Dissertations / Theses on the topic 'Méthodes itératives'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Méthodes itératives.'
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.
Paleo, Pierre. "Méthodes itératives pour la reconstruction tomographique régularisée." Thesis, Université Grenoble Alpes (ComUE), 2017. http://www.theses.fr/2017GREAT070/document.
Full textIn the last years, there have been a diversification of the tomography imaging technique for many applications. However, experimental constraints often lead to limited data - for example fast scans, or medical imaging where the radiation dose is a primary concern. The data limitation may come as a low signal to noise ratio, scarce views or a missing angle wedge.On the other hand, artefacts are detrimental to reconstruction quality.In these contexts, the standard techniques show their limitations.In this work, we explore how regularized tomographic reconstruction methods can handle these challenges.These methods treat the problem as an inverse problem, and the solution is generally found by the means of an optimization procedure.Implementing regularized reconstruction methods entails to both designing an appropriate regularization, and choosing the best optimization algorithm for the resulting problem.On the modelling part, we focus on three types of regularizers in an unified mathematical framework, along with their efficient implementation: Total Variation, Wavelets and dictionary-based reconstruction. On the algorithmic part, we study which state-of-the-art convex optimization algorithms are best fitted for the problem and parallel architectures (GPU), and propose a new algorithm for an increased convergence speed.We then show how the standard regularization models can be extended to take the usual artefacts into account, namely rings and local tomography artefacts. Notably, a novel quasi-exact local tomography reconstruction method is proposed
Sadek, El Mostafa. "Méthodes itératives pour la résolution d'équations matricielles." Thesis, Littoral, 2015. http://www.theses.fr/2015DUNK0434/document.
Full textIn this thesis, we focus in the studying of some iterative methods for solving large matrix equations such as Lyapunov, Sylvester, Riccati and nonsymmetric algebraic Riccati equation. We look for the most efficient and faster iterative methods for solving large matrix equations. We propose iterative methods such as projection on block Krylov subspaces Km(A, V ) = Range{V,AV, . . . ,Am−1V }, or block extended Krylov subspaces Kem(A, V ) = Range{V,A−1V,AV,A−2V,A2V, · · · ,Am−1V,A−m+1V }. These methods are generally most efficient and faster for large problems. We first treat the numerical solution of the following linear matrix equations : Lyapunov, Sylvester and Stein matrix equations. We have proposed a new iterative method based on Minimal Residual MR and projection on block extended Krylov subspaces Kem(A, V ). The extended block Arnoldi algorithm gives a projected minimization problem of small size. The reduced size of the minimization problem is solved by direct or iterative methods. We also introduced the Minimal Residual method based on the global approach instead of the block approach. We projected on the global extended Krylov subspace Kem(A, V ) = Span{V,A−1V,AV,A−2V,A2V, · · · ,Am−1V,A−m+1V }. Secondly, we focus on nonlinear matrix equations, especially the matrix Riccati equation in the continuous case and the nonsymmetric case applied in transportation problems. We used the Newton method and MINRES algorithm to solve the projected minimization problem. Finally, we proposed two new iterative methods for solving large nonsymmetric Riccati equation : the first based on the algorithm of extended block Arnoldi and Galerkin condition, the second type is Newton-Krylov, based on Newton’s method and the resolution of the large matrix Sylvester equation by using block Krylov method. For all these methods, approximations are given in low rank form, wich allow us to save memory space. We have given numerical examples that show the effectiveness of the methods proposed in the case of large sizes
Marinesque, Sébastien. "Méthodes de reconstruction itératives en tomographie thermoacoustique." Toulouse 3, 2012. http://thesesups.ups-tlse.fr/1888/.
Full textWe define, study and implement various iterative reconstruction methods for Thermoacoustic Tomography (TAT): the Back and Forth Nudging (BFN), easy to implement and to use, a variationnal technique (VT) and the Back and Forth SEEK (BF-SEEK), more sophisticated, and a coupling method between Kalman filter (KF) and Time Reversal (TR). A unified formulation is explained for the sequential techniques aforementioned that defines a new class of inverse problem methods: the Back and Forth Filters (BFF). In addition to existence and uniqueness (particularly for backward solutions), we study many frameworks that ensure and characterize the convergence of the algorithms. We give a general theoretical framework for which the BFN is a well-posed problem. Then, in application to TAT, existence and uniqueness of its solutions and geometrical convergence of the algorithm are proved, and an explicit convergence rate and a description of its numerical behaviour are given. Theoretical and numerical studies of more general and realistic framework are led, namely different objects, speeds (with or without trapping), various sensor configurations and samplings, attenuated equations or external sources. Optimal control and best estimate tools are used to characterize the BFN convergence and converging feedbacks for BFF, under observability assumptions. We compare the most flexible and efficient current techniques (TR and an iterative variant) with our various BFF and the VT in several experiments. Thus, robust, with different possible complexities and flexible, the methods that we propose are very interesting reconstruction techniques, particularly in TAT and when observations are degraded
Zhang, Hanyu. "Méthodes itératives à retard pour architecture massivement parallèles." Thesis, Université Paris-Saclay (ComUE), 2016. http://www.theses.fr/2016SACLC068.
Full textWith the increase of architectures composed of multi-cores, many algorithms need to revisited and be modified to exploit the power of these new architectures. These algorithms divide the original problem into “small pieces” and distribute these pieces to different processors at disposal, thus communications among them are indispensible to assure the convergence. My thesis mainly focus on solving large sparse systems of linear equations in parallel with new methods. These methods are based on the gradient methods. Two key parameters of the gradient methods are descent direction and step-length of descent for each iteration. Our methods compute the directions locally, which requires less synchronization and computation, leading to faster iterations and make easy asynchronization possible. Convergence can be proved in both synchronized or asynchronized cases. Numerical tests demonstrate the efficiency of these methods. The other part of my thesis deal with the acceleration of the vector sequences generated by classical iterative algorithms. Though general chaotic sequences may not be accelerated, it is possible to prove that with any fixed retard pattern, then the generated sequence can be accelerated. Different numerical tests demonstrate its efficiency
Tairi, Souhil. "Développement de méthodes itératives pour la reconstruction en tomographie spectrale." Thesis, Aix-Marseille, 2019. http://www.theses.fr/2019AIXM0160/document.
Full textIn recent years, hybrid pixel detectors have paved the way for the development of spectral X ray tomography or spectral tomography (CT). Spectral CT provides more information about the internal structure of the object compared to conventional absorption CT. One of its objectives in medical imaging is to obtain images of components of interest in an object, such as biological markers called contrast agents (iodine, barium, etc.).The state of the art of simultaneous reconstruction and separation of spectral CT data methods remains to this day limited. Existing reconstruction approaches are limited in their performance and often do not take into account the complexity of the acquisition model.The main objective of this thesis work is to propose better quality reconstruction approaches that take into account the complexity of the model in order to improve the quality of the reconstructed images. Our contribution considers the non-linear polychromatic model of the X-ray beam and combines it with an earlier model on the components of the object to be reconstructed. The problem thus obtained is an inverse, non-convex and misplaced problem of very large dimensions.To solve it, we propose a proximal algorithmwith variable metrics. Promising results are shown on real data. They show that the proposed approach allows good separation and reconstruction despite the presence of noise (Gaussian or Poisson). Compared to existing approaches, the proposed approach has advantages over the speed of convergence
Qaddouri, Abdessamad. "Méthodes itératives parallèles, applications en neutronique et en mécanique des fluides." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1998. http://www.collectionscanada.ca/obj/s4/f2/dsk2/tape17/PQDD_0008/NQ33074.pdf.
Full textBerthet, Antoine Olivier. "Méthodes itératives appliquées au décodage efficace de combinaisons de codes en treillis." Paris 6, 2001. http://www.theses.fr/2001PA066498.
Full textDjellab, Housni. "Optimisation combinatoire dans les systèmes de production : hypergraphes et méthodes d'amélioration itératives." Clermont-Ferrand 2, 1997. http://www.theses.fr/1997CLF21952.
Full textBerthet, Antoine Olivier. "Méthodes itératives appliquées au décodage efficace de combinaisons de codes sur treillis." Paris, ENST, 2001. http://www.theses.fr/2001ENST0036.
Full textFar from concentrating on the theory error-correcting codes (e. G. , turbo codes), the renewed interest for iterative methods has spread to the entire communications theory and lead to the advent of a real turbo principle. In the classical theory, the different elements which make up the receiver (detector, equalizer, demodulator, channel decoder, source decoder) are activated sequentially only once in a given order. The propagation of soft decisions between those elements (in opposition to hard decisions) preserves all the information available at the channel output about the variables to estimate. But still remains the fundamental sub-optimality induced by the partitioning of the receiver chain into distinct specific functions, each of them acting with a partial knowledge on the others (especially the first ones on the last ones). The so-called turbo principle aims at recovering the optimality. It substitutes to the classical approach an iterative approach where the different functions of the receiver chain, formally identified to serially concatenated decoders and activated several times according to a given schedule, accept, deliver, and exchange constantly refined probabilistic information (referred to as extrinsic information) about the variables to estimate. The turbo detection is a first instance of the turbo principle. The basic idea consists in modelling the intersymbol interference channel (IIC) as a rate-1 time-varying convolutional code defined by a generator polynomial with complex coefficients. The serial concatenation of the error-correcting code and the IIC suggests the application of an iterative procedure between the two corresponding decoders, which, in effect, allows removing the intersymbol interference completely. Exploiting the highly structured nature of interfering signals, the turbo principle provides excellent results in multiuser detection as well. Other recent and promising applications are the demodulation of nonlinear continuous phase modulations or the decoding of joint source-channel codes. This PhD thesis is mainly focused on the identification and analysis of new instances of the turbo principle. The first part of the thesis is devoted to the design and iterative decoding of highly spectrally-efficient multilevel codes for the Gaussian channel. The proposed schemes involve a multitude of small linear component codes, convolutional or block, and concatenated or not. The optimal symbol-by-symbol decoding of linear block codes, for which finding a representative trellis as reduced as possible in complexity constitutes a fundamental issue, is thoroughly investigated (chapter 2, in French). The parametrization (length, rates at each level, etc. ) and the performance of the multilevel codes are optimized under iterative multistage decoding (chapter 3, in English). The second part of the thesis deals with near-optimal decoding of serially concatenated modulations, bit-interleaved or not, when transmission occurs over frequency-selective channels. We investigate different reduced-complexity approaches to perform detection/equalization, channel decoding and channel estimation in a completely or partially disjoint and iterative fashion (chapter 4, in English). These approaches are then extended to serially concatenated space-time trellis-coded modulations and frequency-selective multiple-input multiple-output (MIMO) channels (chapter 5, in English)
Shahzadeh, Fazeli Seyed Abolfazi. "Stratégies de redémarrage des méthodes itératives d'algèbre linéaire pour le calcul global." Versailles-St Quentin en Yvelines, 2005. http://www.theses.fr/2005VERS0011.
Full textL'objectif de ce travail est de contribuer à la résolution des grands problèmes de valeur propre et/ou des grands systèmes linéaires en utilisant des ressources partagées sur des réseaux plus ou moins larges. La résolution de grands systèmes d'algèbre linéaire s'effectue, à l'aide des méthodes itératives hybrides. Une méthode hybride combine plusieurs méthodes numériques différentes ou bien plusieurs copy d'une même méthode numérique paramétrées différemment afin d'accélérer la convergence de l'une de ces méthodes. L'amélioration de la vitesse de convergence et d'exécution des méthodes hybrides par des méthodologies numériques et/ou des techniques de calcul parallèle et distribué constitue l'objectif principal de cette thèse. La vitesse de convergence de ces méthodes est dépendante de l'approche utilisée lors du redémarrage du processus itératif. Nous présentons une étude sur une méthode hybride appelée Multiple Explicitly Restarted Arnoldi Method (MERAM), et nous proposons deux approches synchrones pour sa mise en oeuvre. Nous proposons également un nouvel algorithme hybride synchrone pour la méthode Implicitly Restarted Arnoldi Method. Des environnements de calcul global basés sur une approche Grid-RPC constituent un bon choix pour élaborer des programmes de résolution de problèmes sur les grilles de calcul. Un exemple typique de tels environnements est le système NetSolve. L'utilisation de ce type d'architectures nécessite la définition de nouveaux algorithmes. Une adaptation de MERAM asynchrone au système de calcul global NetSolve a été conçue. Nous avons montré que les algorithmes asynchrones de type MERAM sont très bien adaptés au calcul global. Nous avons mis en évidence un certain nombre de problèmes ouverts concernant la programmation des algorithmes hybrides en calcul global
Sedrakian, Malhami Ani. "Vers une aide à la décision pour les méthodes itératives hybrides parallèles réutilisables." Paris 6, 2005. http://www.theses.fr/2005PA066074.
Full textMarcotte, Jean-Philippe. "Méthodes itératives pour la résolution, par éléments finis, du problème de Stokes non linéaire." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2000. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp01/MQ57419.pdf.
Full textWane, Bocar Amadou. "Adaptation de maillages et méthodes itératives avec applications aux écoulements à surfaces libres turbulents." Thesis, Université Laval, 2012. http://www.theses.ulaval.ca/2012/29353/29353.pdf.
Full textEl, Maliki Abderrahman. "Résolution de problèmes aux limites à l'aide de méthodes itératives hiérarchiques à préconditionneur variable." Thesis, Université Laval, 2007. http://www.theses.ulaval.ca/2007/24692/24692.pdf.
Full textLaganier, Frank S. "Simulation dynamique de procédés : méthodes itératives dynamiques pour la résolution de systèmes algébro-différentiels." Toulouse, INPT, 1993. http://www.theses.fr/1993INPT038G.
Full textSelva, Gersendre. "Méthodes itératives pour l'intégration implicite des équations de l'aérothermochimie sur des maillages non-structurés." Châtenay-Malabry, Ecole centrale de Paris, 1998. http://www.theses.fr/1998ECAP0647.
Full textEl-Moallem, Rola. "Extrapolation vectorielle et applications aux méthodes itératives pour résoudre des équations algébriques de Riccati." Thesis, Lille 1, 2013. http://www.theses.fr/2013LIL10180/document.
Full textIn this thesis, we are interested in the study of polynomial extrapolation methods and their application as convergence accelerators on iterative methods to solve Algebraic Riccati equations arising in transport theory . In such applications, polynomial extrapolation methods succeed to accelerate the convergence of these iterative methods, even when the convergence turns to be extremely slow.The advantage of these methods of extrapolation is that they use a sequence of vectors which is not necessarily convergent, or which converges very slowly to create a new sequence which can admit a quadratic convergence. Furthermore, the development of restarted (or cyclic) methods allows to limit the cost of computations and storage. An interpretation of the critical case where the Jacobian matrix at the required solution is singular and quadratic convergence turns to linear is made. This problem can be overcome by applying a suitable shift technique. The original equation is transformed into an equivalent Riccati equation where the singularity is removed while the matrix coefficients maintain the same structure as in the original equation. The nice feature of this transformation is that the new equation has the same solution as the original one although the new Jacobian matrix at the solution is nonsingular. Numerical experiments and comparisons which confirm the effectiveness of the new approaches are reported
Dumoulin, Christian. "Sur quelques méthodes itératives et universelles de calcul d'orbites de mouvements képlériens perturbés ou non." Bordeaux 1, 1994. http://www.theses.fr/1994BOR10584.
Full textWeill-Duflos, Christine. "Optimisation de méthodes de résolution itératives de grands systèmes linéaires creux sur machines massivement parallèles." Paris 6, 1994. http://www.theses.fr/1994PA066284.
Full textLecouvez, Matthieu. "Méthodes itératives de décomposition de domaine sans recouvrement avec convergence géométrique pour l'équation de Helmholtz." Palaiseau, Ecole polytechnique, 2015. https://theses.hal.science/tel-01229546/document.
Full textIn this thesis, we are concerned by the mathematical aspects of iterative methods based on domain decomposition and applied to the numerical simulation of wave propagation in frequency domain. More specifically, we are interested in developing optimized transmission conditions that guarantee the exponential convergence of the iterative process. Such a convergence requires non local transmission operators since they should correspond, at least formally, to pseudo differential operators of order 1. A localization method is proposed to reduce the cost caused by these operators, while keeping their properties and thus the exponential convergence of the iterative method. In a general framework, the convergence of the domain decomposition methods is established for a class of operators verifying some properties such as positiveness and isomorphism between Sobolev spaces. Then, we propose several operators, which depend on parameters, that verify the required properties to achieve exponential convergence. A first kind of operator is based on norms of Sobolev spaces of fractional orders, while a second kind of operator is derived from Riesz potential (fractional powers of Laplace-Beltrami operator). Finally, we propose a numerical scheme that allows us to apply the developed theory on a finite elements method. A modal analysis of simple geometries is used to validate the theoretical conclusions of exponential convergence, and then several numerical experiments highlight the advantages of the proposed transmission conditions, especially when high precision is needed
Perez, Saul. "Application à des problèmes d'environnement radar de méthodes itératives de résolution d'un problème électromagnétique par partition." Toulouse, INSA, 2007. http://eprint.insa-toulouse.fr/archive/00000168/.
Full textWith a multiplication of complex and large objects (>>λ) the RCS computation of these objects is important for the aviation community to determine the impact on radio navigation systems like radar. These objects are far greater than the wavelength but asymptotic methods cannot be applied to solve this type of complex problem. The integral method is a popular choice for solving electromagnetic scattering by an arbitrary object. However it is well known that the traditional integral method suffers from the storage requirement (increasing in the order of O(N²)) and computational complexity (increasing in the order of O(N3)) for large scale problem. These limitations impose on the one hand the use of a cluster to share out the storage and the computational complexity cost and on the other hand the use of an iterative method (increasing in the order of O(N²)) combined with an accelerating method to reduce the computational complexity cost. We propose a method for RCS computation of long internally complex, dielectric objects such as wind turbine blades. The method proposed is an improved iterative algorithm whose convergence is proven. Moreover the algorithm can be easily adapted to parallel computation. This method is based into classical integral method and interface decomposition. The integral method consist in the description of electromagnetic field in terms of electric and magnetic currents defined on the surface of the electromagnetic scatterer. The interface decomposition consist in the decomposition of the surface of the scatterer in different zones. In order to accelerate the convergence rate we propose the use of three accelerating methods. The first accelerating method allows the elimination of the internal degrees of freedom. The second accelerating method (matrix compression QR) accelerates all the matrix vector products used in the preconditioning procedure as well as in the GMRES iterative resolution. The third one consists in using a “geometric-neighboring” preconditioner adapted to the physical aspect of the problem
Drummond, Lewis Leroy Anthony. "Résolution de systèmes linéaires creux par des méthodes itératives par blocs dans des environnements distribués hétérogènes." Toulouse, INPT, 1995. http://www.theses.fr/1995INPT098H.
Full textToure, Carine. "Capitalisation pérenne de connaissances industrielles : Vers des méthodes de conception incrémentales et itératives centrées sur l’activité." Thesis, Lyon, 2017. http://www.theses.fr/2017LYSEI095/document.
Full textIn this research, we are interested in the question of sustainability of the use of knowledge management systems (KMS) in companies. KMS are those IT environments that are set up in companies to share and build common expertise through collaborators. Findings show that, despite the rigor employed by companies in the implementation of these KMS, the risk of knowledge management initiatives being unsuccessful, particularly related to the acceptance and continuous use of these environments by users remains prevalent. The persistence of this fact in companies has motivated our interest to contribute to this general research question. As contributions to this problem, we have 1) identified from the state of the art, four facets that are required to promote the perennial use of a platform managing knowledge; 2) proposed a theoretical model of mixed regulation that unifies tools for self-regulation and tools to support change, and allows the continuous implementation of the various factors that stimulate the sustainable use of CMS; 3) proposed a design methodology, adapted to this model and based on the Agile concepts, which incorporates a mixed evaluation methodology of satisfaction and effective use as well as CHI tools for the completion of different iterations of our methodology; 4) implemented the methodology in real context at the Société du Canal de Provence, which allowed us to test its feasibility and propose generic adjustments / recommendations to designers for its application in context. The tool resulting from our implementation was positively received by the users in terms of satisfaction and usages
Haussaire, Jean-Matthieu. "Méthodes variationnelles d'ensemble itératives pour l'assimilation de données non-linéaire : Application au transport et la chimie atmosphérique." Thesis, Paris Est, 2017. http://www.theses.fr/2017PESC1097/document.
Full textData assimilation methods are constantly evolving to adapt to the various application domains. In atmospheric sciences, each new algorithm has first been implemented on numerical weather prediction models before being ported to atmospheric chemistry models. It has been the case for 4D variational methods and ensemble Kalman filters for instance. The new 4D ensemble variational methods (4D EnVar) are no exception. They were developed to take advantage of both variational and ensemble approaches and they are starting to be used in operational weather prediction centers, but have yet to be tested on operational atmospheric chemistry models.The validation of new data assimilation methods on these models is indeed difficult because of the complexity of such models. It is hence necessary to have at our disposal low-order models capable of synthetically reproducing key physical phenomenons from operational models while limiting some of their hardships. Such a model, called L95-GRS, has therefore been developed. It combines the simple meteorology from the Lorenz-95 model to a tropospheric ozone chemistry module with 7 chemical species. Even though it is of low dimension, it reproduces some of the physical and chemical phenomenons observable in real situations. A data assimilation method, the iterative ensemble Kalman smoother (IEnKS), has been applied to this model. It is an iterative 4D EnVar method which solves the full non-linear variational problem. This application validates 4D EnVar methods in the context of non-linear atmospheric chemistry, but also raises the first limits of such methods.After this experiment, results have been extended to a realistic atmospheric pollution prediction model. 4D EnVar methods, via the IEnKS, have once again shown their potential to take into account the non-linearity of the chemistry model in a controlled environment, with synthetic observations. However, the assimilation of real tropospheric ozone concentrations mitigates these results and shows how hard atmospheric chemistry data assimilation is. A strong model error is indeed attached to these models, stemming from multiple uncertainty sources. Two steps must be taken to tackle this issue.First of all, the data assimilation method used must be able to efficiently take into account the model error. However, most methods are developed under the assumption of a perfect model. To avoid this hypothesis, a new method has then been developed. Called IEnKF-Q, it expands the IEnKS to the model error framework. It has been validated on a low-order model, proving its superiority over data assimilation methods naively adapted to take into account model error.Nevertheless, such methods need to know the exact nature and amplitude of the model error which needs to be accounted for. Therefore, the second step is to use statistical tools to quantify this model error. The expectation-maximization algorithm, the naive and unbiased randomize-then-optimize algorithms, an importance sampling based on a Laplace proposal, and a Markov chain Monte Carlo simulation, potentially transdimensional, have been assessed, expanded, and compared to estimate the uncertainty on the retrieval of the source term of the Chernobyl and Fukushima-Daiichi nuclear power plant accidents.This thesis therefore improves the domain of 4D EnVar data assimilation by its methodological input and by paving the way to applying these methods on atmospheric chemistry models
Décamps, Jérôme. "Méthodes itératives par blocs pour la résolution de problèmes linéaires et non linéaires à structures partiellement séparables." Toulouse, INPT, 1997. http://www.theses.fr/1997INPT092H.
Full textLaouar, Abdelhamid. "Aspaect de l'analyse numérique de méthodes itératives de point fixe : : erreurs d'arrondi, accélération de convergence, sous-domaines." Besançon, 1988. http://www.theses.fr/1988BESA2039.
Full textAtallah, Nabil. "Analyse des méthodes itératives par points pour les problèmes de diffusion-convection approchés par les schémas compacts." Toulouse 3, 2002. http://www.theses.fr/2002TOU30010.
Full textZhang, Ye. "Méthodes itératives hybrides asynchrones sur plateformes de calcul hétérogènes pour la résolution accélérée de grands systèmes linéaires." Thesis, Lille 1, 2009. http://www.theses.fr/2009LIL10129/document.
Full textIn this thesis, we have studied an effective parallel hybrid method of solving linear systems, GMRES / LS-Arnoldi, which accelerates the convergence through knowledge of some eigenvalues calculated in paralled by the Arnoldi method in real cases. The asynchronous nature of this method has the advantage of working with a heterogeneous architecture. A study in complex cases is also done by transforming the complex matrix into a real matrix of double dimension. We have implemented our hybrid GMRES method and the general GMRES method on three different types of hardware platforms. They are respectively the IBM SP series supercomputer, a typically centralized hardware platform; Grid5000, a fully distributed hardware platform, and the Tsubame (Tokyo-tech Supercomputer and Ubiquitously Accessible Massstorage Environment) supercomputer, where some nodes are equipped with an accelerator card. We have tested the performance of general GMRES and hybrid GMRES on these three platforms, observing the influence of various parameters for the performance. A number of meaningful results have been obtained; we can not only improve the performance of parallel computing but also specify the direction of our future efforts
Chen, Langshi. "Méthode de Krylov itératives avec communication et efficacité énergétique optimisées sur machine hétérogène." Thesis, Lille 1, 2015. http://www.theses.fr/2015LIL10114/document.
Full textIterative methods are frequently used in extremely large scale linear problems, such solving linear systems or finding eigenvalue/eigenvectors of matrices. As these iterative methods require a substantial computational workload, they are normally deployed on large clusters of distributed memory architectures communicated via MPI. When the problem size scales up, the communication becomes a major bottleneck of reaching a higher scalability because of two reasons: 1) Many of the iterative methods rely on BLAS-2 low level matrix vector kernels that are communication intensive. 2) Data movement (memory access, MPI communication) is much slower than processor's speed. In case of sparse matrix operations such as Sparse Matrix Vector Multiplication (SpMV), the communication even replaces the computation as the dominant time cost. Furthermore, the advent of accelerators/coprocessors like Nvidia's GPU make computation cost more cheaper, while the communication cost remains high in such CPU-coprocessor heterogeneous systems. Thus, the first part of our work focus on the optimization of communication cost of iterative methods on heterogeneous clusters. Besides the communication cost, power wall becomes another bottleneck of future exascale computing in recent time. Researches indicate that a power-aware algorithmic implementation strategy could efficiently reduce the power dissipation of large clusters. We also explore the potential energy saving implementation of iterative methods in our experimentation. Finally, both the communication optimization and energy efficiency implementation would be integrated into a GMRES method, which demands an auto-tuning framework to maximize its performance
Chen, Long. "Méthodes itératives de reconstruction tomographique pour la réduction des artefacts métalliques et de la dose en imagerie dentaire." Thesis, Paris 11, 2015. http://www.theses.fr/2015PA112015/document.
Full textThis thesis contains two main themes: development of new iterative approaches for metal artifact reduction (MAR) and dose reduction in dental CT (Computed Tomography). The metal artifacts are mainly due to the beam-hardening, scatter and photon starvation in case of metal in contrast background like metallic dental implants in teeth. The first issue concerns about data correction on account of these effects. The second one involves the radiation dose reduction delivered to a patient by decreasing the number of projections. At first, the polychromatic spectra of X-ray beam and scatter can be modeled by a non-linear direct modeling in the statistical methods for the purpose of the metal artifacts reduction. However, the reconstruction by statistical methods is too much time consuming. Consequently, we proposed an iterative algorithm with a linear direct modeling based on data correction (beam-hardening and scatter). We introduced a new beam-hardening correction without knowledge of the spectra of X-ray source and the linear attenuation coefficients of the materials and a new scatter estimation method based on the measurements as well. Later, we continued to study the iterative approaches of dose reduction since the over-exposition or unnecessary exposition of irradiation during a CT scan has been increasing the patient's risk of radio-induced cancer. In practice, it may be useful that one can reconstruct an object larger than the field of view of scanner. We proposed an iterative algorithm on super-short-scans on multiple scans in this case, which contain a minimal set of the projections for an optimal dose. Furthermore, we introduced a new scanning mode of variant angular sampling to reduce the number of projections on a single scan. This was adapted to the properties and predefined interesting regions of the scanned object. It needed fewer projections than the standard scanning mode of uniform angular sampling to reconstruct the objet. All of our approaches for MAR and dose reduction have been evaluated on real data. Thanks to our MAR methods, the quality of reconstructed images was improved noticeably. Besides, it did not introduce some new artifacts compared to the MAR method of state of art NMAR [Meyer et al 2010]. We could reduce obviously the number of projections with the proposed new scanning mode and schema of super-short-scans on multiple scans in particular case
Darve, Eric. "Méthodes multipôles rapides : résolution des équations de Maxwell par formulations intégrales." Paris 6, 1999. http://www.theses.fr/1999PA066598.
Full textCrouzet, Laurent. "Résolution des équations de Maxwell tridimensionnelles en régime fréquentiel par éléments finis conformes, multiplicateurs de Lagrange et méthodes itératives." Paris 6, 1994. http://www.theses.fr/1994PA066089.
Full textFaye, Jean-Pierre. "Approche stochastique de la propagation des erreurs d'arrondi dans les méthodes itératives : Application a l'algorithme QR de calcul des valeurs propres." Paris 6, 1987. http://www.theses.fr/1987PA066368.
Full textRoussel, Adrien. "Parallélisation sur un moteur exécutif à base de tâches des méthodes itératives pour la résolution de systèmes linéaires creux sur architecture multi et many coeurs : application aux méthodes de types décomposition de domaines multi-niveaux." Thesis, Université Grenoble Alpes (ComUE), 2018. http://www.theses.fr/2018GREAM010/document.
Full textNumerical methods in reservoir engineering simulations lead to the resolution of unstructured, large and sparse linear systems. The performances of iterative methods employed in simulator to solve these systems are crucial in order to consider many more scenarios.In this work, we present a way to implement efficient parallel iterative methods on top of a task-based runtime system. It enables to simplify the development of methods while keeping control on parallelism management. We propose a linear algebra API which aims to implicitly express task dependencies: the semantic is sequential while the parallelism is implicit.We have extended the HARTS runtime system to monitor executions to better exploit NUMA architectures. Moreover, we implement a scheduling policy which exploits data locality for task placement. We have extended the API for KNL many-core systems while considering the various memory banks available. This work has led to the optimization of the SpMV kernel, one of the most time consuming operation in iterative methods.This work has been evaluated on iterative methods, and particularly on one method coming from domain decomposition. Hence, we demonstrate that the API enables to reach good performances on both multi-core and many-core architectures
Zenadi, Mohamed. "Méthodes hybrides pour la résolution de grands systèmes linéaires creux sur calculateurs parallèles." Thesis, Toulouse, INPT, 2013. http://www.theses.fr/2013INPT0126/document.
Full textWe are interested in solving large sparse systems of linear equations in parallel. Computing the solution of such systems requires a large amount of memory and computational power. The two main ways to obtain the solution are direct and iterative approaches. The former achieves this goal fast but with a large memory footprint while the latter is memory friendly but can be slow to converge. In this work we try first to combine both approaches to create a hybrid solver that can be memory efficient while being fast. Then we discuss a novel approach that creates a pseudo-direct solver that compensates for the drawback of the earlier approach. In the first chapters we take a look at row projection techniques, especially the block Cimmino method and examine some of their numerical aspects and how they affect the convergence. We then discuss the acceleration of convergence using conjugate gradients and show that a block version improves the convergence. Next, we see how partitioning the linear system affects the convergence and show how to improve its quality. We finish by discussing the parallel implementation of the hybrid solver, discussing its performance and seeing how it can be improved. The last two chapters focus on an improvement to this hybrid solver. We try to improve the numerical properties of the linear system so that we converge in a single iteration which results in a pseudo-direct solver. We first discuss the numerical properties of the new system, see how it works in parallel and see how it performs versus the iterative version and versus a direct solver. We finally consider some possible improvements to the solver. This work led to the implementation of a hybrid solver, our "ABCD solver" (Augmented Block Cimmino Distributed solver), that can either work in a fully iterative mode or in a pseudo-direct mode
Langet, Hélène. "Sampling and Motion Reconstruction in Three-dimensional X-ray Interventional Imaging." Phd thesis, Ecole Centrale Paris, 2013. http://tel.archives-ouvertes.fr/tel-00845148.
Full textZiane, Khodja Lilia. "Résolution de systèmes linéaires et non linéaires creux sur grappes de GPUs." Phd thesis, Université de Franche-Comté, 2013. http://tel.archives-ouvertes.fr/tel-00947627.
Full textNachaoui, Abdeljalil. "Contribution à l'analyse et à l'approximation des problèmes d'identification, de reconstruction et des systèmes d'équations elliptiques non linéaires." Habilitation à diriger des recherches, Université de Nantes, 2002. http://tel.archives-ouvertes.fr/tel-00002635.
Full textFerreira, Lago Rafael. "A study on block flexible iterative solvers with applications to Earth imaging problem in geophysics." Phd thesis, Toulouse, INPT, 2013. http://oatao.univ-toulouse.fr/10055/1/Ferreira.pdf.
Full textRuatta, Olivier. "Dualité algébrique, structures et applications." Phd thesis, Université de la Méditerranée - Aix-Marseille II, 2002. http://tel.archives-ouvertes.fr/tel-00002243.
Full textDelvare, Franck. "Une méthode inverse itérative à effet régularisant évanescent." Poitiers, 2000. http://www.theses.fr/2000POIT2312.
Full textHaidar, Azzam. "Sur l'extensibilité parallèle de solveurs linéaires hybrides pour des problèmes tridimensionels de grandes tailles." Phd thesis, Institut National Polytechnique de Toulouse - INPT, 2008. http://tel.archives-ouvertes.fr/tel-00347948.
Full textAhmed, Ouameur Messaoud. "Méthodes d'estimation de canal et de détection itérative pour les communications CDMA." Thèse, Université du Québec à Trois-Rivières, 2006. http://depot-e.uqtr.ca/1811/1/030077881.pdf.
Full textGuo, Jialin. "Estimation de la distribution énergétique induite par un faisceau d'électrons dans un matériau métallique : application au cas du soudage d'un acier." Lorient, 2005. http://www.theses.fr/2005LORIS046.
Full textThis work is concerned with thermal study of the electron beam welding for steel. The objective is to evaluate the energy distribution (source term) in the liquid zone by an inverse approach. The direct thermo-metallurgic problem is treated as non linear in two 2D supplementary approaches : a longitudinal section and a transversal section. The metallurgic transformations are modelled through CCT diagrams. The estimation of source term is based on the measurements form the edge of the fused zone. The parameters of a gaussian source are estimated by the Levenberg Marquardt method. Then in the case of a non defined source, the iterative regularization method is used. A fine experimental work is developed afterward. The results obtained through the experimental measurements present the relevance of the theoretic developments and allow validating the shape of the energy distribution of source term
Poulin, Nicolas M. "Méthode variationnelle itérative pour calculer le spectre vibrationnel de molécules polyatomiques." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1996. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp04/nq21501.pdf.
Full textRavier, Béatrice. "Méthodes performantes de résolution de grands systèmes linéaires." Paris 11, 1986. http://www.theses.fr/1986PA112215.
Full textThis work deals essentially with fast solving methods of large linear systems for an elliptic problem given in a domain included in R2, and composed with one, then several rectangles (domain having L-form or "fork"-domain). In the last chapter, the case of general domain is approached. The aim of the mathematical methods described for effecting the resolution on composite domains is to display subproblems on every component rectangle, where the fast solving algorithms, specially written for those elementary domains can be used (Fourier Analysis (FA), Cyclic Reduction (CH) and FACR(1)). The problems of association of domains are considered in two different ways: the rectangles have or have not the same mesh. For this second case, more interesting, in a first time, is the research of adapted interpolations (linear, of degree 2) required for the discretization of the equation on the in terface entered upon; then the problem is considered in an ether point of view, using the integral equations method which allows not to take care of the mesh, and which also allows to avoid difficulties of approach of the solution on the interface. The elliptic problem in general domains is solved by imbedding of this last one in a rectangle, where the situation is easy
Sanghavi, Chaitanya. "FETI methods for acoustic problems with porous materials." Thesis, Le Mans, 2020. http://www.theses.fr/2020LEMA1021.
Full textSound absorbing materials such as foams are widelyused in many industrial and domestic applications toabsorb undesirable sound. One needs to perform many calculations to get desirable properties of thetreatment using optimization strategies.The state-of-the-art computational models requireprohibitively high computational time. Theproblematic of this PhD is to reduce thecomputational time for such models to speed updesign calculations.This document is a synthesis of the work carried outin this direction. The problem is addressed usingDomain Decompostion methods (DDM). It consists ofsplitting the original problem into small parts referredto as subdomains. A partial solution is computed onthese subdomains to match the global continuity inthe domain of interest. Different DDM methods are benchmarked in termsof performance and scalability , specific for porousmaterials. Any DDM consists of two major costs, thefactorization of the subdomains and iterative part forthe global convergence. A novel factorizationstrategy is implemented and applied in 2D and 3Dto demonstrate savings in time compared toconventional approaches. In the second part, themethod is further improved to reduce the iterativecosts for a series of calculations.A final workflow is proposed to make thecomputationa cost of these models afforable withinindustrial timeframes
Le, Lepvrier Benoît. "Hybridation de la FDTD à Double Grille (DG-FDTD) avec l'Optique Physique Itérative (IPO) - Application à la simulation d'antennes environnées positionnées sur des platesformes de grandes dimensions." Thesis, Rennes, INSA, 2014. http://www.theses.fr/2014ISAR0011/document.
Full textThis thesis aims at extending the Dual-Grid FDTD (DG-FDTD) application domain via its hybridization with the Iterative Physical Optics (IPO) method. This research was motivated by the need to evaluate accurately and efficiently the antenna pattern of surrounded antennas installed on large platforms (satellite, vehicle, space launcher). Overview on numerical method involved in this class of problem revealed DG-FDTD has interesting features. This method allows precise and efficient wide-Band simulations of surrounded antennas. However, this method remains costly for electrically large problems, especially because of its rigorous formulation. This thesis assessed the limitations of DG-FDTD and then put forward its inability to resolve antenna on platform problems. To answer this issue, a hybrid scheme combining DG-FDTD with IPO is proposed in this thesis. DG-FDTD/IPO divides the initial simulation into two successive simulations. The antenna and its vicinity are firstly analyzed with DG-FDTD, and then IPO is used to analyze the platform. The two simulations are interfaced using the equivalence principle. This new method is first validated using a canonical scenario. Then, it is applied to the computation of electromagnetic radiation pattern in two antenna on platform problems (antenna on vehicle especially). The method is then exploited to effectively analyze the radiation pattern of a surrounded antenna mounted on a platform. Two improvements are finely proposed in this thesis for DGFDTD/ IPO. The first one aims at taking into account for the backward coupling between the antenna region and the metallic platform. This improvement implies a coarse description of the antenna region in the IPO simulation. The second improvement concerns the modeling of the currents in the shadow areas of the platform. This improvement answers to the need to analyze precisely antenna-On-Launcher problems. Indeed IPO do not compute currents in shadow areas. Well, for this kind of problem, shadow areas represent almost all the platform. A new method based on IPO and called Domains Sequential Processing is proposed. This method is first validated using a canonical scenario involving a cylinder. Then it is successfully applied to the analysis of a spatial launcher
Borel, Sophie. "Etude d'une équation intégrale stabilisée pour la résolution itérative de problèmes de diffraction d'ondes harmoniques en électromagnétisme." Paris 11, 2006. http://www.theses.fr/2006PA112041.
Full textThis thesis is a contribution to the iterative solution of scattering problems of harmonic electromagnetic waves by perfectly electrically conducting bodies. The aim is to build new integral equations dedicated to this problem that are intrinsically well-conditioned, well-suited to a fast iterative resolution, which is not achievable for classical equations. In order to do this, we parametrize the Maxwell equations solution as the electromagnetic field generated by a combination of electric and magnetic potentials, the same ones appearing in the classical combined sources equation (CSIE), but now coupled with an operator instead of scalar coefficients. The so-built equation can then be considered as a generalization of the CSIE equation. This formulation depends on the choice of the coupling operator, which is designed to approximate the obstacle exterior admittance. We benefit from the increasing localization of the scattering phenomena with the frequency to propose local approximations of the admittance dedicated to the high frequency regime. This new equation is then well-posed provided that the localization is correctly adapted to the frequency. Numerical experiments, part of them being realized for industrial obstacles, show that this formulation yields better conditioned linear systems than classical equations, which translates in a faster iterative resolution
Zounon, Mawussi. "On numerical resilience in linear algebra." Thesis, Bordeaux, 2015. http://www.theses.fr/2015BORD0038/document.
Full textAs the computational power of high performance computing (HPC) systems continues to increase by using huge number of cores or specialized processing units, HPC applications are increasingly prone to faults. This study covers a new class of numerical fault tolerance algorithms at application level that does not require extra resources, i.e., computational unit or computing time, when no fault occurs. Assuming that a separate mechanism ensures fault detection, we propose numerical algorithms to extract relevant information from available data after a fault. After data extraction, well chosen part of missing data is regenerated through interpolation strategies to constitute meaningful inputs to numerically restart the algorithm. We have designed these methods called Interpolation-restart techniques for numerical linear algebra problems such as the solution of linear systems or eigen-problems that are the inner most numerical kernels in many scientific and engineering applications and also often ones of the most time consuming parts. In the framework of Krylov subspace linear solvers the lost entries of the iterate are interpolated using the available entries on the still alive nodes to define a new initial guess before restarting the Krylov method. In particular, we consider two interpolation policies that preserve key numerical properties of well-known linear solvers, namely the monotony decrease of the A-norm of the error of the conjugate gradient or the residual norm decrease of GMRES. We assess the impact of the fault rate and the amount of lost data on the robustness of the resulting linear solvers.For eigensolvers, we revisited state-of-the-art methods for solving large sparse eigenvalue problems namely the Arnoldi methods, subspace iteration methods and the Jacobi-Davidson method, in the light of Interpolation-restart strategies. For each considered eigensolver, we adapted the Interpolation-restart strategies to regenerate as much spectral information as possible. Through intensive experiments, we illustrate the qualitative numerical behavior of the resulting schemes when the number of faults and the amount of lost data are varied; and we demonstrate that they exhibit a numerical robustness close to that of fault-free calculations. In order to assess the efficiency of our numerical strategies, we have consideredan actual fully-featured parallel sparse hybrid (direct/iterative) linear solver, MaPHyS, and we proposed numerical remedies to design a resilient version of the solver. The solver being hybrid, we focus in this study on the iterative solution step, which is often the dominant step in practice. The numerical remedies we propose are twofold. Whenever possible, we exploit the natural data redundancy between processes from the solver toperform an exact recovery through clever copies over processes. Otherwise, data that has been lost and is not available anymore on any process is recovered through Interpolationrestart strategies. These numerical remedies have been implemented in the MaPHyS parallel solver so that we can assess their efficiency on a large number of processing units (up to 12; 288 CPU cores) for solving large-scale real-life problems