Teses / dissertações sobre o tema "Algorithmes de Newton stochastiques"
Crie uma referência precisa em APA, MLA, Chicago, Harvard, e outros estilos
Veja os 50 melhores trabalhos (teses / dissertações) para estudos sobre o assunto "Algorithmes de Newton stochastiques".
Ao lado de cada fonte na lista de referências, há um botão "Adicionar à bibliografia". Clique e geraremos automaticamente a citação bibliográfica do trabalho escolhido no estilo de citação de que você precisa: APA, MLA, Harvard, Chicago, Vancouver, etc.
Você também pode baixar o texto completo da publicação científica em formato .pdf e ler o resumo do trabalho online se estiver presente nos metadados.
Veja as teses / dissertações das mais diversas áreas científicas e compile uma bibliografia correta.
Lu, Wei. "Μéthοdes stοchastiques du secοnd οrdre pοur le traitement séquentiel de dοnnées massives". Electronic Thesis or Diss., Normandie, 2024. http://www.theses.fr/2024NORMIR13.
Texto completo da fonteWith the rapid development of technologies and the acquisition of big data, methods capable of processing data sequentially (online) have become indispensable. Among these methods, stochastic gradient algorithms have been established for estimating the minimizer of a function expressed as the expectation of a random function. Although they have become essential, these algorithms encounter difficulties when the problem is ill-conditioned. In this thesis, we focus on second-order stochastic algorithms, such as those of the Newton type, and their applications to various statistical and optimization problems. After establishing theoretical foundations and exposing the motivations that lead us to explore stochastic Newton algorithms, we develop the various contributions of this thesis. The first contribution concerns the study and development of stochastic Newton algorithms for ridge linear regression and ridge logistic regression. These algorithms are based on the Riccati formula (Sherman-Morrison) to recursively estimate the inverse of the Hessian. As the acquisition of big data is generally accompanied by a contamination of the latter, in a second contribution, we focus on the online estimation of the geometric median, which is a robust indicator, i.e., not very sensitive to the presence of atypical data. More specifically, we propose a new stochastic Newton estimator to estimate the geometric median. In the first two contributions, the estimators of the Hessians' inverses are constructed using the Riccati formula, but this is only possible for certain functions. Thus, our third contribution introduces a new Robbins-Monro type method for online estimation of the Hessian's inverse, allowing us then to develop universal stochastic Newton algorithms. Finally, our last contribution focuses on Full Adagrad type algorithms, where the difficulty lies in the fact that there is an adaptive step based on the square root of the inverse of the gradient's covariance. We thus propose a Robbins-Monro type algorithm to estimate this matrix, allowing us to propose a recursive approach for Full AdaGrad and its streaming version, with reduced computational costs. For all the new estimators we propose, we establish their convergence rates as well as their asymptotic efficiency. Moreover, we illustrate the efficiency of these algorithms using numerical simulations and by applying them to real data
Robert, Philippe. "Réseaux Stochastiques et Algorithmes". Habilitation à diriger des recherches, Université Pierre et Marie Curie - Paris VI, 2006. http://tel.archives-ouvertes.fr/tel-00166813.
Texto completo da fonteles chaînes de caractères. L'étude des réseaux avec un grand nombre de noeuds (limite thermodynamique) ou avec des liens de capacité très grandes (Régime limite de Kelly) est discuté ainsi que les perspectives de ce type d'étude.
L'étude mathématique de plusieurs algorithmes distribués gérant des réseaux stochastiques est présentée. L'accent est mis sur les méthodes probabilistes utilisées dans un contexte qui n'est pas nécessairement probabiliste, elles permettent notamment d'étendre et de simplifier une partie des résultats connus dans ce domaine.
Aguech, Rafik. "Algorithmes stochastiques : etude asymptotique de l'erreur". Paris 6, 2000. http://www.theses.fr/2000PA066004.
Texto completo da fonteArouna, Bouhari. "Méthodes de Monté Carlo et algorithmes stochastiques". Marne-la-vallée, ENPC, 2004. https://pastel.archives-ouvertes.fr/pastel-00001269.
Texto completo da fontePELLETIER, MARIANE. "Sur le comportement asymptotique des algorithmes stochastiques". Paris 11, 1996. http://www.theses.fr/1996PA112380.
Texto completo da fonteTouijar, Driss. "Algorithmes stochastiques et application à l'imagerie électromagnétique baysienne". Lille 1, 1994. http://www.theses.fr/1994LIL10123.
Texto completo da fonteKrueger, Martin. "Méthode d'analyse d'algorithmes d'optimisation stochastiques à l'aide d'algorithmes génétiques /". Paris : École nationale supérieure des télécommunications, 1994. http://catalogue.bnf.fr/ark:/12148/cb35707870b.
Texto completo da fonteMonmarché, Pierre. "Hypocoercivité : approches alternatives et applications aux algorithmes stochastiques". Toulouse 3, 2014. http://thesesups.ups-tlse.fr/2615/.
Texto completo da fonteSome Markov dynamics are considered to sample Gibbs laws in the framework of the simulated annealing algorithm, as possible alternatives to the usual reversible diffusion. The problem of the convergence, at fixed temperature, of these processes toward their equilibrium leads to hypococercivity questions. Since the previous results in this field do not yield sharp asymptotics for the convergence rate at low temperature, new methods are investigated, in particular in the case of piecewise deterministic Markov processes. Finally an optimal condition is given for the cooling
Rochet, Sophie. "Convergence des algorithmes génétiques : modèles stochastiques et épistasie". Aix-Marseille 1, 1998. http://www.theses.fr/1998AIX11032.
Texto completo da fontePREVOST, DONALD. "Retines artificielles stochastiques : algorithmes et mise en uvre". Paris 11, 1995. http://www.theses.fr/1995PA112505.
Texto completo da fonteGodichon-Baggioni, Antoine. "Algorithmes stochastiques pour la statistique robuste en grande dimension". Thesis, Dijon, 2016. http://www.theses.fr/2016DIJOS053/document.
Texto completo da fonteThis thesis focus on stochastic algorithms in high dimension as well as their application in robust statistics. In what follows, the expression high dimension may be used when the the size of the studied sample is large or when the variables we consider take values in high dimensional spaces (not necessarily finite). In order to analyze these kind of data, it can be interesting to consider algorithms which are fast, which do not need to store all the data, and which allow to update easily the estimates. In large sample of high dimensional data, outliers detection is often complicated. Nevertheless, these outliers, even if they are not many, can strongly disturb simple indicators like the mean and the covariance. We will focus on robust estimates, which are not too much sensitive to outliers.In a first part, we are interested in the recursive estimation of the geometric median, which is a robust indicator of location which can so be preferred to the mean when a part of the studied data is contaminated. For this purpose, we introduce a Robbins-Monro algorithm as well as its averaged version, before building non asymptotic confidence balls for these estimates, and exhibiting their $L^{p}$ and almost sure rates of convergence.In a second part, we focus on the estimation of the Median Covariation Matrix (MCM), which is a robust dispersion indicator linked to the geometric median. Furthermore, if the studied variable has a symmetric law, this indicator has the same eigenvectors as the covariance matrix. This last property represent a real interest to study the MCM, especially for Robust Principal Component Analysis. We so introduce a recursive algorithm which enables us to estimate simultaneously the geometric median, the MCM, and its $q$ main eigenvectors. We give, in a first time, the strong consistency of the estimators of the MCM, before exhibiting their rates of convergence in quadratic mean.In a third part, in the light of the work on the estimates of the median and of the Median Covariation Matrix, we exhibit the almost sure and $L^{p}$ rates of convergence of averaged stochastic gradient algorithms in Hilbert spaces, with less restrictive assumptions than in the literature. Then, two applications in robust statistics are given: estimation of the geometric quantiles and application in robust logistic regression.In the last part, we aim to fit a sphere on a noisy points cloud spread around a complete or truncated sphere. More precisely, we consider a random variable with a truncated spherical distribution, and we want to estimate its center as well as its radius. In this aim, we introduce a projected stochastic gradient algorithm and its averaged version. We establish the strong consistency of these estimators as well as their rates of convergence in quadratic mean. Finally, the asymptotic normality of the averaged algorithm is given
Tarrès, Pierre. "Pièges des algorithmes stochastiques et marches aléatoires renforcées par sommets". Cachan, Ecole normale supérieure, 2001. http://www.theses.fr/2001DENS0026.
Texto completo da fonteSaadane, Sofiane. "Algorithmes stochastiques pour l'apprentissage, l'optimisation et l'approximation du régime stationnaire". Thesis, Toulouse 3, 2016. http://www.theses.fr/2016TOU30203/document.
Texto completo da fonteIn this thesis, we are studying severa! stochastic algorithms with different purposes and this is why we will start this manuscript by giving historicals results to define the framework of our work. Then, we will study a bandit algorithm due to the work of Narendra and Shapiro whose objectif was to determine among a choice of severa! sources which one is the most profitable without spending too much times on the wrong orres. Our goal is to understand the weakness of this algorithm in order to propose an optimal procedure for a quantity measuring the performance of a bandit algorithm, the regret. In our results, we will propose an algorithm called NS over-penalized which allows to obtain a minimax regret bound. A second work will be to understand the convergence in law of this process. The particularity of the algorith is that it converges in law toward a non-diffusive process which makes the study more intricate than the standard case. We will use coupling techniques to study this process and propose rates of convergence. The second work of this thesis falls in the scope of optimization of a function using a stochastic algorithm. We will study a stochastic version of the so-called heavy bali method with friction. The particularity of the algorithm is that its dynamics is based on the ali past of the trajectory. The procedure relies on a memory term which dictates the behavior of the procedure by the form it takes. In our framework, two types of memory will investigated : polynomial and exponential. We will start with general convergence results in the non-convex case. In the case of strongly convex functions, we will provide upper-bounds for the rate of convergence. Finally, a convergence in law result is given in the case of exponential memory. The third part is about the McKean-Vlasov equations which were first introduced by Anatoly Vlasov and first studied by Henry McKean in order to mode! the distribution function of plasma. Our objective is to propose a stochastic algorithm to approach the invariant distribution of the McKean Vlasov equation. Methods in the case of diffusion processes (and sorne more general pro cesses) are known but the particularity of McKean Vlasov process is that it is strongly non-linear. Thus, we will have to develop an alternative approach. We will introduce the notion of asymptotic pseudotrajectory in odrer to get an efficient procedure
LADELLI, LUCIA. "Theoremes limites pour les chaines de markov : application aux algorithmes stochastiques". Paris 6, 1989. http://www.theses.fr/1989PA066283.
Texto completo da fonteChèze, Guillaume. "Des méthodes symboliques-numériques et exactes pour la factorisation absolue des polynômes en deux variables". Nice, 2004. http://www.theses.fr/2004NICE4092.
Texto completo da fonteThis thesis is about absolute factorization algorithms. The first chapter is a survey on this topic, and then we describe our contributions. They are divided in two parts. The first part corresponds to the symbolic-numeric study. We give a method which allows us to get an exact absolute factorization from an approximate one. Next, this method is used to get an absolute polynomial factorization algorithm. This algorithm uses ideas developped by A. Galligo, D. Rupprecht, and M. Van Hoeij. Thanks to the LLL algorithm, it allows us to get the absolute factorization for polynomials with big degree (bigger than 100). It was impossible before. In the second part, we give two algorithms. The first one adapt a "lifting and recombination'' scheme to the Gao's algorithm. We get then a new algorithm with a better complexity than the Gao's one. The second one is a Las Vegas algorithm. It is an absolute irreducibility test for polynomials with integer coefficients. This algorithm use modular computations, and the shape of the Newton's polytope
Lelong, Jérôme. "Étude asymptotique des algorithmes stochastiques et calcul du prix des options parisiennes". Phd thesis, Ecole des Ponts ParisTech, 2007. http://pastel.archives-ouvertes.fr/pastel-00003310.
Texto completo da fonteLelong, Jérôme. "Etude asymptotique des algorithmes stochastiques et calcul des prix des options Parisiennes". Phd thesis, Ecole Nationale des Ponts et Chaussées, 2007. http://tel.archives-ouvertes.fr/tel-00201373.
Texto completo da fonteLa seconde partie de cette thèse s'intéresse à l'évaluation des options parisiennes en s'appuyant sur les travaux de Chesney, Jeanblanc et Yor. La méthode d'évaluation se base sur l'obtention de formules fermées pour les transformées de Laplace des prix par rapport à la maturité. Nous établissons ces formules pour les options parisiennes simple et double barrières. Nous étudions ensuite une méthode d'inversion numérique de ces transformées dont nous établissons la précision.
Peel, Thomas. "Algorithmes de poursuite stochastiques et inégalités de concentration empiriques pour l'apprentissage statistique". Thesis, Aix-Marseille, 2013. http://www.theses.fr/2013AIXM4769/document.
Texto completo da fonteThe first part of this thesis introduces new algorithms for the sparse encoding of signals. Based on Matching Pursuit (MP) they focus on the following problem : how to reduce the computation time of the selection step of MP. As an answer, we sub-sample the dictionary in line and column at each iteration. We show that this theoretically grounded approach has good empirical performances. We then propose a bloc coordinate gradient descent algorithm for feature selection problems in the multiclass classification setting. Thanks to the use of error-correcting output codes, this task can be seen as a simultaneous sparse encoding of signals problem. The second part exposes new empirical Bernstein inequalities. Firstly, they concern the theory of the U-Statistics and are applied in order to design generalization bounds for ranking algorithms. These bounds take advantage of a variance estimator and we propose an efficient algorithm to compute it. Then, we present an empirical version of the Bernstein type inequality for martingales by Freedman [1975]. Again, the strength of our result lies in the variance estimator computable from the data. This allows us to propose generalization bounds for online learning algorithms which improve the state of the art and pave the way to a new family of learning algorithms taking advantage of this empirical information
Gavra, Iona Alexandra. "Algorithmes stochastiques d'optimisation sous incertitude sur des structures complexes : convergence et applications". Thesis, Toulouse 3, 2017. http://www.theses.fr/2017TOU30141/document.
Texto completo da fonteThe main topics of this thesis involve the development of stochastic algorithms for optimization under uncertainty, the study of their theoretical properties and applications. The proposed algorithms are modified versions of simulated an- nealing that use only unbiased estimators of the cost function. We study their convergence using the tools developed in the theory of Markov processes: we use properties of infinitesimal generators and functional inequalities to measure the distance between their probability law and a target one. The first part is concerned with quantum graphs endowed with a probability measure on their vertex set. Quantum graphs are continuous versions of undirected weighted graphs. The starting point of the present work was the question of finding Fréchet means on such a graph. The Fréchet mean is an extension of the Euclidean mean to general metric spaces and is defined as an element that minimizes the sum of weighted square distances to all vertices. Our method relies on a Langevin formulation of a noisy simulated annealing dealt with using homogenization. In order to establish the convergence in probability of the process, we study the evolution of the relative entropy of its law with respect to a convenient Gibbs measure. Using functional inequalities (Poincare and Sobolev) and Gronwall's Lemma, we then show that the relative entropy goes to zero. We test our method on some real data sets and propose an heuristic method to adapt the algorithm to huge graphs, using a preliminary clustering. In the same framework, we introduce a definition of principal component analysis for quantum graphs. This implies, once more, a stochastic optimization problem, this time on the space of the graph's geodesics. We suggest an algorithm for finding the first principal component and conjecture the convergence of the associated Markov process to the wanted set. On the second part, we propose a modified version of the simulated annealing algorithm for solving a stochastic global optimization problem on a finite space. Our approach is inspired by the general field of Monte Carlo methods and relies on a Markov chain whose probability transition at each step is defined with the help of mini batches of increasing (random) size. We prove the algorithm's convergence in probability towards the optimal set, provide convergence rate and its optimized parametrization to ensure a minimal number of evaluations for a given accuracy and a confidence level close to 1. This work is completed with a set of numerical experiments and the assessment of the practical performance both on benchmark test cases and on real world examples
Le, Coz Mathieu. "Evaluation de performances des algorithmes liés aux télécommunications". Versailles-St Quentin en Yvelines, 2004. http://www.theses.fr/2004VERS0017.
Texto completo da fonteA classical approach in performance evaluation consists in computing performance indices on a specific system. One may modelise the system, computes the stationnary distribution vector (resolution step) and evaluate the performance indices on this vector. The resolution of the model leads to find the solution of a linear system of equations. The size of this linear system is the size of the model reachable state space. An important difficulty is due to this size since it can be very large (up to several million states). The work of this thesis consists in transforming the initial model so that the resolution of the modified model is faster than the resolution of the initial one. The first part of this thesis consists in transforming the initial model by using techniques of stochastic bounds. During the computation of this bound, we force it to have a particular structure. This structure can be used to reduce time resolution. In particular, algorithm LIMSUB computes a stochastic bound according to a fixed partition of the states by using the definition of strong aggregation. The second part deals with the stochastic automata network (SAN) formalism. It shows how to group automata together before the resolution step. This grouping operation allows to reduce the time resolution of the SAN. The problem of choosing the best possible groups under a given constraint of memory is formulated and is prooved to be an NP-complete problem. An algorithm computing an aggregated bound of a system modelled by a stochastic automata network is then given and some develloped programs are presented
Naja, Dania. "Convergence d'algorithmes stochastiques : 1-recuit simulé sans potentiel. 2-quelques algorithmes d'auto-organisation". Nancy 1, 1998. http://www.theses.fr/1998NAN10261.
Texto completo da fonteHuang, Lorick. "EDS dirigées par des processus stables : Méthode paramétrix pour des estimées de densités et application aux algorithmes stochastiques". Sorbonne Paris Cité, 2015. https://theses.hal.science/tel-01180708.
Texto completo da fonteDensity Estimates for SDEs Driven by Tempered Stable Processes. We study a class of stochastic differential equations driven by a possibly tempered Lévy process, under mild conditions on the coefficients (Wilder continuity). We prove the well-posedness of the associated martingale problem as well as the existence of the density of the solution. Two sided heat kernel estimates are given as well. Our approach is based on the Parametrix series expansion. A Parametrix Approach for some Degenerate Stable Driven SDEs. We consider a stable driven degenerate stochastic differential equation, whose coeffi- cients satisfy a kind of weak Hôrmander condition. Under mild smoothness assumptions we prove the uniqueness of the martingale problem for the associated generator under some dimension constraints. Also, when the driving noise is scalar and tempered, we establish density bounds reflecting the multi-scale behavior of the process. A Multi-step Richardson-Romberg extrapolation method for stochastic approximation We obtain an expansion of the implicit weak discretization error for the target of stochastic approximation algorithms introduced and studied in [30]. This allows us to extend and develop the Richardson-Romberg extrapolation method for Monte Carlo linear estimator (introduced in [79] and deeply studied in [65]) to the framework of stochastic optimization by means of stochastic approximation algorithms. We notably apply the method to the estimation of the quantile of diffusion processes. Numerical results confirm the theoretical analysis and show a significant reduction in the initial computational cost
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.
Texto completo da fonteThis 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
Nsiri, Saïd. "L'utilisation des représentations markoviennes dans l'identification des processus stochastiques ARMA vectoriels". Paris 7, 1985. http://www.theses.fr/1985PA07F134.
Texto completo da fonteMesghouni, Khaled. "Application des algorithmes évolutionnistes dans les problèmes d'optimisation en ordonnancement de la production". Lille 1, 1999. https://pepite-depot.univ-lille.fr/LIBRE/Th_Num/1999/50376-1999-19.pdf.
Texto completo da fonteStazhynski, Uladzislau. "Discrétisation de processus à des temps d’arrêt et Quantification d'incertitude pour des algorithmes stochastiques". Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLX088/document.
Texto completo da fonteThis thesis consists of two parts which study two separate subjects. Chapters 1-4 are devoted to the problem of processes discretization at stopping times. In Chapter 1 we study the optimal discretization error of stochastic integrals, driven by a multidimensional continuous Brownian semimartingale. In this setting we establish a path wise lower bound for the renormalized quadratic variation of the error and we provide a sequence of discretization stopping times, which is asymptotically optimal. The latter is defined as hitting times of random ellipsoids by the semimartingale at hand. In comparison with previous available results, we allow a quite large class of semimartingales and we prove that the asymptotic lower bound is attainable. In Chapter 2 we study the model-adaptive optimal discretization error of stochastic integrals. In Chapter 1 the construction of the optimal strategy involved the knowledge about the diffusion coefficient of the semimartingale under study. In this work we provide a model-adaptive asymptotically optimal discretization strategy that does not require any prior knowledge about the model. In Chapter 3 we study the convergence in distribution of renormalized discretization errors of Ito processes for a concrete general class of random discretization grids given by stopping times. Previous works on the subject only treat the case of dimension 1. Moreover they either focus on particular cases of grids, or provide results under quite abstract assumptions with implicitly specified limit distribution. At the contrast we provide explicitly the limit distribution in a tractable form in terms of the underlying model. The results hold both for multidimensional processes and general multidimensional error terms. In Chapter 4 we study the problem of parametric inference for diffusions based on observations at random stopping times. We work in the asymptotic framework of high frequency data over a fixed horizon. Previous works on the subject consider only deterministic, strongly predictable or random, independent of the process, observation times, and do not cover our setting. Under mild assumptions we construct a consistent sequence of estimators, for a large class of stopping time observation grids. Further we carry out the asymptotic analysis of the estimation error and establish a Central Limit Theorem (CLT) with a mixed Gaussian limit. In addition, in the case of a 1-dimensional parameter, for any sequence of estimators verifying CLT conditions without bias, we prove a uniform a.s. lower bound on the asymptotic variance, and show that this bound is sharp. In Chapters 5-6 we study the problem of uncertainty quantification for stochastic approximation limits. In Chapter 5 we analyze the uncertainty quantification for the limit of a Stochastic Approximation (SA) algorithm. In our setup, this limit is defined as the zero of a function given by an expectation. The expectation is taken w.r.t. a random variable for which the model is assumed to depend on an uncertain parameter. We consider the SA limit as a function of this parameter. We introduce the so-called Uncertainty for SA (USA) algorithm, an SA algorithm in increasing dimension for computing the basis coefficients of a chaos expansion of this function on an orthogonal basis of a suitable Hilbert space. The almost-sure and Lp convergences of USA, in the Hilbert space, are established under mild, tractable conditions. In Chapter 6 we analyse the L2-convergence rate of the USA algorithm designed in Chapter 5.The analysis is non-trivial due to infinite dimensionality of the procedure. Moreover, our setting is not covered by the previous works on infinite dimensional SA. The obtained rate depends non-trivially on the model and the design parameters of the algorithm. Its knowledge enables optimization of the dimension growth speed in the USA algorithm, which is the key factor of its efficient performance
Terrier, Pierre. "Algorithmes stochastiques pour simuler l'évolution microstructurale d'alliages ferritiques : une étude de la dynamique d'amas". Thesis, Paris Est, 2018. http://www.theses.fr/2018PESC1135/document.
Texto completo da fonteWe study ageing of materials at a microstructural level. In particular, defects such as vacancies, interstitials and solute atoms are described by a model called Cluster Dynamics (CD), which characterize the evolution of the concentrations of such defects, on period of times as long as decades. CD is a set of ordinary differential equations (ODEs), which might contain up to hundred of billions of equations. Therefore, classical methods used for solving system of ODEs are not suited in term of efficiency. We first show that CD is well-posed and that physical properties such as the conservation of matter and the preservation of the sign of the solution are verified. We also study an approximation of CD, namely the Fokker--Planck approximation, which is a partial differential equation. We quantify the error between CD and its approximation. We then introduce an algorithm for simulating CD. The algorithm is based on a splitting of the dynamics and couples a deterministic and a stochastic approach of CD. The stochastic approach interprets directly CD as a jump process or its approximation as a Langevin process. The aim is to reduce the number of equations to solve, hence reducing the computation time. We finally apply this algorithm to physical models. The interest of this approach is validated on complex models. Moreover, we show that CD can be efficiently improved thanks to the versatility of the algorithm
Huang, Junbo. "Théorème de Berry-Esseen pour martingales normalisées et algorithmes stochastiques : application en contrôle stochastique". Paris 6, 2009. http://www.theses.fr/2009PA066174.
Texto completo da fonteBouttier, Clément. "Optimisation globale sous incertitude : algorithmes stochastiques et bandits continus avec application aux performances avion". Thesis, Toulouse 3, 2017. http://www.theses.fr/2017TOU30176.
Texto completo da fonteThis PhD thesis is dedicated to the theoretical and numerical analysis of stochastic algorithms for the stochastic flight planning problem. Optimizing the fuel consumption and flight time is a key factor for airlines to be competitive. These companies thus look for flight optimization tools with higher and higher accuracy requirements. However, nowadays available methodologies for flight planning are based on simplified aircraft performance models. In this PhD, we propose to fulfill the accuracy requirements by adapting our methodology to both the constraints induced by the utilization of an industrial aircraft performance computation code and the consideration of the uncertainty about the real flight conditions, i.e., air traffic and weather conditions. Our proposal is supported by three main contributions. First, we design a numerical framework for benchmarking aircraft trajectory optimization tools. This provides us a unified testing procedure for all aircraft performance models. Second, we propose and study (both theoretically and numerically) two global derivative-free algorithms for stochastic optimization problems. The first approach, the NSA algorithm, is highly generic and does not use any prior knowledge about the aircraft performance model. It is an extension of the simulated annealing algorithm adapted to noisy cost functions. We provide an upper bound on the convergence speed of NSA to globally optimal solutions. The second approach, the SPY algorithm, is a Lipschitz bandit algorithm derived from Piyavskii's algorithm. It is more specific as it requires the knowledge of some Lipschitz regularity property around the optimum, but it is therefore far more efficient. We also provide a theoretical study of this algorithm through an upper bound on its simple regret
Segalat, Philippe. "Méthodes de points intérieurs et de quasi-Newton". Limoges, 2002. http://www.theses.fr/2002LIMO0041.
Texto completo da fonteThis thesis is interested in interior point and quasi-Newton methods in nonlinear optimization and with their implementation. One presents the code NOPTIQ using the limited memory BFGS formulas to solve large scale problems. The originality of this approach is the use of these formulas within the framework of interior point methods. The storage requirement and the computing cost of one iteration are then low. One shows that NOPTIQ is robust and that its performance are comparable with the reference codes 1-BFGS-B and LANCELOT. One presents also an infeasible algorithm using the preceding methods to solve a nonlinear problem with inequality constraints and linear equality constraints. The idea to penalize the problem using shift variables and a variant of the big-M method of linear programming. The q-superlinear convergence of the inner iterates and the global convergence of outer iterates are shown
Hajji, Omessaad Brochet Pascal. "Contribution au développement de méthodes d'optimisation stochastiques application à la conception des dispositifs électrotechniques /". [S.l.] : [s.n.], 2003. http://www.univ-lille1.fr/bustl-grisemine/pdf/extheses/50376-2003-237-238.pdf.
Texto completo da fonteBossy, Mireille. "Vitesse de convergence d'algorithmes particulaires stochastiques et application à l'équation de Burgers". Aix-Marseille 1, 1995. http://www.theses.fr/1995AIX11007.
Texto completo da fonteSidaner, Alain. "Contribution à l'étude de l'algorithme de recherche locale stochastique WalkSAT sur des instances SAT aléatoires". Dijon, 2003. http://www.theses.fr/2003DIJOS044.
Texto completo da fonteXayaphoummine, Alain. "Simulations et expériences sur le repliement de l'ARN : prédictions statistiques des pseudonœuds in silico et réalisation de commutateurs ARN par transcription in vitro". Université Louis Pasteur (Strasbourg) (1971-2008), 2004. https://publication-theses.unistra.fr/public/theses_doctorat/2004/XAYAPHOUMMINE_Alain_2004.pdf.
Texto completo da fonteTong, Dinh Quy. "Méthodes d'estimation de paramètres de modèles autorégressifs multivariés". Grenoble 1, 1991. http://www.theses.fr/1991GRE10141.
Texto completo da fonteFijany, Amir. "Algorithmes et architectures parallèles en robotique". Paris 11, 1988. http://www.theses.fr/1988PA112258.
Texto completo da fonteDe, Carvalho Gomes Fernando. "Utilisation d'algorithmes stochastiques en apprentissage". Montpellier 2, 1992. http://www.theses.fr/1992MON20254.
Texto completo da fonteSEGALAT, Philippe. "Méthodes de Points Intérieurs et de quasi-Newton". Phd thesis, Université de Limoges, 2002. http://tel.archives-ouvertes.fr/tel-00005478.
Texto completo da fonteReutenauer, Victor. "Algorithmes stochastiques pour la gestion du risque et l'indexation de bases de données de média". Thesis, Université Côte d'Azur (ComUE), 2017. http://www.theses.fr/2017AZUR4018/document.
Texto completo da fonteThis thesis proposes different problems of stochastic control and optimization that can be solved only thanks approximation. On one hand, we develop methodology aiming to reduce or suppress approximations to obtain more accurate solutions or something exact ones. On another hand we develop new approximation methodology in order to solve quicker larger scale problems. We study numerical methodology to simulated differential equations and enhancement of computation of expectations. We develop quantization methodology to build control variate and gradient stochastic methods to solve stochastic control problems. We are also interested in clustering methods linked to quantization, and principal composant analysis or compression of data thanks neural networks. We study problems motivated by mathematical finance, like stochastic control for the hedging of derivatives in incomplete market but also to manage huge databases of media commonly known as big Data in chapter 5. Theoretically we propose some upper bound for convergence of the numerical method used. This is the case of optimal hedging in incomplete market in chapter 3 but also an extension of Beskos-Roberts methods of exact simulation of stochastic differential equations in chapter 4. We present an original application of karhunen-Loève decomposition for a control variate of computation of expectation in chapter 2
Arouna, Bouhari. "Algotithmes stochastiques et méthodes de Monte Carlo". Phd thesis, Ecole des Ponts ParisTech, 2004. http://pastel.archives-ouvertes.fr/pastel-00001269.
Texto completo da fonteGuo, Chaomei. "Amélioration des propriétés de convergence des algorithmes de simulation des circuits non linéaires microondes". Limoges, 1995. http://www.theses.fr/1995LIMO0024.
Texto completo da fonteHo, Zhen Wai Olivier. "Contributions aux algorithmes stochastiques pour le Big Data et à la théorie des valeurs extrèmes multivariés". Thesis, Bourgogne Franche-Comté, 2018. http://www.theses.fr/2018UBFCD025/document.
Texto completo da fonteThis thesis in divided in two parts. The first part studies models for multivariate extremes. We give a method to construct multivariate regularly varying random vectors. The method is based on a multivariate extension of a Breiman Lemma that states that a product $RZ$ of a random non negative regularly varying variable $R$ and a non negative $Z$ sufficiently integrable is also regularly varying. Replacing $Z$ with a random vector $mathbf{Z}$, we show that the product $Rmathbf{Z}$ is regularly varying and we give a characterisation of its limit measure. Then, we show that taking specific distributions for $mathbf{Z}$, we obtain classical max-stable models. We extend our result to non-standard regular variations. Next, we show that the Pareto model associated with the Hüsler-Reiss max-stable model forms a full exponential family. We show some properties of this model and we give an algorithm for exact simulation. We study the properties of the maximum likelihood estimator. Then, we extend our model to non-standard regular variations. To finish the first part, we propose a numerical study of the Hüsler-Reiss Pareto model.In the second part, we start by giving a lower bound of the smallest singular value of a matrix perturbed by appending a column. Then, we give a greedy algorithm for feature selection and we illustrate this algorithm on a time series dataset. Secondly, we show that an incoherent matrix satisfies a weakened version of the NSP property. Thirdly, we study the problem of column selection of $Xinmathbb{R}^{n imes p}$ given a coherence threshold $mu$. This means we want the largest submatrix satisfying some coherence property. We formulate the problem as a linear program with quadratic constraint on ${0,1}^p$. Then, we consider a relaxation on the sphere and we bound the relaxation error. Finally, we study the projected stochastic gradient descent for online PCA. We show that in expectation, the algorithm converges to a leading eigenvector and we suggest an algorithm for step-size selection. We illustrate this algorithm with a numerical experiment
Kosuch, Stefanie. "Stochastic Optimization Problems with Knapsack Constraint". Paris 11, 2010. http://www.theses.fr/2010PA112154.
Texto completo da fonteGiven a set of objects each having a particular weight and value. The knapsack problem consists of choosing among these items a subset such that (i) the total weight of the chosen items does respect a given weight constraint (the capacity of the knapsack) and (ii) the total value of the chosen items is maximized. In this thesis, we study four stochastic optimization problems with knapsack constraint: the simple recourse knapsack problem, the chance-constrained knapsack problem, the two-stage knapsack problem and a bilievel problem with knapsack chance-constraint. All problems have in common that the item weights in the knapsack constraints are assumed to be random. We propose to solve the simple recourse and the chance-constrained knapsack problems using a branch-&-bound algorithm as framework. Upper bounds are obtained by solving relaxed, i. E. Continuous sub-problems. The latter is done by applying a stochastic gradient algorithm. Concerning the two-stage knapsack problem, we treat, in the first instance, the model where the item weights are assumed to be normally distributed and propose upper and lower bounds on the optimal solution value. Then, we study the problem with discretely distributed weights and show that its deterministic equivalent reformulation does not admit a constant factor approximation algorithm unless P=NP. The studied bilevel problem with knapsack chance-constraint is first of all reformulated as a deterministic equivalent bilinear problem. As the latter is generally hard to solve exactly, we propose to solve a relaxation using a novel iterative algorithm
Cénac, Peggy. "Récursivité au carrefour de la modélisation de séquences, des arbres aléatoires, des algorithmes stochastiques et des martingales". Habilitation à diriger des recherches, Université de Bourgogne, 2013. http://tel.archives-ouvertes.fr/tel-00954528.
Texto completo da fonteLèbre, Sophie. "Analyse de processus stochastiques pour la génomique : étude du modèle MTD et inférence de réseaux bayésiens dynamiques". Evry-Val d'Essonne, 2007. http://www.biblio.univ-evry.fr/theses/2007/interne/2007EVRY0017.pdf.
Texto completo da fonteThis thesis deals with DNA sequence and time series gene expression analysis. First we study the parsimonious Markov model called Mixture Transition Distribution (MTD) model and introduce an EM algorithm for MTD models estimation. Then we propose two approaches for genetic network recovering using Dynamic Bayesian Networks (DBNs). The dependencies are described by a directed graph whose topology has to be inferred despite the overly low number of repeated measurements compared with the number of observed genes. First we assume that the topology is constant across time, we approximate this graph by considering partial order dependencies and we develop a deterministic procedure for DBNs inference. Then we consider a multiple changepoint regression model defining a succession of homogeneous phases. The changepoints location and the structure within each phase are simultaneously inferred thanks to a reversible jump MCMC procedure
Xu, Jian. "Modèles stochastiques évolutionnaires pour la gestion de tournées de véhicules avec fenêtres de temps souples et demandes floues". Artois, 2007. http://www.theses.fr/2007ARTO0202.
Texto completo da fonteThe work of this thesis considers the vehicle routing problem with time windows and fuzzy demands (VRPTWFD). The goal of the problem is to find the routes of the vehicles with a minimal cost so that the vehicles can service each client exactly only once respecting some constraints. The customers specify their demands by a fuzzy number. The VRPTWFD is studied as the static case as well as the dynamic case. We use the possibility theory to handle the constraint of capacity by setting certain thresholds for the degrees of the possibility and necessity. Using this capacity constraint, a chance constrained programming model (CCP) and a two-stages stochastic programming with recourse model (SPR) in stochastic programming were proposed to treat the VRPTWFD. The genetic algorithms integrating these models have been proposed as the optimization approach in order to find the optimal solutions. In the dynamic VRPTWFD, some customers can call in their orders during the daily operation. A simulation platform, which has the capability of simulating the daily operation, has been developed to solve online the dynamic VRPTWFD. In order to assess the performance of the proposed models, we have constructed a benchmark for the static VRPTWFD and a benchmark for the dynamic VRPTWFD adapting from Solomon’s benchmark for the VRPTW, then we have evaluated the quality of the solutions, which were obtained by using these models, by simulating the real world situations with the help of the “test” scenarios
Gratton, Serge. "Outils théoriques d'analyse du calcul à précision finie". Toulouse, INPT, 1998. http://www.theses.fr/1998INPT015H.
Texto completo da fonteRey, 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.
Texto completo da fontePourzandi, Makan. "Etude de l'impact des recouvrements calcul-communication sur des algorithmes parallèles de calcul matriciel". Lyon 1, 1995. http://www.theses.fr/1995LYO19001.
Texto completo da fonteMárquez, David. "Vitesse du recuit simulé vectoriel continu ou discrétisé et algorithmes stochastiques en problèmes inverses bayésiens avec applications à la géophysique". Marne-la-Vallée, 1998. http://www.theses.fr/1998MARN0024.
Texto completo da fonte