Dissertations / Theses on the topic 'Monte-Carlo (Méthode de)'
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 'Monte-Carlo (Méthode de).'
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.
Cornebise, Julien. "Méthodes de Monte Carlo séquentielles adaptatives." Paris 6, 2009. http://www.theses.fr/2009PA066152.
Full textGüçlü, Alev Devrim. "Simulation des dispositifs optoélectroniques par la méthode Monte Carlo." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1999. http://www.collectionscanada.ca/obj/s4/f2/dsk1/tape9/PQDD_0021/MQ54037.pdf.
Full textOunaissi, Daoud. "Méthodes quasi-Monte Carlo et Monte Carlo : application aux calculs des estimateurs Lasso et Lasso bayésien." Thesis, Lille 1, 2016. http://www.theses.fr/2016LIL10043/document.
Full textThe thesis contains 6 chapters. The first chapter contains an introduction to linear regression, the Lasso and the Bayesian Lasso problems. Chapter 2 recalls the convex optimization algorithms and presents the Fista algorithm for calculating the Lasso estimator. The properties of the convergence of this algorithm is also given in this chapter using the entropy estimator and Pitman-Yor estimator. Chapter 3 is devoted to comparison of Monte Carlo and quasi-Monte Carlo methods in numerical calculations of Bayesian Lasso. It comes out of this comparison that the Hammersely points give the best results. Chapter 4 gives a geometric interpretation of the partition function of the Bayesian lasso expressed as a function of the incomplete Gamma function. This allowed us to give a convergence criterion for the Metropolis Hastings algorithm. Chapter 5 presents the Bayesian estimator as the law limit a multivariate stochastic differential equation. This allowed us to calculate the Bayesian Lasso using numerical schemes semi-implicit and explicit Euler and methods of Monte Carlo, Monte Carlo multilevel (MLMC) and Metropolis Hastings algorithm. Comparing the calculation costs shows the couple (semi-implicit Euler scheme, MLMC) wins against the other couples (scheme method). Finally in chapter 6 we found the Lasso convergence rate of the Bayesian Lasso when the signal / noise ratio is constant and when the noise tends to 0. This allowed us to provide a new criteria for the convergence of the Metropolis algorithm Hastings
Arouna, Bouhari. "Algotithmes stochastiques et méthodes de Monte Carlo." Phd thesis, Ecole des Ponts ParisTech, 2004. http://pastel.archives-ouvertes.fr/pastel-00001269.
Full textJaeckel, Alain. "Simulations Monte Carlo de chaînes confinées." Montpellier 2, 1997. http://www.theses.fr/1997MON20206.
Full textChabut, Emmanuel. "Simulation aérothermodynamique en régime d'écoulement raréfié par méthode de Monte-Carlo." Orléans, 2005. http://www.theses.fr/2005ORLE2017.
Full textForster, Simon. "Nouveau matériau semi-conducteur à large bande interdite à base de carbures ternaires - Enquête sur Al4SiC4." Thesis, Université Grenoble Alpes (ComUE), 2019. http://www.theses.fr/2019GREAI095.
Full textWide bandgap semiconductor materials are able to withstand harsh environments and operate over a wide range of temperatures. These make them ideal for many applications such as sensors, high-power and radio-frequencies to name a few.However, more novel materials are required to achieve significant power efficiency of various applications or to develop new applications to complement current wide bandgap semiconductors such as GaN and SiC.In this dissertation, three different methods are used to study one of these novelmaterials, aluminium silicon carbide (Al4SiC4): (1) ensemble Monte Carlo simulationsin order to study the electron transport properties of the novel ternary carbide, (2)experimental studies to determine its material properties, and (3) device simulationsof a heterostructure device made possible by this ternary carbide. All these methodsinterlink with each other. Data from each of them can feed into the other to acquire newresults or refine obtained results thus leading way to attractive electrical properties such as a bandgap of 2.78 eV or a peak drift velocity of 1.35×10 cm s .Ensemble Monte Carlo toolbox, developed in-house for simulations of Si, Ge, GaAs,AlxGa1−xAs, AlAs, and InSb; is adopted for simulations of the ternary carbide by adding anew valley transformation to account for the hexagonal structure of Al4SiC4. We predicta peak electron drift velocity of 1.35×107 cms−1 at electric field of 1400 kVcm−1 and a maximum electron mobility of 82.9 cm V s . We have seen a diffusion constant of 2.14 cm2s−1 at a low electric field and of 0.25 cm2s−1 at a high electric field. Finally, weshow that Al4SiC4 has a critical field of 1831 kVcmsemiconductor crystals are used that had previously been grown at IMGP, one by solution grown and the other by crucible melt. Three different experiments are performed on them; (1) UV, IR and Vis Spectroscopy, (2) X-ray Photo Spectroscopy, and (3) Two- and four-probe measurements where metal contact are grown on the crystals. Here we have found a bandgap of 2.78 ± 0.02 eV UV, IR and Vis Spectroscopy and a thick oxide layer on the samples using XPS. Unfortunately the Two- and four-probe measurements failed to give any results other than noise, most likely due to the thick oxide layer that was found on the samples.In the device simulations, a commercial software Atlas by Silvaco is utilized to predict performance of heterostructure devices, with gates lengths of 5 μm, 2 μm and 1 μm, made possible by the ternary carbide in a combination with SiC. The 5 μm gate length SiC/Al4SiC4 heterostructure transistor delivers a maximum drain current of 1.68×10−4 A/μm, which increases to 2.44×10−4 A/μm and 3.50×10−4 A/μm for gate lengths of 2 μm and 1 μm, respectively. The device breakdown voltage is 59.0 V which reduces to 31.0 V and to 18.0 V for the scaled 2 μm and the 1 μm gate length transistors. The scaled down 1 μm gate length device switches faster because of the higher transconductance of6.51×10−5 S/μmcomparedtoonly1.69×10−6 S/μmforthelargestdevice.Finally,a sub-threshold slope of the scaled devices is 197.3 mV/dec, 97.6 mV/dec, and 96.1 mV/dec for gate lengths of 5 μm, 2 μm, and 1 μm, respectively
Reilhac-Laborde, Anthonin. "Validation et exploitation d'un simulateur TEP de Monte Carlo." Lyon, INSA, 2007. http://theses.insa-lyon.fr/publication/2007ISAL0071/these.pdf.
Full textThe evaluation of algorithms dedicated to process, reconstruct, or analyze PET data is a challenging task. The common strategy is to compare the algorithm output to a controlled gold standard. A part of the difficulty follows from the unavailability of such ground truth with in vivo data. Consequently, validation often relies on the use of simulated data whose geometry and contents are precisely known. This method provides a great flexibility as it allows the use of realistic numerical phantoms. Also, the ability to control the factors that degrade the image formation is a significant advantage. The aim of this PhD research was to conduct first the validation experiments of the simulation model of a Monte Carlo-based PET simulator (PET-SORTEO) and second, to use the simulator as a tool for the design and development of moethods that aim at improving the quantification of the PET data
Maigne, Lydia. "Personnalized dosimetry using GATE Monte Carlo simulations on a grid architecture." Clermont-Ferrand 2, 2005. http://www.theses.fr/2005CLF21607.
Full textSzkolnik, Jean-Jacques. "Application des méthodes de Monte-Carlo séquentielles à l'extraction de trames radar." Brest, 2004. http://www.theses.fr/2004BRES2023.
Full textThe purpose of this study consists in determining an algorithm able to ensure the blind extraction of pulses resulting from the same radar system and to characterize the sequence evolution of their characteristic parameters. We precisely explicit the context and the physical nature of the pulses, the parameters which characterize them and the various parameter modulations taken into account, in order to lead to a radar pulse train state formulation. Then, we deal with the current method limitations implemented on operational equipments and draws up the state of the art of research on the matter through free publications appearing on the subject. We release and justify our own research orientation relying on the application of the particle methods to the previously described problems. We detail Monte-Carlo sequential theories, their limitations and additional techniques used to overcome drawbacks. We stipulate the adaptation of the presented techniques to our problem in order to deduce the formulation of a generic deinterleaving module specified exclusively from pulse times of arrival (TOA). The extraction module adaptation capabilities to news modulations, possibly nonlinear, are improved by integrating the unscented transformation. We extend then the concepts used for the TOA extraction module to new extraction modules specified from the other parameters. We give an example of scenario complexity degree likely to be processed with the association of two extraction modules running for the first from TOA and for the second from an other parameter. Finally the last part is dedicated to a series of extraction module evaluations intended to determine its application field
Dubois, Vincent. "Simulations Monte Carlo de petits agrégats d'argent en solution aqueuse." Paris 11, 2002. http://www.theses.fr/2002PA112164.
Full textWe have studied silver clusters in aqueous solution with the help of Monte Carlo simulations coupled with a Valence-Bond formalism. Silver aggregation has already been investigated, but the mechanism of this aggregation remains poorly known, especially it's the first steps. The aim of this present theoretical work has been to provide informations on the process of aggregation. We have investigated the following two points: The redox potential of the couples Agn(+)/Agn is an essential parameter of the interpretation of numerous reactions, like the photographic development. The potentials of the small clusters are not known. We have evaluated the difference of the redox potential between n=1 and n=2. For the calculation of redox potentials, we have compared three methods of free energy calculation. We have shown that the self-consistent histograms method is the most reliable in the case of mutation from a neutral into a cation. With the help of pulse radiolysis, Heinglein and co-workers have proposed the formation of Ag3(2+). The formation of Ag3(2+) seems mysterious because this cluster displays a large charge repulsion and only one valence electron. We have studied Ag3(2+) in water with geometrical constraints to limit the complexity of the potential. These preliminary works show that on the one hand this cluster is metastable in solution and on the other hand that the formation energy barrier is important. Therefore the reaction Ag(+) + Ag2(+) --> Ag3(2+) seems impossible at room temperature. However, simulations without any constraints may change this conclusion
Gilquin, Laurent. "Échantillonnages Monte Carlo et quasi-Monte Carlo pour l'estimation des indices de Sobol' : application à un modèle transport-urbanisme." Thesis, Université Grenoble Alpes (ComUE), 2016. http://www.theses.fr/2016GREAM042/document.
Full textLand Use and Transportation Integrated (LUTI) models have become a norm for representing the interactions between land use and the transportation of goods and people in a territory. These models are mainly used to evaluate alternative planning scenarios, simulating their impact on land cover and travel demand.LUTI models and other mathematical models used in various fields are most of the time based on complex computer codes. These codes often involve poorly-known inputs whose uncertainty can have significant effects on the model outputs.Global sensitivity analysis methods are useful tools to study the influence of the model inputs on its outputs. Among the large number of available approaches, the variance based method introduced by Sobol' allows to calculate sensitivity indices called Sobol' indices. These indices quantify the influence of each model input on the outputs and can detect existing interactions between inputs.In this framework, we favor a particular method based on replicated designs of experiments called replication method. This method appears to be the most suitable for our application and is advantageous as it requires a relatively small number of model evaluations to estimate first-order or second-order Sobol' indices.This thesis focuses on extensions of the replication method to face constraints arising in our application on the LUTI model Tranus, such as the presence of dependency among the model inputs, as far as multivariate outputs.Aside from that, we propose a recursive approach to sequentially estimate Sobol' indices. The recursive approach is based on the iterative construction of stratified designs, latin hypercubes and orthogonal arrays, and on the definition of a new stopping criterion. With this approach, more accurate Sobol' estimates are obtained while recycling previous sets of model evaluations. We also propose to combine such an approach with quasi-Monte Carlo sampling.An application of our contributions on the LUTI model Tranus is presented
Brunet, Charles. "Parallélisation des algorithmes de Monte-Carlo multicanoniques." Thesis, Université Laval, 2012. http://www.theses.ulaval.ca/2012/29397/29397.pdf.
Full textEl, Haddad Rami. "Méthodes quasi-Monte Carlo de simulation des chaînes de Markov." Chambéry, 2008. http://www.theses.fr/2008CHAMS062.
Full textMonte Carlo (MC) methods are probabilistic methods based on the use of random numbers in repeated simulations to estimate some parameter. Their deterministic versions are called Quasi-Monte Carlo (QMC) methods. The idea is to replace pseudo-random points by deterministic quasi-random points (also known as low-discrepancy point sets or sequences). In this work, we propose and analyze QMC-based algorithms for the simulation of multidimensional Markov chains. The quasi-random points we use are (T,S)-sequences in base B. After recalling the principles of MC and QMC methods and their main properties, we introduce some plain financial models, to serve in the following as numerical examples to test the convergence of the proposed schemes. We focus on problems where the exact solution is known, in order to be able to compute the error and to compare the efficiency of the various schemes In a first part, we consider discrete-time Markov chains with S-dimensional state spaces. We propose an iterative QMC scheme for approximating the distribution of the chain at any time. The scheme uses a (T,S+1)-sequence in base b for the transitions. Additionally, one needs to re-order the copies of the chain according to their successive components at each time-step. We study the convergence of the scheme by making some assumptions on the transition matrix. We assess the accuracy of the QMC algorithm through financial examples. The results show that the new technique is more efficient than the traditional MC approach. Then, we propose a QMC algorithm for the simulation of Markov chains with multidimensional continuous state spaces. The method uses the same re-ordering step as in the discrete setting. We provide convergence results in the case of one dimensional chains and then in the case of multidimensional chains, by making additional assumptions. We illustrate the convergence of the algorithm through numerical experiments. The results show that the new method converges faster than the MC algorithm. In the last part, we consider the problem of the diffusion equation in a spatially nonhomogeneous medium. We use a random walk algorithm, in conjunction with a correction of the Gaussian Steplength. We write a QMC variant of the algorithm, by adapting the principles seen for the simulation of the Markov chains. We test the method in dimensions 1, 2 and 3 on a problem involving the diffusion of calcium ions in a biological medium. In all the simulations, the results of QMC computations show a strong improvement over MC outcomes. Finally, we give some perspectives and directions for future work
Chabot, Martin. "Le Bootstrap comme méthode d'estimation du r de Pearson, une étude Monte Carlo." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1997. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp05/mq25286.pdf.
Full textRimmel, Arpad. "Improvements and Evaluation of the Monte Carlo Tree Search Algorithm." Paris 11, 2009. http://www.theses.fr/2009PA112223.
Full textMy thesis deals with planification in a discrete environment with finite horizon and with a number of states too large to be explored entirely. The goal is to maximize a reward function that associates a value to final states. This thesis focuses on particular on improving and evaluating a new algorithm: bandit-based Monte Carlo tree search. After presenting the state of the art (Minimax and Alphabeta for the two-players case; nested Monte Carlo and Dynamic Programing for the one-player case), I describe the principle of the algorithm. Then, I propose an efficient parallelization method for the case of separated memories. This method can be combined with classical parallelization methods for shared memories. I propose also a way of constructing an opening book and show its efficiency in the concrete case of the game of Go. I introduce also several ways of using expert knowledge, in the part concerning bandits as well as in the Monte Carlo part. Finally, I show that this algorithm that gives very good results in the context of two-players applications is also efficient in a one-player context. I propose an adaptation of the algorithm in order to handle graphs and use a different bandit formula in order to solve the problem of the automatic generation of linear transforms libraries. I obtain results much better than by using a classical dynamic programming algorithm
Andral, Charly. "Quelques contributions aux méthodes de Monte Carlo en statistique." Electronic Thesis or Diss., Université Paris sciences et lettres, 2024. http://www.theses.fr/2024UPSLD049.
Full textMonte Carlo methods are extensively utilized in statistics for estimating quantities that cannot be analytically. In this thesis, we investigate various approaches to enhance the efficiency of Monte Carlo methods and establish connections between different techniques.In the first part, we connect Markov chain Monte Carlo methods with rejection sampling and important sampling to propose a new algorithm, the Importance Markov Chain. For this algorithm, we provide a theoretical analysis of its convergence properties by establishing a law of large numbers, a central limit theorem and a geometric ergodicity result. We also provide a numerical study to illustrate the performance of the Importance Markov Chain.In the second part, we combine normalizing flows with randomized quasi-Monte Carlo methods to improve the convergence rate of the estimator. We test our method on some examples and show that it can outperform standard Monte Carlo methods on low dimensional problems. However, the performance of the method decays when the dimension of the problem increases.In the third and final part, we cover another class of Monte Carlo methods, the piecewise deterministic Markov processes. We propose a new algorithm to sample from these processes and provide a theoretical proof of the correctness of the method. We provide several numerical experiments to illustrate the performance of the algorithm. It outperforms other PMDPs sampling algorithms in terms of computational cost and robustness when the potential is not convex
Karagulyan, Avetik. "Sampling with the Langevin Monte-Carlo." Electronic Thesis or Diss., Institut polytechnique de Paris, 2021. http://www.theses.fr/2021IPPAG002.
Full textSampling from probability distributions is a problem of significant importance in Statistics and Machine Learning. The approaches for the latter can be roughly classified into two main categories, that is the frequentist and the Bayesian. The first is the MLE or ERM which boils down to optimization, while the other requires the integration of the posterior distribution. Approximate sampling methods are hence applied to estimate the integral. In this manuscript, we focus mainly on Langevin sampling which is based on discretizations of Langevin SDEs. The first half of the introductory part presents the general mathematical framework of statistics and optimization, while the rest aims at the historical background and mathematical development of sampling algorithms.The first main contribution provides non-asymptotic bounds on convergence LMC in Wasserstein error. We first prove the bounds for LMC with the time-varying step. Then we establish bounds in the case when the gradient is available with a noise. In the end, we study the convergence of two versions of discretization, when the Hessian of the potential is regular.In the second main contribution, we study the sampling from log-concave (non-strongly) distributions using LMC, KLMC, and KLMC with higher-order discretization. We propose a constant square penalty for the potential function. We then prove non-asymptotic bounds in Wasserstein distances and provide the optimal choice of the penalization parameter. In the end, we highlight the importance of scaling the error for different error measures.The third main contribution focuses on the convergence properties of convex Langevin diffusions. We propose to penalize the drift with a linear term that vanishes over time. Explicit bounds on the convergence error in Wasserstein distance are proposed for the PenalizedLangevin Dynamics and Penalized Kinetic Langevin Dynamics. Also, similar bounds are proved for the Gradient Flow of convex functions
Helmstetter, Bernard. "Analyses de dépendances et méthodes Monte-Carlo dans les jeux de réflexion." Paris 8, 2007. http://octaviana.fr/document/126279233#?c=0&m=0&s=0&cv=0.
Full textWe explore two families of game programming methods: Monte-Carlo methods, and methods that exploit the weak dependencies between parts of a game. Those methods are applied to one and two-player games: a solitaire card game called Montana, the game of Go, and a one-player puzzle called Morpion solitaire. We describe an algorithm called incremental transpositions which we first apply to Montana; we also apply an algorithm called block search. We study the transitivity of connections in the game of Go and we develop the Monte-Carlo approach, which make a particularly simple program. On Morpion solitaire, applying the algorithm of incremental transpositions and combining with a parallelized search allows us to find a new record for a variant of the game
Janati, el idrissi Yazid. "Monte Carlo Methods for Machine Learning : Practical and Theoretical Contributions for Importance Sampling and Sequential Methods." Electronic Thesis or Diss., Institut polytechnique de Paris, 2023. http://www.theses.fr/2023IPPAS008.
Full textThis thesis contributes to the vast domain of Monte Carlo methods with novel algorithms that aim at adressing high dimensional inference and uncertainty quantification. In a first part, we develop two novel methods for Importance Sampling. The first algorithm is a lightweight optimization based proposal for computing normalizing constants and which extends into a novel MCMC algorithm. The second one is a new scheme for learning sharp importance proposals. In a second part, we focus on Sequential Monte Carlo methods. We develop new estimators for the asymptotic variance of the particle filter and provide the first estimator of the asymptotic variance of a particle smoother. Next, we derive a procedure for parameter learning within hidden Markov models using a particle smoother with provably reduced bias. Finally, we devise a Sequential Monte Carlo algorithm for solving Bayesian linear inverse problems with generative model priors
Coulibaly, Ibrahim. "Contributions à l'analyse numérique des méthodes quasi-Monte Carlo." Phd thesis, Université Joseph Fourier (Grenoble), 1997. http://tel.archives-ouvertes.fr/tel-00004933.
Full textMelenev, Petr. "Simulations Monte Carlo de propriétés structurales et magnétiques d'agrégats de nanoparticules." Paris 6, 2011. http://www.theses.fr/2011PA066162.
Full textNemirovsky, Danil. "Monte Carlo methods and Markov chain based approaches for PageRank computation." Nice, 2010. http://www.theses.fr/2010NICE4018.
Full textMaire, Sylvain. "Quelques Techniques de Couplage entre Méthodes Numériques Déterministes et Méthodes de Monte-Carlo." Habilitation à diriger des recherches, Université du Sud Toulon Var, 2007. http://tel.archives-ouvertes.fr/tel-00579977.
Full textMertz, Helene. "Modélisation des réactions chimiques dans un code de simulation par la méthode Monte Carlo." Thesis, Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLV012/document.
Full textDirect Simulation Monte Carlo (DSMC) methods are used in Ariane group to compute aerodynamic forces and moments and heat fluxes on space objects for hypersonic flows in rarefied regimes.To caracterise the dislocation of the stages and the debris footprints, a precise modelisation of the mechanism that contribute to the heat flux is necessary. The contribution of the chemical reactions is important for the determination of the heat flux. The purpose of this thesis is to develop the in house IEMC tool using the DSMC method so that it can compute reactive flows.The different steps of the developments are presented in this work. The first step is the presentation, implementation and verification of two different chemistry models. They are validated for simulations on real test cases. Different models are tested in order to evaluate their effect. Chemical models implemented in the code depend on new input parameters, whose numerical data are uncertain. Using a uncertainty quantification study, it is shown that the output data of the reactive simulation, especially the heat flux, is weakly impacted by the tested uncertain parameters
Coulot, Jérémy. "Dosimétrie des émetteurs bêta à l'échelle tissulaire et cellulaire par méthode de Monte-Carlo." Paris 11, 2004. http://www.theses.fr/2004PA11T061.
Full textInternal dosimetry is necessary both for radiation protection and for therapy. To calculate the dose, the physicist uses different dosimetric tools, including mathematical phantoms and algorithm, and a reference formalism, provided in part by the Medical Internal Radiation Dose (MIRD) comity. However, these tools are limited when focusing on beta emitters, particularly when they are heterogeneously distributed in tissues. Although many studies have been performed to assess the dosimetric consequences of such heterogeneity at the tissue and cellular level, there is a lack of tools as general as those developed at the organ level. In the present work, we established a global method to solve these problems, based on the Monte Carlo method. First, we adapted and validated at the required length scales a Monte Carlo particle transport code written at the Institut Gustave-Roussy (DOSE3D). Then, we developed a software dedicated to the mathematical description of biological structures, CLUSTER3D. These tools allowed us to define a mathematical model of the follicular structure of the thyroid gland, and to investigate the dosimetric consequences of a heterogeneous distribution of radioiodine isotopes in the thyroid. Finally, we present the study of the biological distribution of a radiolabeled monoclonal antibody (⁹⁰Y-Zevalin™), using digital autoradiography. Results represent the first mandatory step of a dosimetric study which will allows, using the developed tools, a better understanding of radioimmunotherapy treatments consequences
El, Idrissi Hassan. "Etude de la capacité SIS GaAs/AlGaAs/GaAs par la méthode de Monte-Carlo." Lille 1, 1993. http://www.theses.fr/1993LIL10016.
Full textMorel, Sylvie. "La régression logistique : comparaison avec l'analyse probit à l'aide de la méthode Monte Carlo." Master's thesis, Université Laval, 1986. http://hdl.handle.net/20.500.11794/29182.
Full textTessé, Lionel. "Modélisation des transferts radiatifs dans les flammes turbulentes par une méthode de Monte Carlo." Châtenay-Malabry, Ecole centrale de Paris, 2001. http://www.theses.fr/2001ECAP0739.
Full textChen, Yuting. "Inférence bayésienne dans les modèles de croissance de plantes pour la prévision et la caractérisation des incertitudes." Thesis, Châtenay-Malabry, Ecole centrale de Paris, 2014. http://www.theses.fr/2014ECAP0040/document.
Full textPlant growth models aim to describe plant development and functional processes in interaction with the environment. They offer promising perspectives for many applications, such as yield prediction for decision support or virtual experimentation inthe context of breeding. This PhD focuses on the solutions to enhance plant growth model predictive capacity with an emphasis on advanced statistical methods. Our contributions can be summarized in four parts. Firstly, from a model design perspective, the Log-Normal Allocation and Senescence (LNAS) crop model is proposed. It describes only the essential ecophysiological processes for biomass budget in a probabilistic framework, so as to avoid identification problems and to accentuate uncertainty assessment in model prediction. Secondly, a thorough research is conducted regarding model parameterization. In a Bayesian framework, both Sequential Monte Carlo (SMC) methods and Markov chain Monte Carlo (MCMC) based methods are investigated to address the parameterization issues in the context of plant growth models, which are frequently characterized by nonlinear dynamics, scarce data and a large number of parameters. Particularly, whenthe prior distribution is non-informative, with the objective to put more emphasis on the observation data while preserving the robustness of Bayesian methods, an iterative version of the SMC and MCMC methods is introduced. It can be regarded as a stochastic variant of an EM type algorithm. Thirdly, a three-step data assimilation approach is proposed to address model prediction issues. The most influential parameters are first identified by global sensitivity analysis and chosen by model selection. Subsequently, the model calibration is performed with special attention paid to the uncertainty assessment. The posterior distribution obtained from this estimation step is consequently considered as prior information for the prediction step, in which a SMC-based on-line estimation method such as Convolution Particle Filtering (CPF) is employed to perform data assimilation. Both state and parameter estimates are updated with the purpose of improving theprediction accuracy and reducing the associated uncertainty. Finally, from an application point of view, the proposed methodology is implemented and evaluated with two crop models, the LNAS model for sugar beet and the STICS model for winter wheat. Some indications are also given on the experimental design to optimize the quality of predictions. The applications to real case scenarios show encouraging predictive performances and open the way to potential tools for yield prediction in agriculture
Pistrui-Maximean, Simona Anca. "Modeling and simulation of the response of scintillation screens for X-ray imaging." Lyon, INSA, 2007. http://theses.insa-lyon.fr/publication/2006ISAL0094/these.pdf.
Full textCette thèse est consacrée à l'étude de la réponse des écrans scintillateurs utilisés en imagerie par rayons X, à l'aide d'une approche de modélisation/simulation basée sur le code de Monte Carlo Geant4. Cette approche permet de prendre en compte la diversité des processus physiques mis en jeu (interactions des photons X, électrons et photons visibles). Après une synthèse sur l'état de l'art concernant les propriétés des scintillateurs, ainsi que leur modélisation, nous montrons tout d'abord que les résultats fournis par notre outil de simulation sont en adéquation avec le modèle analytique de Swank qui explicite la fonction de transfert de modulation (FTM) dans des cas simples. La simulation Monte Carlo permet d'isoler le rôle joué par chacun des processus physiques mis en jeu : dégradation mineure de la réponse spatiale par les processus électroniques et amélioration sensible causée par l'absorption et la diffusion des photons visibles. Des validations expérimentales sont effectuées sur des écrans scintillateurs commerciaux à l'aide d'un faisceau X collimaté par une fente en uranium. Après une étape d'estimation des coefficients optiques par la technique du gradient stochastique, nous montrons que les réponses spatiales simulée et mesurée s'accordent pour différentes énergies du rayonnement. Dans la dernière partie de cette recherche, nous étudions le rôle (en termes d'efficacité et de résolution spatiale) de couches métalliques ou plastiques servant de support au scintillateur. À sensibilité égale, l'effet renforçateur du substrat permet de réduire l'épaisseur du scintillateur sans dégradation notable de la résolution spatiale
Vaïana, Florian. "Couplage neutronique - thermohydraulique : application au réacteur à neutrons rapides refroidi à l'hélium." Grenoble INPG, 2009. http://www.theses.fr/2009INPG0132.
Full textThis thesis focuses en the study of the interactions between neutron-kinetics and thermal-hydraulics. Neutron-kinetics allow to calculate the power in a nuclear reactor and the temperature evolution of materials where this power is deposited is known thanks to thermal-hydraulics. The first part of this work corresponds to the study and development of a method which allows to simulate transients in nuclear reactor and especially with a Monte-Carlo code for neutron-kinetics. The second part deals with the study of a misplaced control rod withdrawing in a GFR, a fourth generation reactor. Some models allowing to calculate neutron-kintics and thermal-hydraulics in the core (which contains assemblies built up with fuel plates) were defined
Saucier, Marie Annie. "Développement d'un outil de simulation par Monte Carlo du rayonnement diffusé en tomodensitométrie." Master's thesis, Université Laval, 2018. http://hdl.handle.net/20.500.11794/31391.
Full textThe goal of this project is to develop an application to correct the scattered radiation in a cone beam computed tomography scan (CBCT). A Monte Carlo simulation is used to estimate the scattered radiation which is a numerical replication of a CBCT acquisition. This project has been divided into two sections : the validation of the physics for this specific application and the development of the application. The validation consisted in reproducing the results obtained with Geant4 in GPUMCD. Geant4 is the reference platform and GPUMCD is the platform studied. Both are Monte Carlo simulators of the passage of particles through matter.The elements studied are the cross sections, the materials, the Rayleigh scattering algorithm and the Compton scattering algorithm. Although some errors are still present, a great improvement of the results between GPUMCD and Geant4 was obtained. The difference between the two simulations was greater than 100 % for complex geometries and dropped below 10% after corrections of the physics. In addition, it was possible to identify some other problems such as a theoretical difference in the Compton scattering algorithms. Regarding the second part of the project, although the correction could not be implemented in a reconstruction, all elements are present to estimate the scattered radiation for an actual CBCT reconstruction. The parameters and strategies studied in order to optimize the computation time while maintaining the accuracy of the results are : ray tracing, Gaussian smoothing of scattered radiation, reduction of the number of pixels on the detector, interpolation of between the simulated projections, symmetry and reduction of number of voxels in the patient. In addition, considering a correction of high quality is 2 % error and less per implemented strategy, a simulation time of less than 2 minutes is obtained. For a low quality simulation (5% error and less per parameter), a simulation time of less than 15 seconds per simulation was obtained. Those are clinically acceptable simulation times.
Chan, Shio Christian Paul. "Échantillonner les solutions de systèmes différentiels." Thesis, Nice, 2014. http://www.theses.fr/2014NICE4114/document.
Full textThis work addresses two complementary problems when studying differential systems with random coefficients using a simulation approach. In the first part, we look at the problem of computing the law of the solution at time t* of a differential equation with random coefficients. It is shown that even in simplest cases, one will usually obtain a random variable where the pdf cannot be computed explicitly, and for which we need to rely on Monte Carlo simulation. As this simulation may not always be possible due to the explosion of the solution, several workarounds are presented. This includes displaying the histogram on a compact manifold using two charts and approximating the distribution using a polynomial chaos expansion. The second part considers the problem of estimating the coefficients in a system of differential equations when a trajectory of the system is known at a set of times. To do this, we use a simple Monte Carlo sampling method, known as the rejection sampling algorithm. Unlike deterministic methods, it does not provide a point estimate of the coefficients directly, but rather a collection of values that “fits” the known data well. An examination of the properties of the method allows us not only to better understand how to choose the different parameters when implementing the method, but also to introduce more efficient methods. This includes a new approach which we call sequential rejection sampling and methods based on the Markov Chain Monte Carlo and Sequential Monte Carlo algorithms. Several examples are presented to illustrate the performance of all these methods
Gonçalves, Thomas. "Contributions à la parallélisation de méthodes de type transport Monte-Carlo." Thesis, Université Grenoble Alpes (ComUE), 2017. http://www.theses.fr/2017GREAM047/document.
Full textMonte Carlo particle transport applications consist in studying the behaviour of particles moving about a simulation domain. Particles distribution among simulation domains is not uniform and change dynamically during simulation. The parallelization of this kind of applications on massively parallel architectures leads to solve a complex issue of workloads and data balancing among numerous compute cores.We started by identifying parallelization pitfalls of Monte Carlo particle transport applications using theoretical and experimental analysis of reference parallelization methods. A semi-dynamic based on partitioning techniques has been proposed then. Finally, we defined a dynamic approach able to redistribute workloads and data keeping a low communication volume. The dynamic approach obtains speedups using strong scaling and a memory footprint reduction compared to the perfectly balanced domain replication method
Bouabça, Thomas. "Introduction d'orbitales corrélées dans les approches Monte-Carlo quantiques." Toulouse 3, 2009. http://thesesups.ups-tlse.fr/847/.
Full textQuantum chemistry is the branch of theoretical chemistry which applies quantum mechanics to chemistry. The computation of chemical properties is a huge challenge for many scientific and technological fields. (biochemistry, nanosciences. . . ). Nevertheless, for now, no methods can accurately study any systems according to their size or their nature. Based on a stochastic resolution of the Schrödinger equation, Quantum Monte Carlo methods (QMC) represent an original and efficient way for this matter. They are especially suited for big molecular systems. For instance, QMC methods are known to be the most powerful algorithms for computing total ground-state energies. However, some quantities can still not be properly computed with QMC methods. Thus, one of the main issue that remains is the evaluation of differences of energies. Solving this problem is an important step for QMC methods to be considered as standard ones. Indeed, roughly speaking, differences of energies are at the heart of the whole chemistry : any chemical problem can be interpreted as a difference of energies. The purpose of this thesis is to propose a way to compute those differences with QMC methods. Our approach is particularly motivated by a deep concern : using simple preoptimized wavefunctions. In order to achieve this, we propose here two strategies : First, a new wavefunction is introduced. This wavefunction is composed of preoptimized modular elements. With this new wavefunction, any system can be recomposed piece by piece. Second, a set of coherent wavefunctions is used for a controlled compensation of errors
Saggadi, Samira. "Simulation d'évènements rares par Monte Carlo dans les réseaux hautement fiables." Thesis, Rennes 1, 2013. http://www.theses.fr/2013REN1S055.
Full textNetwork reliability determination, is an NP-hard problem. For instance, in telecommunications, it is desired to evaluate the probability that a selected group of nodes communicate or not. In this case, a set of disconnected nodes can lead to critical financials security consequences. A precise estimation of the reliability is, therefore, needed. In this work, we are interested in the study and the calculation of the reliability of highly reliable networks. In this case the unreliability is very small, which makes the standard Monte Carlo approach useless, because it requires a large number of iterations. For a good estimation of system reliability with minimum cost, we have developed new simulation techniques based on variance reduction using importance sampling
Roucairol, Milo. "Monte-Carlo tree search applied to structure generation." Electronic Thesis or Diss., Université Paris sciences et lettres, 2024. http://www.theses.fr/2024UPSLD029.
Full textThis document gathers the articles published during my PhD thesis directed by Tristan Cazenave at LAMSADE. Monte Carlo search refers to a class of stochastic search algorithms that return a solution with a guarantee of time, but no guarantee of result. These algorithms use reinforcement learning techniques based on random or guided exploration. The capabilities of Monte Carlo algorithms are limited in recently highlighted application domains, such as image and text generation, where neural networks, LLM and other algorithms trained on large databases dominate. On the other hand,they excel in more classic, defined problems.The best-known use of a Monte Carlo search algorithm is its use in 2017 to beat a Go champion for the first time, somethingno other algorithm family had managed to do. But the uses of Monte Carlo search algorithms also extend far beyond gaming. Monte Carlo search algorithms are widely used in chemistry, operations research, transportation, mathematics,and gaming. They can be applied to any sequential decision and state-space search problem, as long as the functions for evaluating and modifying a state are defined. The definition of structure for this thesis is “a system defined by the elements that compose it and the interactions betweenthese elements”. This thesis explores several applications of Monte Carlo search in the context of structure generation.Many search spaces can be represented as structures outside of games, such as the circuit of the traveling sales man problem, but also molecules, crystals, coalitions, graphs, etc. The highlights of this thesis are: - Comparisons between algorithms on various problems showing the superiority of the“nested” family of algorithms. - A new variant of Nested Monte Carlo Tree Search (NMCS) with improved performance.- A library of Monte Carlo algorithms coded in Rust. - A project to refute graph conjectures. - An NMCS implementation for AiZynthFinder, AstraZeneca's open source retrosynthesis software. - A program for generating valid, synthesizable molecules. The topics covered can be divided into two groups. On the one hand, chemistry, with HP-model, retrosynthesis andmolecule generation. On the other, mathematics, with coalition structures, spectral graph theory, transport networks and nonograms. Although this thesis is devoted solely to applications of Monte Carlo search, it also provides more general insights: a comparison of algorithm families showing the superiority of “nested”, a new variant of NMCS, and heuristics and modifications generally useful with NP problems
Chamberland, Ève. "Évaluation d'un algorithme de calcul de dose par méthode Monte Carlo pour des faisceaux d'électrons." Thesis, Université Laval, 2010. http://www.theses.ulaval.ca/2010/27413/27413.pdf.
Full textSénégas, Julien. "Méthode de Monte Carlo en vision stéréoscopique : Application à l'étude de modèles numériques de terrain." Paris, ENMP, 2002. https://pastel.archives-ouvertes.fr/tel-00005637.
Full textBenassi, Romain. "Nouvel algorithme d'optimisation bayésien utilisant une approche Monte-Carlo séquentielle." Phd thesis, Supélec, 2013. http://tel.archives-ouvertes.fr/tel-00864700.
Full textTarhini, Ali. "Analyse numérique des méthodes quasi-Monte Carlo appliquées aux modèles d'agglomération." Chambéry, 2008. http://www.theses.fr/2008CHAMS015.
Full textMonte Carlo (MC) methods are probabilistic methods based on the use of random numbers in repeated experiments. Quasi-Monte Carlo (QMC) methods are deterministic versions of Monte Carlo methods. Random sequences are replaced by low discrepancy sequences. These sequences ha ve a better uniform repartition in the s-dimensional unit cube. We use a special class of low discrepany sequences called (t,s)-sequences. In this work, we develop and analyze Monte Carlo and quasi-Monte Carlo particle methods for agglomeration phenomena. We are interested, in particular, in the numerical simulation of the discrete coagulation equations (the Smoluchowski equation), the continuous coagulation equation, the continuous coagulation-fragmentation equation and the general dynamics equation (GDE) for aerosols. In all these particle methods, we write the equation verified by the mass distribution density and we approach this density by a sum of n Dirac measures ; these measures are weighted when simulating the GDE equation. We use an explicit Euler disretiza tion scheme in time. For the simulation of coagulation and coagulation-fragmentation, the numerical particles evolves by using random numbers (for MC simulations) or by quasi-Monte Carlo quadratures. To insure the convergence of the numerical scheme, we reorder the numerical particles by their increasing mass at each time step. In the case of the GDE equation, we use a fractional step iteration scheme : coagulation is simulated as previously, other phenomena (like condensation, evaporation and deposition) are integrated by using a deterministic particle method for solving hyperbolic partial differential equation. We prove the convergence of the QMC numerical scheme in the case of the coagulation equation and the coagulation-fragmentation equation, when the number n of numerical particles goes to infinity. All our numerical tests show that the numerical solutions calculated by QMC algorithms converges to the exact solutions and gives better results than those obtained by the corresponding Monte Carlo strategies
Galdin-Retailleau, Sylvie. "Etude du transistor bipolaire npn a double heterojonction si/sige/si par simulations monte-carlo." Paris 11, 1992. http://www.theses.fr/1992PA112132.
Full textThauvoye, Christophe. "Simulation numérique d'écoulements turbulents réactifs par une méthode hybride à fonction densité de probabilité transportée." Poitiers, 2005. http://www.theses.fr/2005POIT2276.
Full textThis work concerns the field of numerical simulation of turbulent reactive flows. The aim of this work is to study a hybrid method based on the use of a lagrangian transported Probability Density Function (PDF) method coupled with a eulerian method which solves the Reynolds Averaged Navier-Stokes equations (R. A. N. S). The first part is devoted to the description of the RANS and the transported PDF methods. The latter is more precisely detailed : it allows to highlight both advantages and drawbacks of the two approaches. In this context, we will develop all the aspects related to the modelling and resolution of the transported joint PDF equation. Its resolution generally uses a Monte-Carlo numerical simulation. We also show how the statistical nature of Monte-Carlo methods induces numerical difficulties, which led to the development of hybrid methods associating RANS method with a transported PDF approach. In the second part of this study, theoretical and numerical aspects of the hybrid methods are detailed, and more precisely the PEUL+ model developed at ONERA. A new – instationary – way of coupling is proposed. It improves the stability and precision of the model in comparison with the stationary way of coupling. It is then tested and validated on two configurations : a methane-air nonpremixed flame stabilised by a piloted flame ; and a premixed flame in a sudden symmetric plane expansion
Buchholz, Alexander. "High dimensional Bayesian computation." Electronic Thesis or Diss., Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLG004.
Full textComputational Bayesian statistics builds approximations to the posterior distribution either bysampling or by constructing tractable approximations. The contribution of this thesis to the fieldof Bayesian statistics is the development of new methodology by combining existing methods. Ourapproaches either scale better with the dimension or result in reduced computational cost com-pared to existing methods. Our first contribution improves approximate Bayesian computation(ABC) by using quasi-Monte Carlo (QMC). ABC allows Bayesian inference in models with in-tractable likelihoods. QMC is a variance reduction technique that yields precise estimations ofintegrals. Our second contribution takes advantage of QMC for Variational Inference (VI). VIis a method for constructing tractable approximations to the posterior distribution. The thirdcontribution develops an approach for tuning Sequential Monte Carlo (SMC) samplers whenusing Hamiltonian Monte Carlo (HMC) mutation kernels. SMC samplers allow the unbiasedestimation of the model evidence but tend to struggle with increasing dimension. HMC is aMarkov chain Monte Carlo technique that has appealing properties when the dimension of thetarget space increases but is difficult to tune. By combining the two we construct a sampler thattakes advantage of the two
Samba, Gérald. "Analyse asymptotique de schémas de résolution de l'équation du transport en régime diffusif." Phd thesis, École normale supérieure de Cachan - ENS Cachan, 2008. http://tel.archives-ouvertes.fr/tel-00363723.
Full textBuchholz, Alexander. "High dimensional Bayesian computation." Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLG004/document.
Full textComputational Bayesian statistics builds approximations to the posterior distribution either bysampling or by constructing tractable approximations. The contribution of this thesis to the fieldof Bayesian statistics is the development of new methodology by combining existing methods. Ourapproaches either scale better with the dimension or result in reduced computational cost com-pared to existing methods. Our first contribution improves approximate Bayesian computation(ABC) by using quasi-Monte Carlo (QMC). ABC allows Bayesian inference in models with in-tractable likelihoods. QMC is a variance reduction technique that yields precise estimations ofintegrals. Our second contribution takes advantage of QMC for Variational Inference (VI). VIis a method for constructing tractable approximations to the posterior distribution. The thirdcontribution develops an approach for tuning Sequential Monte Carlo (SMC) samplers whenusing Hamiltonian Monte Carlo (HMC) mutation kernels. SMC samplers allow the unbiasedestimation of the model evidence but tend to struggle with increasing dimension. HMC is aMarkov chain Monte Carlo technique that has appealing properties when the dimension of thetarget space increases but is difficult to tune. By combining the two we construct a sampler thattakes advantage of the two
Jaillon, Franck. "Caractérisation optique des milieux diffusants : simulation Monte Carlo et mesures en rétrodiffusion polarisée." Lyon 1, 2003. http://www.theses.fr/2003LYO10055.
Full textDoan, Viet Dung. "Grid computing for Monte Carlo based intensive calculations in financial derivative pricing applications." Nice, 2010. http://www.theses.fr/2010NICE4078.
Full textIn this thesis, we provide a grid programming framework, named PicsouGrid, which specially targets financial derivative pricing applications. We are interested in the evaluation of high dimensional option contracts (based on a basket of underlying assets) Such a framework includes fault tolerance, load balancing, dynamic task distribution and deployment mechanisms, and as such targets heterogeneous computing environments like computing grids. We also propose a parallelisation of the Classification Monte Carlo (CMC) algorithm, originally devised by Picazo (2002), for high dimensional American option pricing, which scales in a grid environment. We describe experimental results obtained running this parallel algorithm using PicsouGrid. We are able to run this algorithm for pricing American options requiring important computation times, because of a large number of opportunity dates (e. G. Up to 100 such dates), and because of large dimension (e. G. Up to 40 underlying assets). We also compare and analyse the option price accuracy obtained with this CMC algorithm when using different classification algorithms such as AdaBoost, Gradient Boost and Support Vector Machines. Additionally, we define a financial benchmark suite for performance evaluation and analysis of grid computing middlewares. Such benchmark suite is conceptually simple and easy to understand for both the grid computing and financial communities. The benchmark suite was successfully used in the 2008 SuperQuant Monte Carlo challenge - the Fifth Grid Plugtest and Contest. Within the context of this challenge, we developed the ProActive Monte Carlo API (MC API). Key works : Grid computing, high dimensional American option pricing, Monte Carlo methods, classification techniques, benchmark suite, grid iddleware
Haeseler, Geoffroy. "Transitions de phases topologiques classiques et quantiques dans les systèmes magnétiques frustrés." Electronic Thesis or Diss., Lyon, École normale supérieure, 2025. http://www.theses.fr/2025ENSL0008.
Full textIl est bien connu que les glaces de spins comme Ho2Ti2O7 présentent plusieurs plateaux dans leur aimantation lorsqu’un champ magnétique leur est appliqué. Le premier d’entre eux est du aux gèles des spins apicaux qui laisse une entropie extensive dans les plans de kagomé perpendiculaire au champ et isolés les uns des autres. Dans ce cas, les spins du réseau kagomé sont contraint dans leur phase K11. En utilisant le processus de fragmentation, les spins des plans mappent, en partie, sur un modèle de dimer sur réseau nid d’abeille. Ce système est étudié sous plusieurs conditions. Nous commençons par ajouter un champ magnétique dans les plans kagomé. Dans le langage de dimer, le champ prend la forme d’affinités chimiques différentes en fonction de la direction du dimer. Dans ce cas, le système présente une transition de Kasteleyn avec des corrélations critiques ajustables par la direction du champ. Ensuite nous ajoutons un terme de plaquette, qui compte le nombre de boucle de trois dimers pouvant être ajouté, conduisant à une transition BKT vers une star phase ordonnée lorsque ce terme est négatif et une transition de Kasteleyn vers une columnar phase lorsque qu’il est positif. Nous ajoutons finalement ces deux termes en même temps pour trouver un point tri critique à partir duquel la transition BKT devient premier ordre. Nous étudions ensuite une généralisation de ce modèle de dimer sur réseau diamant sous un champs qui peut être incliné, conduisant à d’autres transitions de Kasteleyn en 3D. Ce modèle peut être réalisé expérimentalement dans des pyrochlores qui satisfont, dans leur état fondamental, la règle 3-in-1-out and 3-out-1-in similaire à celle des glaces de spins comme Ho2Ir2O7. Ce modèle de dimer présente des points triples intéressant autour desquels la température tends vers zéros et l’état de plus basse énergie présente une entropie résiduelle proportionnel à celle de la glace de kagomé. Finalement nous ajoutons des termes quantiques, hors-diagonaux, sur le modèle de dimer 2D, ce qui favorise des superpositions quantiques des boucles à trois dimer autour d’un hexagone similaire à des anneaux de benzène. Ce modèle est étudié numériquement à l’aide d’un algorithme de Monte-Carlo basé sur le formalisme de Suzuki-Trotter. Nous confirmons l’existence d’une transition de premier ordre entre la star phase classique et la phase plaquette résonnante quantique. Nous ajoutons un champ conjugué au paramètre d’ordre de la star phase pour trouver un point critique quantique entre ces deux phases. Une analyse fine de la phase plaquette révèle des corrélations similaires à celle d’un liquide de Coulomb qui disparaissent en loi de puissance avec la température, ce qui indique l’existence d’un spectre d’excitations continues au-dessus d’un état fondamental ordonné. Enfin, nous étudions cette phase quantique avec un secteur topologique non zéro qui frustre l’ordre à longue portée de la star phase, laissant des corrélations rappelant un liquide quantique. Les corrélations sont présentées sous la forme de figures de diffraction de neutrons polarisés interagissant avec les spins du réseau kagomé équivalent au modèle de dimer et fragmenté pour en extraire le champ transverse émergent des dimers