Cette thèse présente de nouveaux algorithmes de diagonalisation conjointe par similitude. Cesalgorithmes permettent, entre autres, de résoudre le problème de décomposition canonique polyadiquede tenseurs. Cette décomposition est particulièrement utilisée dans les problèmes deséparation de sources. L’utilisation de la diagonalisation conjointe par similitude permet de paliercertains problèmes dont les autres types de méthode de décomposition canonique polyadiquesouffrent, tels que le taux de convergence, la sensibilité à la surestimation du nombre de facteurset la sensibilité aux facteurs corrélés. Les algorithmes de diagonalisation conjointe par similitudetraitant des données complexes donnent soit de bons résultats lorsque le niveau de bruit est faible,soit sont plus robustes au bruit mais ont un coût calcul élevé. Nous proposons donc en premierlieu des algorithmes de diagonalisation conjointe par similitude traitant les données réelles etcomplexes de la même manière. Par ailleurs, dans plusieurs applications, les matrices facteursde la décomposition canonique polyadique contiennent des éléments exclusivement non-négatifs.Prendre en compte cette contrainte de non-négativité permet de rendre les algorithmes de décompositioncanonique polyadique plus robustes à la surestimation du nombre de facteurs ou lorsqueces derniers ont un haut degré de corrélation. Nous proposons donc aussi des algorithmes dediagonalisation conjointe par similitude exploitant cette contrainte. Les simulations numériquesproposées montrent que le premier type d’algorithmes développés améliore l’estimation des paramètresinconnus et diminue le coût de calcul. Les simulations numériques montrent aussi queles algorithmes avec contrainte de non-négativité améliorent l’estimation des matrices facteurslorsque leurs colonnes ont un haut degré de corrélation. Enfin, nos résultats sont validés à traversdeux applications de séparation de sources en télécommunications numériques et en spectroscopiede fluorescence. / This thesis introduces new joint eigenvalue decomposition algorithms. These algorithms allowamongst others to solve the canonical polyadic decomposition problem. This decomposition iswidely used for blind source separation. Using the joint eigenvalue decomposition to solve thecanonical polyadic decomposition problem allows to avoid some problems whose the others canonicalpolyadic decomposition algorithms generally suffer, such as the convergence rate, theoverfactoring sensibility and the correlated factors sensibility. The joint eigenvalue decompositionalgorithms dealing with complex data give either good results when the noise power is low, orthey are robust to the noise power but have a high numerical cost. Therefore, we first proposealgorithms equally dealing with real and complex. Moreover, in some applications, factor matricesof the canonical polyadic decomposition contain only nonnegative values. Taking this constraintinto account makes the algorithms more robust to the overfactoring and to the correlated factors.Therefore, we also offer joint eigenvalue decomposition algorithms taking advantage of thisnonnegativity constraint. Suggested numerical simulations show that the first developed algorithmsimprove the estimation accuracy and reduce the numerical cost in the case of complexdata. Our numerical simulations also highlight the fact that our nonnegative joint eigenvaluedecomposition algorithms improve the factor matrices estimation when their columns have ahigh correlation degree. Eventually, we successfully applied our algorithms to two blind sourceseparation problems : one concerning numerical telecommunications and the other concerningfluorescence spectroscopy.

Made available in DSpace on 2014-06-11T19:33:33Z (GMT). No. of bitstreams: 0 Previous issue date: 2007-12-18Bitstream added on 2014-06-13T20:45:15Z : No. of bitstreams: 1 pimentel_ecg_dr_jabo.pdf: 351573 bytes, checksum: 4ed13e3cadb5bcefe457a98cb28d9baf (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / O objetivo da pesquisa descrita nesta tese foi estudar possíveis aplicações do método das rotações modificadas de Givens na solução de sistemas de equações lineares tipicamente observados em problemas de melhoramento animal. Duas aplicações foram consideradas: a predição de valores genéticos com base em informação fenotípica e genealógica, por meio da metodologia dos modelos mistos; e a predição de valores genéticos com base em informação molecular, obtida pela genotipagem de painéis densos de SNPs. Na primeira aplicação, delineou-se o emprego de um modelo animal reduzido, combinado a uma ordenação do sistema que permitiu uma abordagem multi-frontal de decomposição. As matrizes frontais foram definidas como sendo as partes da triangular superior pertinentes a cada rebanho. Com isso, o problema pôde ser desmembrado em n subproblemas em que n é o número de rebanhos. Um conjunto de programas foi desenvolvido de modo a decompor as matrizes de dados de cada rebanho independentemente, e depois combinar as informações de todos eles na solução do sistema triangular geral, por retro-substituição. Concluiu-se que o método pode ser empregado em um sistema para atualização de predições de valor genético sob modelo animal reduzido, em que se aninham os efeitos de vacas dentro de rebanhos. Na segunda aplicação, comparou-se o emprego das rotações de Givens com o método do Gradiente Conjugado, na solução de sistemas lineares envolvidos na estimação de efeitos de SNPs em valores genéticos. O método das rotações demandou menos tempo de processamento e mais memória. Concluiu-se que, dado o crescente avanço em capacidade computacional, o método das rotações pode ser um método numérico viável e apresenta a vantagem de permitir o cálculo dos erros-padrão das estimativas. / The aim of this study was to investigate possible applications of the modified Givens rotations on the solution of linear systems that typically arise in animal breeding problems. Two applications were considered: prediction of breeding values based on phenotypes and relationships, using mixed model methods; and prediction of breeding values based on molecular information, using genotypes from high density SNP chips. In the first application, the use of a reduced animal model, combined with a specific ordering of the system, made it possible to apply a multi-frontal decomposition approach. The frontal matrices were defined as the parts of the upper triangular corresponding to each herd. In this way, the problem could be partitioned into n subproblems, where n is the number of herds. A set of programs was developed in order to factorize the data matrix of each herd independently, and then combine the information from all of them while solving the overall triangular system, by back-substitution. The conclusion was that Givens rotations can be used as a numerical method for updating predicted breeding values under a reduced animal model, if dam effects are nested within herds. In the second application, the modified Givens rotations were compared to the Conjugate Gradient method for solving linear systems that arise in the estimation of SNP effects on breeding values. Givens rotations required less processing time but a greater amount of high speed memory. The conclusion was that, given the increasing rate of advance in computer power, Givens rotations can be regarded as a feasible numerical method which presents the advantage that it allows for the calculation of standard errors of estimates.

Dans ce manuscrit de thèse, nous introduisons les avancées récentes sur la décomposition en matrices (et tenseurs) de rang faible et parcimonieuse ainsi que les contributions pour faire face aux principaux problèmes dans ce domaine. Nous présentons d’abord un aperçu des méthodes matricielles et tensorielles les plus récentes ainsi que ses applications sur la modélisation d’arrière-plan et la segmentation du premier plan. Ensuite, nous abordons le problème de l’initialisation du modèle de fond comme un processus de reconstruction à partir de données manquantes ou corrompues. Une nouvelle méthodologie est présentée montrant un potentiel intéressant pour l’initialisation de la modélisation du fond dans le cadre de VSI. Par la suite, nous proposons une version « double contrainte » de l’ACP robuste pour améliorer la détection de premier plan en milieu marin dans des applications de vidéo-surveillance automatisées. Nous avons aussi développé deux algorithmes incrémentaux basés sur tenseurs afin d’effectuer une séparation entre le fond et le premier plan à partir de données multidimensionnelles. Ces deux travaux abordent le problème de la décomposition de rang faible et parcimonieuse sur des tenseurs. A la fin, nous présentons un travail particulier réalisé en conjonction avec le Centre de Vision Informatique (CVC) de l’Université Autonome de Barcelone (UAB). / This thesis introduces the recent advances on decomposition into low-rank plus sparse matrices and tensors, as well as the main contributions to face the principal issues in moving object detection. First, we present an overview of the state-of-the-art methods for low-rank and sparse decomposition, as well as their application to background modeling and foreground segmentation tasks. Next, we address the problem of background model initialization as a reconstruction process from missing/corrupted data. A novel methodology is presented showing an attractive potential for background modeling initialization in video surveillance. Subsequently, we propose a double-constrained version of robust principal component analysis to improve the foreground detection in maritime environments for automated video-surveillance applications. The algorithm makes use of double constraints extracted from spatial saliency maps to enhance object foreground detection in dynamic scenes. We also developed two incremental tensor-based algorithms in order to perform background/foreground separation from multidimensional streaming data. These works address the problem of low-rank and sparse decomposition on tensors. Finally, we present a particular work realized in conjunction with the Computer Vision Center (CVC) at Autonomous University of Barcelona (UAB).

<p dir="ltr">Matrix decompositions play a pivotal role in matrix computation and applications. While general dense matrix-vector multiplications and linear equation solvers are prohibitively expensive, matrix decompositions offer fast alternatives for matrices meeting specific properties. This dissertation delves into my contributions to two fast matrix multiplication algorithms and one fast linear equation solver algorithm tailored for certain matrices and applications, all based on efficient matrix decompositions. Fast dimensionality reduction methods in spectral clustering, based on efficient eigen-decompositions, are also explored.</p><p dir="ltr">The first matrix decomposition introduced is the "kernel-independent" interpolative decomposition butterfly factorization (IDBF), acting as a data-sparse approximation for matrices adhering to a complementary low-rank property. Constructible in $O(N\log N)$ operations for an $N \times N$ matrix via hierarchical interpolative decompositions (IDs), the IDBF results in a product of $O(\log N)$ sparse matrices, each with $O(N)$ non-zero entries. This factorization facilitates rapid matrix-vector multiplication in $O(N \log N)$ operations, making it a versatile framework applicable to various scenarios like special function transformation, Fourier integral operators, and high-frequency wave computation.</p><p dir="ltr">The second matrix decomposition accelerates matrix-vector multiplication for computing multi-dimensional Jacobi polynomial transforms. Leveraging the observation that solutions to Jacobi's differential equation can be represented through non-oscillatory phase and amplitude functions, the corresponding matrix is expressed as the Hadamard product of a numerically low-rank matrix and a multi-dimensional discrete Fourier transform (DFT) matrix. This approach utilizes $r^d$ fast Fourier transforms (FFTs), where $r = O(\log n / \log \log n)$ and $d$ is the dimension, resulting in an almost optimal algorithm for computing the multidimensional Jacobi polynomial transform.</p><p dir="ltr">An efficient numerical method is developed based on a matrix decomposition, Hierarchical Interpolative Factorization, for solving modified Poisson-Boltzmann (MPB) equations. Addressing the computational bottleneck of evaluating Green's function in the MPB solver, the proposed method achieves linear scaling by combining selected inversion and hierarchical interpolative factorization. This innovation significantly reduces the computational cost associated with solving MPB equations, particularly in the evaluation of Green's function.</p><p dir="ltr"><br></p><p dir="ltr">Finally, eigen-decomposition methods, including the block Chebyshev-Davidson method and Orthogonalization-Free methods, are proposed for dimensionality reduction in spectral clustering. By leveraging well-known spectrum bounds of a Laplacian matrix, the Chebyshev-Davidson methods allow dimensionality reduction without the need for spectrum bounds estimation. And instead of the vanilla Chebyshev-Davidson method, it is better to use the block Chebyshev-Davidson method with an inner-outer restart technique to reduce total CPU time and a progressive polynomial filter to take advantage of suitable initial vectors when available, for example, in the streaming graph scenario. Theoretically, the Orthogonalization-Free method constructs a unitary isomorphic space to the eigenspace or a space weighting the eigenspace, solving optimization problems through Gradient Descent with Momentum Acceleration based on Conjugate Gradient and Line Search for optimal step sizes. Numerical results indicate that the eigenspace and the weighted eigenspace are equivalent in clustering performance, and scalable parallel versions of the block Chebyshev-Davidson method and OFM are developed to enhance efficiency in parallel computing.</p>

Dans cette thèse, nous nous intéressons au système d’identification automatique spatial lequel est dédié à la surveillancemaritime par satellite. Ce système couvre une zone bien plus large que le système standard à terre correspondant àplusieurs cellules traditionnelles ce qui peut entraîner des risques de collision des données envoyées par des navireslocalisés dans des cellules différentes et reçues au niveau de l’antenne du satellite. Nous présentons différentes approchesafin de répondre au problème de collision considéré. Elles ne reposent pas toujours sur les mêmes hypothèses en ce quiconcerne les signaux reçus, et ne s’appliquent donc pas toutes dans les mêmes contextes (nombre de capteurs utilisés,mode semi-supervisé avec utilisation de trames d’apprentissage et information a priori ou mode aveugle, problèmes liés àla synchronisation des signaux, etc...).Dans un premier temps, nous proposons des méthodes permettant la séparation/dé-collision des messages en modèle surdéterminé(plus de capteurs que de messages). Elles sont fondées sur des algorithmes de décompositions matriciellesconjointes combinés à des détecteurs de points temps-fréquence (retard-fréquence Doppler) particuliers permettant laconstruction d’ensembles de matrices devant être (bloc) ou zéro (bloc) diagonalisées conjointement. En ce qui concerneles algorithmes de décompositions matricielles conjointes, nous proposons quatre nouveaux algorithmes de blocdiagonalisation conjointe (de même que leur version à pas optimal) fondés respectivement sur des algorithmesd’optimisation de type gradient conjugué, gradient conjugué pré-conditionné, Levenberg-Marquardt et Quasi-Newton. Lecalcul exact du gradient matriciel complexe et des matrices Hessiennes complexes est mené. Nous introduisonségalement un nouveau problème dénommé zéro-bloc diagonalisation conjointe non-unitaire lequel généralise le problèmedésormais classique de la zéro-diagonalisation conjointe non-unitaire. Il implique le choix d’une fonction de coût adaptéeet à nouveau le calcul de quantités telles que gradient matriciel complexe et les matrices Hessiennes complexes. Nousproposons ensuite trois nouveaux algorithmes à pas optimal fondés sur des algorithmes d’optimisation de type gradientconjugué, gradient conjugué pré-conditionné et Levenberg-Marquardt.Finalement, nous terminons par des approches à base de techniques de détection multi-utilisateurs conjointe susceptiblesde fonctionner en contexte sous-déterminé dans lequel nous ne disposons plus que d’un seul capteur recevantsimultanément plusieurs signaux sources. Nous commençons par développer une première approche par déflationconsistant à supprimer successivement les interférences. Nous proposons ensuite un deuxième mode opératoire fondéquant à lui sur l’estimateur du maximum de vraisemblance conjoint qui est une variante de l’algorithme de VITERBI. / This PHD thesis concerns the spatial automatic identification system dedicated to marine surveillance by satellite. Thissystem covers a larger area than the traditional system corresponding to several satellite cells. In such a system, there arerisks of collision of the messages sent by vessels located in different cells and received at the antenna of the samesatellite. We present different approaches to address the considered problem. They are not always based on the sameassumptions regarding the received signals and are not all applied in the same contexts (they depend on the number ofused sensors, semi-supervised mode with use of training sequences and a priori information versus blind mode, problemswith synchronization of signals, etc.). Firstly, we develop several approaches for the source separation/de-collision in theover-determined case (more sensors than messages) using joint matrix decomposition algorithms combined withdetectors of particular time-frequency (delay-Doppler frequency) points to build matrix sets to be joint (block) or zero(block) diagonalized. Concerning joint matrix decomposition algorithms, four new joint block-diagonalization algorithms(with optimal step-size) are introduced based respectively on conjugate gradient, preconditioned conjugate gradient,Levenberg-Marquardt and Quasi-Newton optimization schemes. Secondly, a new problem called non-unitary joint zeroblockdiagonalization is introduced. It encompasses the classical joint zero diagonalization problem. It involves thechoice of a well-chosen cost function and the calculation of quantities such as the complex gradient matrix and thecomplex Hessian matrices. We have therefore proposed three new algorithms (and their optimal step-size version) basedrespectively on conjugate gradient, preconditioned conjugate gradient and Levenberg-Marquardt optimization schemes.Finally, we suggest other approaches based on multi-user joint detection techniques in an underdetermined context wherewe have only one sensor receiving simultaneously several signals. First, we have developed an approach by deflationbased on a successive interferences cancelation technique. Then, we have proposed a second method based on the jointmaximum likelihood sequence estimator which is a variant of the VITERBI algorithm.

Submitted by Rosivalda Pereira (mrs.pereira@ufma.br) on 2017-06-23T21:14:46Z No. of bitstreams: 1 JonathanQueiroz.pdf: 631286 bytes, checksum: 2cab2a7d6e496bf574ddef1f49a77440 (MD5) / Made available in DSpace on 2017-06-23T21:14:46Z (GMT). No. of bitstreams: 1 JonathanQueiroz.pdf: 631286 bytes, checksum: 2cab2a7d6e496bf574ddef1f49a77440 (MD5) Previous issue date: 2016-03-08 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPQ) / The discrete Riccati algebraic equation has played an increasingly important role in optimal control theory and adaptive ltering. For this reason, various techniques have been developed to solve the DARE, for example the approach based on self vectors or approaches related to invariant subspaces [1], which require mathematical rigor and precision. However, these approaches present a number of problems, among them the fact that they can not be implemented in real-time due to its high computational cost to estimate the solution of DARE in many systems, especially systems with higher order three. In order to overcomes this problem, we propose to solve the DARE using as an estimator based on the sum of potential error pairs. The estimator is similar to the Recursive Least Squares (RLS), but with a better performance in terms of convergence speed and estimation accuracy without a signi- cant increase in computational complexity. The estimator is called Recursive Least Non-Squares (RLNS). One other aspect in unraveling the general DARE is to ensure that DARE is numerically well conditioned. To perform the numerical conditioning of DARE, a matrix decomposition technique known as Moore-Penrose inverse or generalized inverse is used. The proposed method is evaluated in a multivariate system 6th order corresponding to the wind generator. The method is evaluated under the numerical stability point of view and speed of convergence. / A equação algébrica Riccati discreta (discrete algebraic Riccati equation (DARE)) tem desempenhado uma papel cada vez mais importante na teoria de controle ótimo. Por esse motivo, varias técnicas tem sido desenvolvidas para solucionar a DARE, por exemplo a abordagem baseada em auto vetores ou ainda abordagens relacionadas a subespaços invariantes, as quais requerem rigor e precisão matemáticas. No entanto, estas abordagens apresentam uma serie de problemas, dentre eles, o fato de não poderem ser implementadas em tempo real devido ao seu alto custo computacional para estimar a solução da DARE em diversos sistemas, sobretudo sistemas com ordem superior a três. Com o intuito de contorna este problema, propomos solucionar a DARE utilizando um estimador baseado na soma das potencias pares do erro. O estimador e similar ao Recursive least squares (RLS), mas com um desempenho melhor em termos de velocidade de convergência e precisão de estimativa, sem aumento significativo da complexidade computacional. O estimador é denominado Recursive Least Non-Squares (RLNS). Um outra aspecto para que possamos solucionar a DARE de forma geral, e garantir que a DARE seja numericamente bem condicionada. Para efetuar o condicionamento numérico da DARE, será utilizada uma técnica de decomposição matricial conhecida como inversa de Moore-Penrose ou inversa generalizada. A metodologia proposta e avaliada em um sistema multivariavel de 6th ordem correspondente ao gerador eólico.

Banco do Nordeste do Brasil / A avaliaÃÃo educacional dispÃe de mÃtodos para a obtenÃÃo de dados que podem ser Ãteis para avaliar grupos de indivÃduos (alunos, professores, administradores, tÃcnicos e outros), projetos, produtos e materiais, instituiÃÃes e sistemas educacionais, nos seus diversos nÃveis e competÃncias. No campo da educaÃÃo em engenharia, os processos avaliativos podem ajudar os gestores a tomarem decisÃes e a realizarem mudanÃas em cursos de graduaÃÃo. Esta tese investiga de forma inÃdita uma nova abordagem para a anÃlise e interpretaÃÃo de dados no campo da educaÃÃo em engenharia com Ãnfase no processo de avaliaÃÃo, levando em consideraÃÃo dois aspectos de modo integrado: a) a percepÃÃo/opiniÃo dos estudantes sobre o contexto/ambiente educacional (Learning Context - LC) e b) os resultados/rendimentos obtidos pelos mesmos discentes (Learning Outcomes - LO). Para a realizaÃÃo desta pesquisa, foram coletados dados de estudantes do curso de graduaÃÃo em Engenharia de TeleinformÃtica (ETI) do Centro de Tecnologia (CT) da Universidade Federal do Cearà (UFC). Os dados de LC foram coletados a partir da aplicaÃÃo do instrumento SEEQ (Studentâs Evaluation of Educational Quality) da metodologia SETE (Student Evaluate Teaching Effetivecness). Os dados de LO foram coletados a partir das informaÃÃes dos resultados de desempenho da aprendizagem dos mesmos discentes. Na realizaÃÃo do processamento da informaÃÃo dos dados matriciais e tensoriais obtidos, foram utilizadas duas ferramentas matemÃticas: a decomposiÃÃo bilinear, por meio da AnÃlise de Componentes Principais (Principal Component Analysis - PCA) e a decomposiÃÃo multilinear tensorial por meio da AnÃlise de Fatores Paralelos (Parallel Factor Analysis - PARAFAC). Os resultados obtidos permitem identificar caracterÃsticas comuns e semelhanÃas em componentes curriculares, tanto em termos da percepÃÃo quanto do desempenho dos estudantes. Os modelos PCA e PARAFAC tambÃm demonstraram um potencial significativo para extrair informaÃÃes de dados relacionados com variÃveis latentes em contextos educativos. / Educational evaluation provides methods to obtain data that can be useful for evaluating groups of individuals (students, teachers, administrators, technicians and others), projects, products and materials, educational institutions and systems at different levels and skills. In engineering education, evaluation processes can help managers to make decisions and changes in undergraduate courses. This thesis investigates in unprecedented way a new approach to the analysis and interpretation of data in the field of engineering education with emphasis in the evaluation process, taking into account two aspects in an integrated manner: a) perception / opinion of students about the context / educational environment (Learning Context - LC) and b) the results / income earned by the same students (Learning outcomes - LO). For this research, we collected data related to undergraduate students in Teleinformatics Engineering (TEI), at Technology Center (CT) of the Federal University of Cearà (UFC). LC data were collected from the application of SEEQ (Studentâs Evaluation of Educational Quality) instrument of SETE (Student Teaching Evaluate Effetivecness) methodology. The LO data was collected from the information of the performance of the studentsâ learning outcomes. Carrying out the information processing of the obtained tensor and matrix data, we have used two mathematical tools: the bilinear decomposition, called Principal Component Analysis - PCA decomposition and the multilinear tensor decomposition by Parallel Factor Analysis - PARAFAC. The results allow us to identify common features and similarities in curriculum components, both in terms of perception as the performance of students. The PCA and PARAFAC models also showed significant potential to extract data information related to latent variables in educational settings.

Tesis por compendio / [ES] La idea de seguir una secuencia de pasos para lograr un resultado deseado es inherente a la naturaleza humana: desde que empezamos a andar, siguiendo una receta de cocina o aprendiendo un nuevo juego de cartas. Desde la antigüedad se ha seguido este esquema para organizar leyes, corregir escritos, e incluso asignar diagnósticos. En matemáticas a esta forma de pensar se la denomina 'algoritmo'. Formalmente, un algoritmo es un conjunto de instrucciones definidas y no-ambiguas, ordenadas y finitas, que permite solucionar un problema. Desde pequeños nos enfrentamos a ellos cuando aprendemos a multiplicar o dividir, y a medida que crecemos, estas estructuras nos permiten resolver diferentes problemas cada vez más complejos: sistemas lineales, ecuaciones diferenciales, problemas de optimización, etcétera. Hay multitud de algoritmos que nos permiten hacer frente a este tipo de problemas, como métodos iterativos, donde encontramos el famoso Método de Newton para buscar raíces; algoritmos de búsqueda para localizar un elemento con ciertas propiedades en un conjunto mayor; o descomposiciones matriciales, como la descomposición LU para resolver sistemas lineales. Sin embargo, estos enfoques clásicos presentan limitaciones cuando se enfrentan a problemas de grandes dimensiones, problema que se conoce como `la maldición de la dimensionalidad'. El avance de la tecnología, el uso de redes sociales y, en general, los nuevos problemas que han aparecido con el desarrollo de la Inteligencia Artificial, ha puesto de manifiesto la necesidad de manejar grandes cantidades de datos, lo que requiere el diseño de nuevos mecanismos que permitan su manipulación. En la comunidad científica, este hecho ha despertado el interés por las estructuras tensoriales, ya que éstas permiten trabajar eficazmente con problemas de grandes dimensiones. Sin embargo, la mayoría de métodos clásicos no están pensados para ser empleados junto a estas operaciones, por lo que se requieren herramientas específicas que permitan su tratamiento, lo que motiva un proyecto como este. El presente trabajo se divide de la siguiente manera: tras revisar algunas definiciones necesarias para su comprensión, en el Capítulo 3, se desarrolla la teoría de una nueva descomposición tensorial para matrices cuadradas. A continuación, en el Capítulo 4, se muestra una aplicación de dicha descomposición a grafos regulares y redes de mundo pequeño. En el Capítulo 5, se plantea una implementación eficiente del algoritmo que proporciona la nueva descomposición matricial, y se estudian como aplicación algunas EDP de orden dos. Por último, en los Capítulos 6 y 7 se exponen unas breves conclusiones y se enumeran algunas de las referencias consultadas, respectivamente. / [CA] La idea de seguir una seqüència de passos per a aconseguir un resultat desitjat és inherent a la naturalesa humana: des que comencem a caminar, seguint una recepta de cuina o aprenent un nou joc de cartes. Des de l'antiguitat s'ha seguit aquest esquema per a organitzar lleis, corregir escrits, i fins i tot assignar diagnòstics. En matemàtiques a aquesta manera de pensar se la denomina algorisme. Formalment, un algorisme és un conjunt d'instruccions definides i no-ambigües, ordenades i finites, que permet solucionar un problema. Des de xicotets ens enfrontem a ells quan aprenem a multiplicar o dividir, i a mesura que creixem, aquestes estructures ens permeten resoldre diferents problemes cada vegada més complexos: sistemes lineals, equacions diferencials, problemes d'optimització, etcètera. Hi ha multitud d'algorismes que ens permeten fer front a aquesta mena de problemes, com a mètodes iteratius, on trobem el famós Mètode de Newton per a buscar arrels; algorismes de cerca per a localitzar un element amb unes certes propietats en un conjunt major; o descomposicions matricials, com la descomposició DL. per a resoldre sistemes lineals. No obstant això, aquests enfocaments clàssics presenten limitacions quan s'enfronten a problemes de grans dimensions, problema que es coneix com `la maledicció de la dimensionalitat'. L'avanç de la tecnologia, l'ús de xarxes socials i, en general, els nous problemes que han aparegut amb el desenvolupament de la Intel·ligència Artificial, ha posat de manifest la necessitat de manejar grans quantitats de dades, la qual cosa requereix el disseny de nous mecanismes que permeten la seua manipulació. En la comunitat científica, aquest fet ha despertat l'interés per les estructures tensorials, ja que aquestes permeten treballar eficaçment amb problemes de grans dimensions. No obstant això, la majoria de mètodes clàssics no estan pensats per a ser emprats al costat d'aquestes operacions, per la qual cosa es requereixen eines específiques que permeten el seu tractament, la qual cosa motiva un projecte com aquest. El present treball es divideix de la següent manera: després de revisar algunes definicions necessàries per a la seua comprensió, en el Capítol 3, es desenvolupa la teoria d'una nova descomposició tensorial per a matrius quadrades. A continuació, en el Capítol 4, es mostra una aplicació d'aquesta descomposició a grafs regulars i xarxes de món xicotet. En el Capítol 5, es planteja una implementació eficient de l'algorisme que proporciona la nova descomposició matricial, i s'estudien com a aplicació algunes EDP d'ordre dos. Finalment, en els Capítols 6 i 7 s'exposen unes breus conclusions i s'enumeren algunes de les referències consultades, respectivament. / [EN] The idea of following a sequence of steps to achieve a desired result is inherent in human nature: from the moment we start walking, following a cooking recipe or learning a new card game. Since ancient times, this scheme has been followed to organize laws, correct writings, and even assign diagnoses. In mathematics, this way of thinking is called an algorithm. Formally, an algorithm is a set of defined and unambiguous instructions, ordered and finite, that allows for solving a problem. From childhood, we face them when we learn to multiply or divide, and as we grow, these structures will enable us to solve different increasingly complex problems: linear systems, differential equations, optimization problems, etc. There is a multitude of algorithms that allow us to deal with this type of problem, such as iterative methods, where we find the famous Newton Method to find roots; search algorithms to locate an element with specific properties in a more extensive set; or matrix decompositions, such as the LU decomposition to solve some linear systems. However, these classical approaches have limitations when faced with large-dimensional problems, a problem known as the `curse of dimensionality'. The advancement of technology, the use of social networks and, in general, the new problems that have appeared with the development of Artificial Intelligence, have revealed the need to handle large amounts of data, which requires the design of new mechanisms that allow its manipulation. This fact has aroused interest in the scientific community in tensor structures since they allow us to work efficiently with large-dimensional problems. However, most of the classic methods are not designed to be used together with these operations, so specific tools are required to allow their treatment, which motivates work like this. This work is divided as follows: after reviewing some definitions necessary for its understanding, in Chapter 3, the theory of a new tensor decomposition for square matrices is developed. Next, Chapter 4 shows an application of said decomposition to regular graphs and small-world networks. In Chapter 5, an efficient implementation of the algorithm provided by the new matrix decomposition is proposed, and some order two PDEs are studied as an application. Finally, Chapters 6 and 7 present some brief conclusions and list some of the references consulted. / María Mora Jiménez acknowledges funding from grant (ACIF/2020/269) funded by the Generalitat Valenciana and the European Social Found / Mora Jiménez, M. (2023). Numerical approximations with tensor-based techniques for high-dimensional problems [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/202604 / Compendio