiques dans le cadre sous-analytique." Antilles-Guyane, 2008. http://www.theses.fr/2008AGUY0239.
Повний текст джерелаThis work deals with seminumerical methods for soIving variational inclusions of the form zero in f(x)+F(x) where the function f and the set-valued map F are bath acting in a Banach space. The fisrt part, we recall sorne resulls on Upschitz continuity, directional derivatives, semialgebraic and subanalytic sets and functions and divided differences; then we give some results on set-valued analysis. In the second part, we present the muItipoint iterative method and we deveIop the soIving of variational inclusions by this method in the Lipschitz case, the HöIder case and the center-Hölder case. The third part is dedicated to the study of classical methods in the subanalytic case. Newton's method was the subject of many works concerning the sollving of equations or variational inclusions, but the case where fis not Frechet derivable or does not admit divided dilferences has not been studied 50 far; that is why we investigate a Newton type method when f is subanalytic. Afterwards, our study concerns a perturbed probIem of the form zero in f(x)+g(x)+F(x), where all the functions involved are acting in R^n. We analyse a NewIon-secant type method and two variants: a regula-falsi method and an acceleration of the previous method. Then we present a Stetrensen type method and finally an iterative method in the case where the function g is Lipschitz
Da, Cunha Joao Paulo. "Diagnostic thermique de la machine à courant continu par identification paramétrique." Poitiers, 1999. http://www.theses.fr/1999POIT2353.
Повний текст джерелаAbouir, Jilali. "Problèmes concernant l'approximation rationnelle à plusieurs variables." Lille 1, 1992. http://www.theses.fr/1992LIL10054.
Повний текст джерелаMurray, Maxime. "Existence de connexions homoclines pour l'équation du pont suspendu : une preuve assistée par ordinateur." Master's thesis, Université Laval, 2016. http://hdl.handle.net/20.500.11794/26716.
Повний текст джерелаDans ce mémoire, une méthode assistée numériquement est introduite et utilisée afin de montrer l'existence d'une connexion homocline à zéro pour l'équation du pont suspendu. Cette méthode, basée sur l'utilisation du théorème de contraction de Banach, permet d'obtenir les points fixes de l'opérateur de Newton légèrement modifié. La méthode ainsi que son cadre théorique sont introduits au premier chapitre. L'espace de Banach sur lequel sera définit l'opérateur ainsi que la manière de construire l'approximation de l'inverse utilisée pour l'opérateur sont les éléments majeurs du cadre théorique. Par la suite, la méthode est utilisée dans le Chapitre 2 pour prouver rigoureusement la validité de l'approximation numérique utilisée pour la variété stable locale. Puis cette approximation est réutilisée pour prouver l'existence de la connexion homocline. Cette preuve est à nouveau effectuée en utilisant la méthode introduite au premier chapitre. Finalement, certains résultats des calculs numériques sont présentés pour conclure ce mémoire.
In this work, a numerically assisted technique is introduced in order to prove the existence of a homoclinic connexion to zero for the suspension bridge equation. This technique, based on the use of the Banach fixed point theorem, can provide the fixed point of a slightly modified version of the Newton operator. The technique and its theorical background are introduced in the first chapter. The Banach space on which the operator is defined and the way to construct the approximation of the inverse needed to define the operator are the major parts of the theoretical background. The method is then used to rigorously validate the numerical approximation used to parametrize the local stable manifold. This parametrization is used to find the homoclinic connexion we are looking for. This proof is also completed using the technique from the first chapter. Finally, some results and numerical approximations will be presented in the last chapter.
Cherif, Riheb. "Développement de solveurs non linéaires robustes pour la méthode des éléments finis appliquée à des problèmes électromagnétiques basses fréquences." Thesis, Lille 1, 2019. http://www.theses.fr/2019LIL1I088.
Повний текст джерелаNon-linear electromagnetic problems are widely encountered in electrical engineering like applications of electrical machines. The computation of the magnetic fields requires the resolution of nonlinear problems due to the saturation of ferromagnetic materials. The finite element method (FEM) is a widely used numerical technique in the modeling field thanks to its high precision and robustness for solving systems with complex geometries. The discretization of these problems leads to a large system of nonlinear equations that can be solved by two important iterative processes: the fixed point method which is more robust, but can be very slow, due to its linear convergence rate and the Newton method which has been widely preferred for nonlinear field problems, thanks to its quadratic convergence speed. However, this convergence remains local, it means that the initial estimation must be rather close to the solution. Thus, several globalization techniques are introduced to acheive an acceptable level of robustness. In addition, an iteration of Newton can be very expensive. In fact, at each iteration, it requires the evaluation of the Jacobian and the resolution of the linearized problem involving this last matrix. To reduce these resolution costs, several variants of this method are introduced as the Inexact-Newton methods and Quasi-Newton methods
Vu, Duc Thach Son. "Numerical resolution of algebraic systems with complementarity conditions : Application to the thermodynamics of compositional multiphase mixtures." Electronic Thesis or Diss., université Paris-Saclay, 2020. http://www.theses.fr/2020UPASG006.
Повний текст джерелаIn reservoir simulators, it is usually delicate to take into account the laws of thermodynamic equilibrium for multiphase hydrocarbon mixtures. The difficulty lies in handling the appearance and disappearance of phases for different species. The traditional dynamic approach, known as variable switching, consists in considering only the unknowns and equations of the present phases. It is cumbersome and costly, insofar as "switching" occurs constantly, even from one Newton iteration to another.An alternative approach, called unified formulation, allows a fixed set of unknowns and equations to be maintained during the calculations. From a theoretical point of view, this is an major advance. On the practical level, because of the nonsmoothness of the complementarity equations involved in the new formulation, the discretized equations have to be solved by the semi-smooth Newton-min method, whose behavior is often pathological.In order to fully exploit the interest of the unified approach, this thesis aims at circumventing this numerical obstacle by means of more robust resolution algorithms, with a better convergence. To this end, we draw inspiration from the methods that have proven their worth in constrained optimization and we try to transpose them to general systems. This gives rise to interior-point methods, of which we propose a nonparametric version called NPIPM. The results appear to be superior to those of Newton-min.Another contribution of this doctoral work is the understanding and (partial) resolution of another obstruction to the proper functioning of the unified formulation, hitherto unidentified in the literature. This is the limitation of the domain of definition of Gibbs' functions associated with cubic equations of state. To remedy the possible non-existence of a system solution, we advocate a natural extension of Gibbs' functions
Ach, Karim. "Etude du comportement de structures polyarticulées avec ou sans jeu : techniques d'homogénéisation et modélisation numérique." Montpellier 2, 2000. http://www.theses.fr/2000MON20047.
Повний текст джерелаBlanchard, Louis. "Conception d´antenne avec optimisation des lobes réseau : application au partitionnement en sous réseaux d´une antenne radar." Paris, ENMP, 2007. http://www.theses.fr/2007ENMP1474.
Повний текст джерелаThe synthesis of an array antenna is a inverse problem. The applications of this problem are civil : telecommunication and satellites, but also military : air defense radars. In an array antenna, each element is weighted using the ponderation term and consequently the array antenna beamforming can evolves toward a desired pattern without any mechanical movement of the antenna. The aim of this study is the analysis of a new concept in the antenna synthesis problem, based on a bi-criteria formulation of the antenna beamforming, which is an alternative to the problem of pattern synthesis. Moreover, in order to solve antenna synthesis problem based on this bi-criteria formulation, we need to address two major areas of mathematical analysis. The first area concerns differentiable optimization problems under constraints, consisting in optimizing both ponderation and geometry of the array antenna, and for which a deterministic method based on a tangential Newton-Raphson algorithm is used. The second area concerns the topological optimization problem by moving domains used in the problem of optimal partitioning of an array antenna into sub-array, in the case of radar made by a thousand elements. To conclude, the bi-criteria optimization method can be used, both as in a civil context for array antennas dedicated to mobiles phone, and as in a military context for air defense radars
Bezier, Florence. "Problèmes de transport-diffusion par éléments finis." Compiègne, 1990. http://www.theses.fr/1990COMPD255.
Повний текст джерелаDelbary, Fabrice. "Identification de fissures par ondes acoustiques." Paris 6, 2006. http://www.theses.fr/2006PA066605.
Повний текст джерелаPicart, Philippe. "Contribution à la résolution numérique des problèmes élasto-plastiques et élasto-viscoplastiques en transformations finies." Valenciennes, 1986. https://ged.uphf.fr/nuxeo/site/esupversions/c0aa3577-2ebe-4da6-82da-563b9f60f2ea.
Повний текст джерелаKaboul, Hanane. "Méthodes d'intégration produit pour les équations de Fredholm de deuxième espèce : cas linéaire et non linéaire." Thesis, Lyon, 2016. http://www.theses.fr/2016LYSES024.
Повний текст джерелаThe product integration method has been proposed for solving singular linear Fredholm equations of the second kind whose exact solution is smooth, at least continuous. In this work, we adapt this method to the case where the solution is only integrable. We also study the nonlinear case in the space of integrable functions. Then, we propose a new version of the method in the nonlinear framework : we first linearize the eqaution by a Newton type method and then discretize the Newton iterations by the product integration method