Dissertations / Theses on the topic 'Méthodes de collocation stochastiques'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Méthodes de collocation stochastiques.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
Oumouni, Mestapha. "Analyse numérique de méthodes performantes pour les EDP stochastiques modélisant l'écoulement et le transport en milieux poreux." Phd thesis, Université Rennes 1, 2013. http://tel.archives-ouvertes.fr/tel-00904512.
Full textSilly-Carette, Jessica. "Modélisation avancée de l'absorption des ondes électromagnétiques dans les tissus biologiques : schémas en temps, approches adjointe et stochastique." Paris 6, 2008. http://www.theses.fr/2008PA066368.
Full textJannet, Basile. "Influence de la non-stationnarité du milieu de propagation sur le processus de Retournement Temporel (RT)." Thesis, Clermont-Ferrand 2, 2014. http://www.theses.fr/2014CLF22436/document.
Full textThe aim of this thesis is to measure and quantify the impacts of uncertainties in the Time Reversal (TR) process. These random variations, coming from diverse sources, can have a huge influence if they happen between the TR steps. On this perspective, the Stochastique Collocation (SC) method is used. Very good results in terms of effectiveness and accuracy had been noticed in previous studies in ElectroMagnetic Compatibility (EMC). The conclusions are still excellent here on TR problems. Although, when the problem dimension rises (high number of Random Variables (RV)), the SC method reaches its limits and the efficiency decreases. Therefore a study on Sensitivity Analysis (SA) techniques has been carried out. Indeed, these methods emphasize the respective influences of the random variables of a model. Among the various quantitative or qualitative SA techniques the Morris method and the Sobol total sensivity indices have been adopted. Since only a split of the inputs (point out of the predominant RV) is expected, they bring results at a lesser cost. That is why a novel method is built, combining SA techniques and the SC method. In a first step, the model is reduced with SA techniques. Then, the shortened model in which only the prevailing inputs remain, allows the SC method to show once again its efficiency with a high accuracy. This global process has been validated facing Monte Carlo results on several analytical and numerical TR cases subjet to random variations
Rhayma, Noureddine. "Contribution à l'évolution des méthodologies de caractérisation et d'amélioration des voies ferrées." Phd thesis, Université Blaise Pascal - Clermont-Ferrand II, 2010. http://tel.archives-ouvertes.fr/tel-00719102.
Full textCharrier, Julia. "Analyse numérique d’équations aux dérivées aléatoires, applications à l’hydrogéologie." Thesis, Cachan, Ecole normale supérieure, 2011. http://www.theses.fr/2011DENS0030/document.
Full textThis work presents some results about probabilistic and deterministic numerical methods for partial differential equations with stochastic coefficients, with applications to hydrogeology. We first consider the steady flow equation in porous media with a homogeneous lognormal permeability coefficient, including the case of a low regularity covariance function. We establish error estimates, both in strong and weak senses, of the error in the solution resulting from the truncature of the Karhunen-Loève expansion of the coefficient. Then we establish finite element error estimates, from which we deduce an extension of the existing error estimate for the stochastic collocation method along with an error estimate for a multilevel Monte-Carlo method. We finally consider the coupling of the previous flow equation with an advection-diffusion equation, in the case when the uncertainty is important and the correlation length is small. We propose the numerical analysis of a numerical method, which aims at computing the mean velocity of the expansion of a pollutant. The method consists in a Monte-Carlo method, combining a finite element method for the flow equation and an Euler scheme for the stochastic differential equation associated to the advection-diffusion equation, seen as a Fokker-Planck equation
Charrier, Julia. "Analyse numérique d'équations aux dérivées aléatoires, applications à l'hydrogéologie." Phd thesis, École normale supérieure de Cachan - ENS Cachan, 2011. http://tel.archives-ouvertes.fr/tel-00625092.
Full textperrin, nicolas. "Méthodes stochastiques en dynamique moléculaire." Phd thesis, Université Nice Sophia Antipolis, 2013. http://tel.archives-ouvertes.fr/tel-00833886.
Full textPerrin, Nicolas. "Méthodes stochastiques en dynamique moléculaire." Thesis, Nice, 2013. http://www.theses.fr/2013NICE4011/document.
Full textThis thesis presents two independent research topics. Both are related to the application of stochastic problems to molecular dynamics. In the first part, we present a work related to the probabilistic interpretation of the Poisson-Boltzmann equation. This equation describes the electrostatic potential of a molecular system. After an introduction to the Poisson-Boltzmann equation, we focus on the parabolic and linear equation. After extending an existence and uniqueness result for backward stochastic differential equations, we establish a probabilistic interpretation of the nonlinear Poisson-Boltzmann equation with backward stochastic differential equations. Finally, in a more prospective second part, we initiate a study of a slow and fast variables detection method due to Paul Malliavin
Arouna, Bouhari. "Algotithmes stochastiques et méthodes de Monte Carlo." Phd thesis, Ecole des Ponts ParisTech, 2004. http://pastel.archives-ouvertes.fr/pastel-00001269.
Full textArouna, Bouhari. "Méthodes de Monté Carlo et algorithmes stochastiques." Marne-la-vallée, ENPC, 2004. https://pastel.archives-ouvertes.fr/pastel-00001269.
Full textMercier, Sophie. "Modèles stochastiques et méthodes numériques pour la fiabilité." Habilitation à diriger des recherches, Université Paris-Est, 2008. http://tel.archives-ouvertes.fr/tel-00368100.
Full textNous nous intéressons ensuite au remplacement préventif de composants devenus obsolescents, du fait de l'apparition de nouveaux composants plus performants. Le problème est ici de déterminer la stratégie optimale de remplacement des anciens composants par les nouveaux. Les résultats obtenus conduisent à des stratégies très différentes selon que les composants ont des taux de panne constants ou non.
Les travaux suivants sont consacrés à l'évaluation numérique de différentes quantités fiabilistes, les unes liées à des sommes de variables aléatoires indépendantes, du type fonction de renouvellement par exemple, les autres liées à des systèmes markoviens ou semi-markoviens. Pour chacune de ces quantités, nous proposons des bornes simples et aisément calculables, dont la précision peut être ajustée en fonction d'un pas de temps. La convergence des bornes est par ailleurs démontrée, et des algorithmes de calcul proposés.
Nous nous intéressons ensuite à des systèmes hybrides, issus de la fiabilité dynamique, dont l'évolution est modélisée à l'aide d'un processus de Markov déterministe par morceaux (PDMP). Pour de tels systèmes, les quantités fiabilistes usuelles ne sont généralement pas atteignables analytiquement et doivent être calculées numériquement. Ces quantités s'exprimant à l'aide des lois marginales du PDMP (les lois à t fixé), nous nous attachons plus spécifiquement à leur évaluation. Pour ce faire, nous commençons par les caractériser comme unique solution d'un système d'équations intégro-différentielles. Puis, partant de ces équations, nous proposons deux schémas de type volumes finis pour les évaluer, l'un explicite, l'autre implicite, dont nous démontrons la convergence. Nous étudions ensuite un cas-test issu de l'industrie gazière, que nous modélisons à l'aide d'un PDMP, et pour lequel nous calculons différentes quantités fiabilistes, d'une part par méthodes de volumes finis, d'autre part par simulations de Monte-Carlo. Nous nous intéressons aussi à des études de sensibilité : les caractéristiques d'un PDMP sont supposées dépendre d'une famille de paramètres et le problème est de comparer l'influence qu'ont ces différents paramètres sur un critère donné, à horizon fini ou infini. Cette étude est faite au travers des dérivées du critère d'étude par rapport aux paramètres, dont nous démontrons l'existence et que nous calculons.
Enfin, nous présentons rapidement les travaux effectués par Margot Desgrouas lors de sa thèse consacrée au comportement asymptotique des PDMP, et nous donnons un aperçu de quelques travaux en cours et autres projets.
Papadopoulos, Christos. "Méthodes de simulation stochastiques rapides et applications en fiabilité." Compiègne, 1999. http://www.theses.fr/1999COMP1213.
Full textDettori, Lucia. "La méthode de Galerkin nonlinéaire en discrétisation par collocationL'équation de Cahn-Hilliard avec une énergie libre logarithmique." Paris 11, 1994. http://www.theses.fr/1994PA112034.
Full textZein, Samih. "Application des méthodes d’optimisation stochastiques à deux problèmes d’inversion sismique." Rennes 1, 2007. http://www.theses.fr/2007REN1S043.
Full textIn this thesis we are interested in solving two seismic inverse problems using Monte Carlo type methods. The first problem is on estimating the density and the Lamé's coefficients of a thin layered solid from measures of the pressure variation due to the propagation of a seismic wave. We use MCMC and SPSA methods and a finite elements code to perform the inversion. We give an estimation of the treshold of the noise in the pressure measures with respect to the condition number of the jacobien of the foward problem. The second inverse problem is on the estimation of the wave's velocity from the seismic traveltime measures. We propose a reduction of the computations time by taking a formulation that avoids seeking the rays that are hard to find. We use the genetic algorithms and the ray tracing technique. We study the robustness of the solution by doing a SVD of the jacobien to find the solution
Lévy, Véhel Jacques. "Analyse et synthèse d'objets bi-dimensionnels par des méthodes stochastiques." Paris 11, 1988. http://www.theses.fr/1988PA112404.
Full textSchmied, Aurélien. "Méthodes stochastiques d'optimisation appliquées à la mise au point moteur." Paris 12, 2003. https://athena.u-pec.fr/primo-explore/search?query=any,exact,990003948640204611&vid=upec.
Full textThis thesis deals with stochastic methods of optimisation applied to engine calibration. The purpose of engine calibration is to lower each car's consumption and to met the regulation on the emission of different pollutants. This thesis presents two modifications of the methodology actually used at Renault an a new approach which is different of the current process. The first modofication consists of reformuling the optimisation problem using a model issued from Fuzzy logic and the second modification proposes a new stochastic optimisation algorithm named "Multistoch", for which convergence's results are presented. The new approach is based on designing dynamic experiments (curves in a plan) using constrained functional quantification. For this purpose, we propose a new type of constrained distortion which is minimised using different stochastic optimisation algorithms
Bailly, Raphael. "Méthodes spectrales pour l'inférence grammaticale probabiliste de langages stochastiques rationnels." Thesis, Aix-Marseille 1, 2011. http://www.theses.fr/2011AIX10125/document.
Full textOur framework is the probabilistic grammatical inference. That is, given an unknown distribution p on a set of string S∗ , to infer a probabilistic model for p from a sample S of observations assumed to be i.i.d. according to p. Grammatical inference focuses primarily on the structure of the probabilistic model, and the convergence of parameter estimate. Probabilistic models which will be considered here are weighted automata, or WA. The series they model are called rational series. Initially, we study the possibility of finding an absolute convergence criterion for such series. Subsequently, we introduce a algorithm for the inference of rational distrbutions (i.e. distributions modeled by WA), based on spectral methods. We will show how to fit this algorithm to the domain, fairly close, of rational distributions on trees. Finally, we will try to see how to use the spectral algorithm in a more statistical way, in a density estimation task
Zaytsev, Victor. "Méthodes stochastiques pour la modélisation d'incertitudes sur les maillages non structurés." Thesis, Paris Sciences et Lettres (ComUE), 2016. http://www.theses.fr/2016PSLEM094/document.
Full textSimulations of physical phenomenon often require discretizing the medium with a mesh. An example of this type of simulation is the simulation of fluid flow through a porous medium and the evaluation of the geomechanical stress in the petroleum reservoir. The studied medium is often not homogeneous and applying a homogeneity hypothesis can lead to incorrect simulation results. That makes simulation of heterogeneities important for this kind of problems.This thesis is devoted to geostatistical simulations of heterogeneities on unstructured grids using methods of non-linear geostatistics. The objective of this work is the development of algorithms for simulating heterogeneities directly on unstructured grids without using intermediate fine scale regular grids and upscaling. We present two theoretical models for geostatistical simulations of continuous parameters on unstructured grids which are different generalizations of the Discrete Gaussian model (DGM) – DGM 1 and DGM 2. The proposed theoretical models enable converting the problem of geostatistical simulation on an unstructured grid into the well-studied problem of simulating multivariate Gaussian random vectors followed by application of block-dependent transformation functions. The problem of simulating facies is also addressed in this work, for which generalizations of pluri-Gaussian and truncated Gaussian simulation models for unstructured grids are proposed.An application of the proposed methods is demonstrated on a case study X, which is an offshore gas reservoir with a tartan-meshed grid
Lacomme, Philippe. "Optimisation des systèmes de production : méthodes stochastiques et approche multi-agents." Clermont-Ferrand 2, 1998. http://www.theses.fr/1998CLF21986.
Full textRamdane, Saïd. "Identification automatique de types de formulaires par des méthodes stochastiques markoviennes." Le Havre, 2002. http://www.theses.fr/2002LEHA0018.
Full textIdentification of forms is a significant operation of an automatic system of reading. No distinctive sign is supposed to mark the form. The treatment starts with the extraction of the rectangular blocks of texts or rectangles including the drawings or the images. Since the forms include handwritten fields, the position, dimensions of the rectangular blocks present are variable. The phenomena of merging and fragmentation resulting from the segmentation induce an additional variability in the number of the rectangles. This double variability of the rectangles is naturally random. A first statistical method carries out the recognition by the calculation of a distance, which generalizes the Mahalanobis distance. Learning requires taking care of the phenomenon of merging/fragmentation. This statistical model appears to be actually a Markovian stochastic model of order 0. A second stochastic method rests on the construction of planar hidden Markov models (PHMM: Pseudo-2D Hidden Markov Model). We describe in particular a new unsupervised training of the states number by a dynamic aggregation method. The recognition is based on the estimation of the conditional probability calculated by an extension of a doubly imbricated Viterbi algorithm. For the two methods, we sought to make automatic all the phases of the training and the recognition. The experimental results confirm the validity of the two methods
Tong, Dinh Quy. "Méthodes d'estimation de paramètres de modèles autorégressifs multivariés." Grenoble 1, 1991. http://www.theses.fr/1991GRE10141.
Full textBréhier, Charles-Edouard. "Analyse numérique d'EDP Stochastiques hautement oscillantes." Phd thesis, École normale supérieure de Cachan - ENS Cachan, 2012. http://tel.archives-ouvertes.fr/tel-00763340.
Full textChaouche, Ali Mahmoud. "Une méthode de collocation pour la prédiction et l'analyse des écoulements et des transferts dans les milieux en rotation rapide." Aix-Marseille 2, 1991. http://www.theses.fr/1991AIX22065.
Full textPoëtte, Gaël. "Propagation d'incertitudes pour les systèmes de lois de conservation, méthodes spectrales stochastiques." Paris 6, 2009. http://www.theses.fr/2009PA066801.
Full textIn this thesis, we are interested in the quantification of uncertainty with stochastic spectral methods applied to systems of conservation laws. These are nonlinear and develop discontinuities: these difficulties can foster a loss of the hyperbolicity of the truncated system resulting from the application of sG-gPC (Galerkin projection). We introduce a well suited formalisme based on both kinetic theory and moment theory so as to ensure, by construction, the hyperbolicity of the resulting truncated system. The idea is to close the truncated thanks to a nonlinear Galerkin projection via introduction of a system entropy - strictly convex function. In the case this entropy is the mathematical entropy of the non truncated system, hyperbolicity is ensured. We establish several properties of the truncated system obtained with the new method for an arbitrary conservation law. We apply the new method to Burgers equation and Euler. In the vicinity of discontinuities, the new method bounds the oscillations due to Gibbs phenomenon without the use of an adaptive procedure. The method reveals to be more precise than classical intrusive method on several test problems. In a last chapter, we present two prospective tracks: we suggest an uncertainty quantification method coupling truncated system and non truncated system. We finally emphasize the potential of modelization of the intrusive approach so as to take into account tridimensional perturbations of a monodimensional mean flow
Martinez, Miguel. "Interprétations probabilistes d'opérateurs sous forme divergence et analyse de méthodes numériques probabilistes associées." Aix-Marseille 1, 2004. http://www.theses.fr/2004AIX11068.
Full textIlland, Camille. "Contrôle stochastique par méthodes de quantification et applications à la finance." Phd thesis, Université Pierre et Marie Curie - Paris VI, 2012. http://tel.archives-ouvertes.fr/tel-00768570.
Full textDia, Baye Moussa. "Méthodes et modèles d'évaluation d'options avec dividende stochastique." Paris 1, 2007. http://www.theses.fr/2007PA010035.
Full textDescombes, Xavier. "Méthodes stochastiques en analyse d'image : des champs de Markov aux processus ponctuels marqués." Habilitation à diriger des recherches, Université de Nice Sophia-Antipolis, 2004. http://tel.archives-ouvertes.fr/tel-00506084.
Full textOdasso, Cyril. "Méthodes de couplage pour des équations stochastiques de type Navier-Stokes et Schrödinger." Phd thesis, Université Rennes 1, 2005. http://tel.archives-ouvertes.fr/tel-00011214.
Full textDans un deuxième temps, nous considérerons les équations de Navier-Stokes stochastiques tridimensionnelles (NS3D). Nous établirons la régularité Hp et Gevrey des solutions stationnaires de NS3D et nous en déduirons des informations sur l'échelle de dissipation de Kolmogorov (K41). Puis, nous établirons le caractère exponentiellement mélangeant des solutions de NS3D lorsque le bruit est à la fois suffisament régulier et non-dégénéré.
Giffard, Françòis-Xavier. "Développements utilisant des méthodes stochastiques et déterministes pour l'analyse de systèmes nucléaires complexes." Evry-Val d'Essonne, 2000. http://www.theses.fr/2000EVRY0007.
Full textGao, Yueyuan. "Méthodes de volumes finis pour des équations aux dérivées partielles déterministes et stochastiques." Thesis, Université Paris-Saclay (ComUE), 2015. http://www.theses.fr/2015SACLS187/document.
Full textThis thesis bears on numerical methods for deterministic and stochastic partial differential equations; we perform numerical simulations by means of finite volume methods and prove convergence results.In Chapter 1, we apply a semi-implicit time scheme together with the generalized finite volume method SUSHI for the numerical simulation of density driven flows in porous media; it amounts to solve a nonlinear convection-diffusion parabolic equation for the concentration coupled with an elliptic equation for the pressure. We then propose a numerical scheme to simulate density driven flows in porous media coupled to heat transfer. We use adaptive meshes, based upon square or cubic volume elements.In Chapter 2, We perform Monte-Carlo simulations in the one-dimensional torus for the first order Burgers equation forced by a stochastic source term with zero spatial integral. We suppose that this source term is a white noise in time, and consider various regularities in space. We apply a finite volume scheme combining the Godunov numerical flux with the Euler-Maruyama integrator in time. It turns out that the empirical mean converges to the space-average of the deterministic initial condition as t → ∞. The empirical variance also stabilizes for large time, towards a limit which depends on the space regularity and on the intensity of the noise.In Chapter 3, we study a time explicit finite volume method with an upwind scheme for a first order conservation law with a monotone flux function and a multiplicative source term involving a Q-Wiener process. We present some a priori estimates including a weak BV estimate. After performing a time interpolation, we prove two entropy inequalities for the discrete solution and show that it converges up to a subsequence to a stochastic measure-valued entropy solution of the conservation law in the sense of Young measures.In Chapter 4, we obtain similar results as in Chapter 3, in the case that the flux function is non-monotone, and that the convection term is discretized by means of a monotone scheme
Ehrenstein, Uwe. "Méthodes spectrales de résolution des équations de Stokes et de Navier-Stokes : application à des écoulements de convection double diffusive." Nice, 1986. http://www.theses.fr/1986NICE4056.
Full textSène, Mbaye. "Méthodes et outils d'évaluation des systèmes transactionnels répartis." Paris 9, 2002. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=2002PA090049.
Full textMennouni, Abdelaziz. "Sur la résolution des équations intégrales singulières à noyau de Cauchy." Thesis, Saint-Etienne, 2011. http://www.theses.fr/2011STET4005/document.
Full textThe purpose of this thesis is to develop and illustrate various new methods for solving many classes of Cauchy singular integral and integro-differential equations. We study the successive approximation method for solving Cauchy singular integral equations of the first kind in the general case, then we develop a collocation method based on trigonometric polynomials combined with a regularization procedure, for solving Cauchy integral equations of the second kind. In the same perspective, we use a projection method for solving operator equation with bounded noncompact operators in Hilbert spaces. We apply a collocation and projection methods for solving Cauchy integro-differential equations, using airfoil and Legendre polynomials
Deaconu, Madalina. "Processus stochastiques associés aux équations d'évolution linéaires ou non-linéaires et méthodes numériques probabilistes." Habilitation à diriger des recherches, Université Henri Poincaré - Nancy I, 2008. http://tel.archives-ouvertes.fr/tel-00590778.
Full textSlaoui, Yousri. "Application des méthodes d'approximations stochastiques à l'estimation de la densité et de la régression." Phd thesis, Université de Versailles-Saint Quentin en Yvelines, 2006. http://tel.archives-ouvertes.fr/tel-00131964.
Full textTryoen, Julie. "Méthodes de Galerkin stochastiques adaptatives pour la propagation d'incertitudes paramétriques dans les modèles hyperboliques." Phd thesis, Université Paris-Est, 2011. http://pastel.archives-ouvertes.fr/pastel-00795322.
Full textRubenthaler, Sylvain. "Méthodes de Monte-Carlo en filtrage non linéaire et pour certaines équations différentielles stochastiques." Paris 6, 2002. http://www.theses.fr/2002PA066546.
Full textHajji, Omessaad. "Contribution au développement de méthodes d'optimisation stochastiques : application à la conception des dispositifs électrotechniques." Lille 1, 2003. https://pepite-depot.univ-lille.fr/LIBRE/Th_Num/2003/50376-2003-237-238.pdf.
Full textBuslig, Leticia. "Méthodes stochastiques de modélisation de données : application à la reconstruction de données non régulières." Thesis, Aix-Marseille, 2014. http://www.theses.fr/2014AIXM4734/document.
Full textBen, Mamoun Mouad. "Encadrements stochastiques et évaluation de performances des réseaux." Versailles-St Quentin en Yvelines, 2002. http://www.theses.fr/2002VERS0012.
Full textJeannet, Bernard. "Autour de l'usage de méthodes probabilistes en mécanique des milieux continus." Paris 13, 1992. http://www.theses.fr/1992PA132002.
Full textBouhaddane, Abdelaziz. "Application de l'intégrale de Cauchy à la méthode des discontinuités de déplacement et autres méthodes de collocation." Lille 1, 1987. http://www.theses.fr/1987LIL10207.
Full textVernie, Pascale. "Géothermobarométrie : applications des méthodes d'inversion stochastique à la construction des bases de données thermodynamiques et à la modélisation des solutions solides." Paris 6, 1990. http://www.theses.fr/1990PA066711.
Full textPham, Huyên. "Applications des méthodes probabilistes et de contrôle stochastique à la finance mathématique." Paris 9, 1995. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=1995PA090001.
Full textNorre, Sylvie. "Problèmes de placement de taches sur des architectures multiprocesseurs : méthodes stochastiques et évaluation des performances." Clermont-Ferrand 2, 1993. http://www.theses.fr/1993CLF21511.
Full textMokdad, Nadia Lynda. "Méthodes et outils pour l'évaluation des performances des réseaux informatiques." Versailles-St Quentin en Yvelines, 1997. http://www.theses.fr/1997VERS0011.
Full textDiniz, Alberto. "Extension des méthodes de synthèse modale aux structures non-déterministes." Ecully, Ecole centrale de Lyon, 2000. http://www.theses.fr/2000ECDL0022.
Full textThe modal synthesis method and the stochastic finite elements method are applied to model structures with large and stochastic interfaces between its substructures. The inherent randomness of the technological systems of substructures connection requires a statistical approach, which increases the size and complexity of the problem. In the same way, the large interfaces involve a similar problem in the application of the traditional methods of modal synthesis. In order ot avoid these difficulties, we propose an innovating method of double modal synthesis, using branch modes defined in a particular way. These branch modes, very representative of the link interface dynamics, lead to equivalent results of those ontained by traditional methods, but with an importanting time computational reduction. This method will be extended to the structures with random properties on the interfaces by using the perturbations approach. For these interfaces, we use additional link substructures which gather the random parameters independently of the principal substructures. The proposed method is compared to traditional approaches by numerical tests. The examples show the advantages of the use of the sub-structuring to treat stochastic structures. Lastly, the method is applied to an industrial example (a car muffler) with a large interface ad a random rigidity
Glaude, Hadrien. "Méthodes des moments pour l’inférence de systèmes séquentiels linéaires rationnels." Thesis, Lille 1, 2016. http://www.theses.fr/2016LIL10107/document.
Full textLearning stochastic models generating sequences has many applications in natural language processing, speech recognitions or bioinformatics. Multiplicity Automata (MA) are graphical latent variable models that encompass a wide variety of linear systems. In particular, they can model stochastic languages, stochastic processes and controlled processes. Traditional learning algorithms such as the one of Baum-Welch are iterative, slow and may converge to local optima. A recent alternative is to use the Method of Moments (MoM) to design consistent and fast algorithms with pseudo-PAC guarantees. However, MoM-based algorithms have two main disadvantages. First, the PAC guarantees hold only if the size of the learned model corresponds to the size of the target model. Second, although these algorithms learn a function close to the target distribution, most do not ensure it will be a distribution. Thus, a model learned from a finite number of examples may return negative values or values that do not sum to one. This thesis addresses both problems. First, we extend the theoretical guarantees for compressed models, and propose a regularized spectral algorithm that adjusts the size of the model to the data. Then, an application in electronic warfare is proposed to sequence of the dwells of a super-heterodyne receiver. Finally, we design new learning algorithms based on the MoM that do not suffer the problem of negative probabilities. We show for one of them pseudo-PAC guarantees
Detournay, Sylvie. "Méthodes multigrilles pour les jeux stochastiques à deux joueurs et somme nulle, en horizon infini." Phd thesis, Ecole Polytechnique X, 2012. http://pastel.archives-ouvertes.fr/pastel-00762010.
Full text