As redes ópticas de comunicação têm possibilitado, cada vez mais, o oferecimento de serviços do tipo faixa larga. A rede de comunicação totalmente óptica está se tornando a meta tecnológica mais ambiciosa. Grandes esforços têm sido concentrados no desenvolvimento e aperfeiçoamento de materiais e dispositivos não-recíprocos, como isoladores e circuladores constituem uma importante classe de dispositivos ópticos. Os isoladores são utilizados em sistemas ópticos para prevenir a reflexão de luz em lasers e amplificadores. Os circuladores são empregados em esquemas de derivação de sinais que utilizam multiplexação em comprimento de onda (WDM). O funcionamento destes dispositivos é baseado nas propriedades de materiais magnetoópticos. As metas desta dissertação são apresentar as principais características dos materiais magnetoópticos, explorando as características não-recíprocas dos modos TM. Guias planares e tridimensionais são analisados neste trabalho. Para tanto, são obtidas as expressões das componentes dos campos e as equações características dos modos de interesse em estruturas planares por meio da técnica de matriz de transferência (TMT). A análise de propagação de onda em guias planares com materiais magnetoópticos é feita por meio de simulação numérica empregando o método de propagação de feixe (BPM) baseado em diferenças finitas (FD) e o esquema de Crank-Nicholson (CN) na discretização da solução da equação de onda. A condição de fronteira transparente (TBC) é incorporada ao algoritmo FD-BPM com a finalidade de se evitar reflexões de ondas eletromagnéticas para dentro da janela computacional. O método do índice efetivo é empregado na análise de guias de onda tridimensionais do tipo rib. / Optical communication networks have allowed a continuous increase of broadband services offer. The all-optical communication networks are becoming the most ambitious technological goal. Great efforts have been concentrated on the materiaIs and devices development and improvement to make it possible. Nonreciprocal devices, such as isolators and circulators constitute an important class of optical devices. Isolators are used in optical systems to avoid reflection of light in lasers and amplifiers. Circulators are used in signal derivation schemes that use wavelength division multiplexing (WDM). The operation of these devices is based on the properties of magnetooptic materiaIs. The purposes of this dissertation are to present the main features of the magnetooptic materiaIs as well as to analyze the eletromagnetic wave propagation in magnetooptic waveguides, exploring nonreciprocal features of TM modes. Planar and three-dimensional waveguides are analysed in this present study. Therefore expressions of electromagnetic field components and characteristic equations of the modes of interest in planar structures are obtained by using transfer matrix technique (TMT). The wave propagation analysis in planar magnetooptic waveguides is realized by using the finite-difference beam propagation method (FD-BPM) and Crank-Nicholson scheme (CN) applied to wave equation solution discretization. In order to avoid electromagnetic wave reflection into computational window, the transparent boundary condition (TBC) is incorporated to the FD-BPM formalism. The effective index method (EIM) is used in the analysis of three-dimensional rib magnetooptic waveguides.

Este trabalho visa o projeto, modelagem e fabricação de estruturas multicamadas baseadas em polímeros para aplicações como sensores ópticos integrados. A grande motivação para este trabalho está no fato de que estas estruturas, diferente da geometria Mach-Zehnder, dispensam o uso de litografia pois são completamente planares. Isto permite uma diminuição no custo de fabricação dos dispositivos além de permitir que estruturas mais curtas sejam utilizadas. Em se tratando de óptica integrada, as dimensões reduzidas da estrutura impõem severas penalidades no processo de lançamento de potência óptica na mesma. Por isso, será considerada neste estudo a utilização de prismas para o acoplamento de entrada do acoplador. Esta técnica, além de reduzir drasticamente os problemas de alinhamento decorrentes de acoplamento convencional do tipo \"End Fire\", permite uma transferência de potência óptica superior a 80% entre a fibra e o guia retangular. As variações na transferência de potência entre os guias de ondas da estrutura multicamada serão medidas através de um fotodetetor MSM integrado ao sensor. Este detetor, além de sua extrema facilidade de fabricação e baixos custos, torna o conjunto sensor mais robusto. Em se tratando de uma proposta de plataforma para sensores, serão também investigadas possíveis aplicações para esta estrutura, como por exemplo: refratômetro integrado, sensor de glicose, sensor de adulteração de combustível, etc. A análise das estruturas será efetuada por meio de técnicas de modelagem analíticas (Técnica da Matriz de Transferência - TMT e Teoria de Modos Acoplados - TMA), e numéricas (Método da Propagação de Feixe de Ângulo Largo formulado em Diferenças Finitas - WA-FD-BPM). Esta última permite que a estrutura do fotodetetor seja levada em consideração simultaneamente nas simulações. / This work concerns with design, modeling and fabrication of polymer based planar multilayer structures for integrated optic sensor applications. The motivation for this work is that planar multilayer structures, differently from the Mach-Zehnder geometry, do not require a lithographic process. As a consequence, significantly cheaper and shorter structures can be realized. The reduced dimensions of the structure, by its turn, pose a severe penalty in terms of optical power coupling. Therefore, this investigation will focus primarily on input (and output) prism coupling configuration. This technique, besides reducing the alignment requirements observed for conventional end-fire coupling, allows optical power coupling efficiency as high as 80% from fiber to rectangular waveguide. Any optical power transfer between the waveguides of the multilayer structure will be detected by an MSM photodetector integrated with the sensor. This low cost photodetector, besides improving the structure robustness, is quite ease to fabricate. Since the idea of this work is to develop a platform for integrated optic sensors, it will also be investigated possible applications for this structure, such as: integrated optic refractometer, glucose sensor and fuel adulteration sensor. The analysis of such structures will be carried out by means of analythical (Transfer Matrix Technique-TMT and Coupled Mode Theory-CMT) and numerical (Wide-Angle Finite Difference Beam Propagation Method-WA-FD-BPM) modeling techniques. The WA-FD-BPM technique allows one to simulate the multilayer waveguide and the MSM photodetector simultaneously.

Utilizando a técnica da matriz de transferência e as ideias de finite-size scaling, de renormalização fenomenológica e de invariância conforme estudamos dois modelos de polímeros interagentes na rede quadrada. Em ambos, a atividade de um monômero pertencente ao polímero vale x = e. Quando as interações são entre as ligações primeiras-vizinhas que pertencem ao polímero definimos um fator de Boltzmann associado à interação como sendo = e-l, onde l é a energia de interação entre ligações. Se as interações são entre os monômeros que pertencem a sítios primeiros-vizinhos mas não-consecutivos o fator de Boltzmann associado à interação é z = e-m , sendo m a energia de interação entre os monômeros. Através do estudo de pares de tiras de larguras L-L obtivemos estimativas para os diagramas de fases dos dois modelos. Ambos apresentam três fases: uma fase não-polimerizada, uma fase polimerizada usual e uma fase polimerizada densa, na qual o polímero se encontra colapsado. Nessa fase, o polímero assume uma configuração que maximiza o número de interações, de modo que a densidade de sítios ocupados pelo polímero tende a 1 e a densidade de interações assume valores muito próximos de 1. Os diagramas de fases dos dois modelos são qualitativamente semelhantes, havendo diferenças quantitativas, já esperadas. As transições entre a fase não polimerizada e a fase polimerizada densa são de primeira ordem. A transição entre a fase não polimerizada e a fase polimerizada usual é de segunda ordem, e um ponto dessa fronteira, que corresponde ao modelo sem interações ( = 1 ou z = 1, dependendo do modelo), é bem conhecido [1-3] e vale xc = 0,37905227 ± 0,00000012. A transição entre as fases polimerizadas é de primeira ordem para valores pequenos de x, mas muda para uma transição de segunda ordem quando x aumenta. Nessa fronteira existe um ponto tri crítico, que foi estimado em ( xPTC = 1,5 ± 0,1, yPTC = 1,1 ± 0,1 ), para o modelo de ligações interagentes. No caso do modelo de monômeros interagentes, não foi possível obter uma estimativa conclusiva a respeito da localização do ponto tri crítico. No encontro das três fronteiras existe um ponto crítico terminal, no qual terminam a linha de transições de segunda ordem entre a fase não polimerizada e a fase polimerizada usual, a linha de transições descontínuas entre as duas fases polimerizadas e a linha de transições de primeira ordem entre as fases não polimerizadas e a polimerizada densa. Os valores estimados por nós são (xPCT = 0,244 ± 0,002, yPCT = 3,86 ± 0,03) para o modelo de ligações interagentes e (xPCT = 0,345 ± 0,001, zPCT = 1,52 ± 0,001) para monômeros interagentes. O ponto , no qual termina a fronteira de segunda ordem entre a fase não-polimeriza e a fase polimerizada usual e onde ocorre pela primeira vez a transição de colapso é um ponto crítico terminal em ambos os modelos. Os expoentes críticos e associados à fronteira entre a fase não polimerizada e polimerizada usual também foram calculados, e encontramos os valores = 0,7507 ± 0,0008 e = 0,2082 ± 0,0004, para y = 1, e = 0,7498 ± 0,0004 e = 0,205 ± 0,003, para y = 1,2, para o modelo de ligações interagentes. Para o modelo de monômero interagentes, os dados foram = 0,7507 ± 0,0007 e = 0,2089 ± 0,0009, para z = 1, e = 0,7500 ± 0,0004 e = 0,205 ± 0,008, para z = 1,2. Observando os valores dos expoentes, vemos que eles ficam constantes dentro das barras de erros, de modo que a transição é uma transição de segunda ordem usual. Os valores concordam muito bem com os valores esperados, que são (exatamente) = 3/4 e = 5/24 [4]. / Using the transfer matrix technique, finite-size scaling, phenomenological renormalization group, and conformal invariance ideas, we studied the thermodynamic behavior of two interacting models of polymers on the square lattice. In both models one monomer that belongs to the polymer has an activity x = e. When the interactions are between first neighbor bonds that belong to the polymer, we define a Boltzmann factor y = e-l, where l is the interaction energy between two bonds. If the interactions are between monomers located at first neighbor but nonconsecutive sites, the associated Boltzmann factor is z = e-m, where m is the interaction energy between two monomers. We consider pairs of strips of widths L-L\' and found estimates for the phase diagrams of both models. They have three phases: a non-polymerized phase, an usual polymerized phase and a dense phase, in which the polymer is colapsed. In this phase, the configuration of the polymer is that maximizes the number of interactions, and the density of sites occupied by the polymer goes to 1, while the density of interactions is very close to 1. The phase diagrams of two models are qualitatively similar, but. there are quantitative differences between them, as we already expected. The transition between non polymerized phase and dense phase is of first order. The transition between non-polymerized phase and usual polymerized phase is of second order, and one point of this frontier, which corresponds to the non-interacting model (y = 1 ou z = 1, depending on the specific model), is well known [1-3J and has the value xc = 0,37905227 ± 0,00000012. The transition between the two polymerized phases is of first order for small values of x, and it changes to a second order transition when x increases. At this frontier there is a tri critical point, and we found ( xTCP = 1,5 ± 0,1, yTCP = 1,1 ± 0,1 ) for the interacting bond model. It was not possible to obtain a conclusive estimation of the location of the tri critical point for the model of interacting monomers. At the point that all transition lines meet there is a critical endpoint, in which the second order transition line between non polymerized phase and usual polymerized phase, the first order transition line between polymerized phases and the first order transition line between non polymerized phase and dense polymerized phases finish. We found (xCEP = 0,244 ± 0,002, yCEP = 3,86 ± 0,03) for the interacting bond model and (xCEP = 0,345 ± 0,001, zCEP = 1,52 ± 0,001) for the interacting monomer model. The point is where ends the second order transition between non polymerized phase and usual polymerized phase and at this point the collapse transition happens at the first time. Then, in our models, the point is a critical endpoint. We also found the critical exponents and of the second order transition line between non polymerized phase and usual 1\'olymerizcd phase. The values we obtained are = 0,7507 ± 0,0008 and = 0,2082 ± 0,0004, for y = 1, and = 0,7498 ± 0,0004 and = 0,205 ± 0,003, for y = 1,2, to the model of interacting bonds. The interacting monomers model has = 0,7507 ± 0,0007 and = 0,2089 ± 0,0009, for z = 1, and = 0,7500 ± 0,0004 and = 0,205 ± 0,008, for z = 1,2. Looking at these results we can see that the exponents remain constant within error bars, thus the transition is a usual second order transition. Furthermore, these values are in a very good agreement with the expected values, which are = 3/4 and = 5/24 [4].

This thesis is concentrated on mathematical modeling of high frequency pulsations in pump turbines, which are the source of high-cycle continuous stress of the spiral casing cover, wicket gates and runner. There are proposed the solutions using the transfer matrix for the tube with a constant and conical cross-section. The paper compares variations of cylindrical and conical tubes, changes in boundary conditions. There are the models of PSPP Dlouhé Stráně made only of cylindrical tubes comparing to the model with cylindrical and conical tubes

Untersucht wird der Auswirkung von Anisotropie auf den unordnungsinduzierten Metall-Isolator-Übergang (MIÜ) im Rahmen des dreidimensionalen Anderson-Modells der Lokalisierung für (schwach) gekoppelte Ebenen bzw. Ketten. Mittels numerischer Verfahren (Lanczos- und Transfer-Matrix-Methode) werden Eigenwerte und -vektoren bzw. die Lokalisierungslänge berechnet. Zur Bestimmung des kritischen Exponenten dieses Phasenüberganges 2. Ordnung wird ein allgemeiner Skalenansatz verwendet, der auch den Einfluss einer irrelevanten Skalenvariablen und Nichtlinearitäten berücksichtigt. Ein Kapitel untersucht die verwendeten numerischen Verfahren, verschiedene Methoden werden verglichen und die Portierbarkeit zu Parallelrechnern diskutiert. Der MIÜ wird mit zwei unabhängigen Methoden charakterisiert: Eigenwertstatistik und Transfer-Matrix-Methode. Die Systemgrößenunabhängigkeit der betrachteten Größen am Phasenübergang wird benutzt um den MIÜ zu identifizieren. Sie resultiert aus der Multifraktalität der kritischen Eigenzustände, die für den isotropen Fall bis zu einer Systemgröße von 111^3 Gitterplätzen gezeigt wird. Es stellt sich heraus, daß der MIÜ auch bei sehr starker Anisotropie existiert und bereits bei geringerer Potentialunordnung als im isotropen Fall auftritt. Für den Fall sehr schwach gekoppelter Ebenen wird gezeigt, daß der kritische Exponent mit dem des isotropen Falles übereinstimmt und damit die übliche Einteilung in Universalitätsklassen bestätigt.

No description available.

Intake and Exhaust system noise makes a huge contribution to the interior and exterior noise of automobiles. There are a number of linear acoustic tools developed by institutions and industries to predict the acoustic properties of intake and exhaust systems. The present project discusses and validates, through measurements, the proper modelling of these systems using BOOST-SID and discusses the ideas to properly convert a geometrical model of an exhaust muffler to an acoustic model. The various elements and their properties are also discussed. When it comes to Acoustic properties there are several parameters that describe the performance of a muffler, the Transmission Loss (TL) can be useful to check the validity of a mathematical model but when we want to predict the actual acoustic behavior of a component after it is installed in a system and subjected to operating conditions then we have to determine other properties like Attenuation, Insertion loss etc,. Zero flow and Mean flow (M=0.12) measurements of these properties were carried out for mufflers ranging from simple expansion chambers to complex geometry using two approaches 1) Two Load technique 2) Two Source location technique. For both these cases, the measured transmission losses were compared to those obtained from BOOST-SID models. The measured acoustic properties compared well with the simulated model for almost all the cases.

Les modèles sur réseau comme ceux de la percolation, d’Ising et de Potts servent à décrire les transitions de phase en deux dimensions. La recherche de leur solution analytique passe par le calcul de la fonction de partition et la diagonalisation de matrices de transfert. Au point critique, ces modèles statistiques bidimensionnels sont invariants sous les transformations conformes et la construction de théories des champs conformes rationnelles, limites continues des modèles statistiques, permet un calcul de la fonction de partition au point critique. Plusieurs chercheurs pensent cependant que le paradigme des théories des champs conformes rationnelles peut être élargi pour inclure les modèles statistiques avec des matrices de transfert non diagonalisables. Ces modèles seraient alors décrits, dans la limite d’échelle, par des théories des champs logarithmiques et les représentations de l’algèbre de Virasoro intervenant dans la description des observables physiques seraient indécomposables. La matrice de transfert de boucles D_N(λ, u), un élément de l’algèbre de Temperley- Lieb, se manifeste dans les théories physiques à l’aide des représentations de connectivités ρ (link modules). L’espace vectoriel sur lequel agit cette représentation se décompose en secteurs étiquetés par un paramètre physique, le nombre d de défauts. L’action de cette représentation ne peut que diminuer ce nombre ou le laisser constant. La thèse est consacrée à l’identification de la structure de Jordan de D_N(λ, u) dans ces représentations. Le paramètre β = 2 cos λ = −(q + 1/q) fixe la théorie : β = 1 pour la percolation et √2 pour le modèle d’Ising, par exemple. Sur la géométrie du ruban, nous montrons que D_N(λ, u) possède les mêmes blocs de Jordan que F_N, son plus haut coefficient de Fourier. Nous étudions la non diagonalisabilité de F_N à l’aide des divergences de certaines composantes de ses vecteurs propres, qui apparaissent aux valeurs critiques de λ. Nous prouvons dans ρ(D_N(λ, u)) l’existence de cellules de Jordan intersectorielles, de rang 2 et couplant des secteurs d, d′ lorsque certaines contraintes sur λ, d, d′ et N sont satisfaites. Pour le modèle de polymères denses critique (β = 0) sur le ruban, les valeurs propres de ρ(D_N(λ, u)) étaient connues, mais les dégénérescences conjecturées. En construisant un isomorphisme entre les modules de connectivités et un sous-espace des modules de spins du modèle XXZ en q = i, nous prouvons cette conjecture. Nous montrons aussi que la restriction de l’hamiltonien de boucles à un secteur donné est diagonalisable et trouvons la forme de Jordan exacte de l’hamiltonien XX, non triviale pour N pair seulement. Enfin nous étudions la structure de Jordan de la matrice de transfert T_N(λ, ν) pour des conditions aux frontières périodiques. La matrice T_N(λ, ν) a des blocs de Jordan intrasectoriels et intersectoriels lorsque λ = πa/b, et a, b ∈ Z×. L’approche par F_N admet une généralisation qui permet de diagnostiquer des cellules intersectorielles dont le rang excède 2 dans certains cas et peut croître indéfiniment avec N. Pour les blocs de Jordan intrasectoriels, nous montrons que les représentations de connectivités sur le cylindre et celles du modèle XXZ sont isomorphes sauf pour certaines valeurs précises de q et du paramètre de torsion v. En utilisant le comportement de la transformation i_N^d dans un voisinage des valeurs critiques (q_c, v_c), nous construisons explicitement des vecteurs généralisés de Jordan de rang 2 et discutons l’existence de blocs de Jordan intrasectoriels de plus haut rang. / Lattice models such as percolation, the Ising model and the Potts model are useful for the description of phase transitions in two dimensions. Finding analytical solutions is done by calculating the partition function, which in turn requires finding eigenvalues of transfer matrices. At the critical point, the two dimensional statistical models are invariant under conformal transformations and the construction of rational conformal field theories, as the continuum limit of these lattice models, allows one to compute the partition function at the critical point. Many researchers think however that the paradigm of rational conformal conformal field theories can be extended to include models with non diagonalizable transfer matrices. These models would then be described, in the scaling limit, by logarithmic conformal field theories and the representations of the Virasoro algebra coming into play would be indecomposable. We recall the construction of the double-row transfer matrix D_N(λ, u) of the Fortuin-Kasteleyn model, seen as an element of the Temperley-Lieb algebra. This transfer matrix comes into play in physical theories through its representation in link modules (or standard modules). The vector space on which this representation acts decomposes into sectors labelled by a physical parameter d, the number of defects, which remains constant or decreases in the link representations. This thesis is devoted to the identification of the Jordan structure of D_N(λ, u) in the link representations. The parameter β = 2 cos λ = −(q + 1/q) fixes the theory : for instance β = 1 for percolation and √2 for the Ising model. On the geometry of the strip with open boundary conditions, we show that D_N(λ, u) has the same Jordan blocks as its highest Fourier coefficient, F_N. We study the non-diagonalizability of F_N through the divergences of some of the eigenstates of ρ(F_N) that appear at the critical values of λ. The Jordan cells we find in ρ(D_N(λ, u)) have rank 2 and couple sectors d and d′ when specific constraints on λ, d, d′ and N are satisfied. For the model of critical dense polymers (β = 0) on the strip, the eigenvalues of ρ(D_N(λ, u)) were known, but their degeneracies only conjectured. By constructing an isomorphism between the link modules on the strip and a subspace of spin modules of the XXZ model at q = i, we prove this conjecture. We also show that the restriction of the Hamiltonian to any sector d is diagonalizable, and that the XX Hamiltonian has rank 2 Jordan cells when N is even. Finally, we study the Jordan structure of the transfer matrix T_N(λ, ν) for periodic boundary conditions. When λ = πa/b and a, b ∈ Z×, the matrix T_N(λ, ν) has Jordan blocks between sectors, but also within sectors. The approach using F_N admits a generalization to the present case and allows us to probe the Jordan cells that tie different sectors. The rank of these cells exceeds 2 in some cases and can grow indefinitely with N. For the Jordan blocks within a sector, we show that the link modules on the cylinder and the XXZ spin modules are isomorphic except for specific curves in the (q, v) plane. By using the behavior of the transformation i_N^d in a neighborhood of the critical values (q_c, v_c), we explicitly build Jordan partners of rank 2 and discuss the existence of Jordan cells with higher rank.

Structural colours arise from the interaction of visible light with nano-structured materials. The occurrence of such structures in nature has been known for over a century, but it is only in the last few decades that the study of natural photonic structures has fully matured due to the advances in imagining techniques and computational modelling. Even though a plethora of different colour-producing architectures in a variety of species has been investigated, a few significant questions are still open: how do these structures develop in living organisms? Does disorder play a functional role in biological photonics? If so, is it possible to say that the optical response of natural disordered photonics has been optimised under evolutionary pressure? And, finally, can we exploit the well-adapted photonic design principles that we observe in Nature to fabricate functional materials with optimised scattering response? In my thesis I try to answer the questions above: I microscopically investigate $\textit{in vivo}$ the growth of a cuticular multilayer, one of the most common colour-producing strategies in nature, in the green beetles $\textit{Gastrophysa viridula}$ showing how the interplay between different materials varies during the various life stages of the beetles; I further investigate two types of disordered photonic structures and their biological role, the random array of spherical air inclusions in the eggshells of the honeyguide $\textit{Prodotiscus regulus}$, a species under unique evolutionary pressure to produce blue eggs, and the anisotropic chitinous network of fibres in the white beetle $\textit{Cyphochilus}$, the whitest low-refractive index material; finally, inspired by these natural designs, I fabricate and study light transport in biocompatible highly-scattering materials.

Les modèles sur réseau comme ceux de la percolation, d’Ising et de Potts servent à décrire les transitions de phase en deux dimensions. La recherche de leur solution analytique passe par le calcul de la fonction de partition et la diagonalisation de matrices de transfert. Au point critique, ces modèles statistiques bidimensionnels sont invariants sous les transformations conformes et la construction de théories des champs conformes rationnelles, limites continues des modèles statistiques, permet un calcul de la fonction de partition au point critique. Plusieurs chercheurs pensent cependant que le paradigme des théories des champs conformes rationnelles peut être élargi pour inclure les modèles statistiques avec des matrices de transfert non diagonalisables. Ces modèles seraient alors décrits, dans la limite d’échelle, par des théories des champs logarithmiques et les représentations de l’algèbre de Virasoro intervenant dans la description des observables physiques seraient indécomposables. La matrice de transfert de boucles D_N(λ, u), un élément de l’algèbre de Temperley- Lieb, se manifeste dans les théories physiques à l’aide des représentations de connectivités ρ (link modules). L’espace vectoriel sur lequel agit cette représentation se décompose en secteurs étiquetés par un paramètre physique, le nombre d de défauts. L’action de cette représentation ne peut que diminuer ce nombre ou le laisser constant. La thèse est consacrée à l’identification de la structure de Jordan de D_N(λ, u) dans ces représentations. Le paramètre β = 2 cos λ = −(q + 1/q) fixe la théorie : β = 1 pour la percolation et √2 pour le modèle d’Ising, par exemple. Sur la géométrie du ruban, nous montrons que D_N(λ, u) possède les mêmes blocs de Jordan que F_N, son plus haut coefficient de Fourier. Nous étudions la non diagonalisabilité de F_N à l’aide des divergences de certaines composantes de ses vecteurs propres, qui apparaissent aux valeurs critiques de λ. Nous prouvons dans ρ(D_N(λ, u)) l’existence de cellules de Jordan intersectorielles, de rang 2 et couplant des secteurs d, d′ lorsque certaines contraintes sur λ, d, d′ et N sont satisfaites. Pour le modèle de polymères denses critique (β = 0) sur le ruban, les valeurs propres de ρ(D_N(λ, u)) étaient connues, mais les dégénérescences conjecturées. En construisant un isomorphisme entre les modules de connectivités et un sous-espace des modules de spins du modèle XXZ en q = i, nous prouvons cette conjecture. Nous montrons aussi que la restriction de l’hamiltonien de boucles à un secteur donné est diagonalisable et trouvons la forme de Jordan exacte de l’hamiltonien XX, non triviale pour N pair seulement. Enfin nous étudions la structure de Jordan de la matrice de transfert T_N(λ, ν) pour des conditions aux frontières périodiques. La matrice T_N(λ, ν) a des blocs de Jordan intrasectoriels et intersectoriels lorsque λ = πa/b, et a, b ∈ Z×. L’approche par F_N admet une généralisation qui permet de diagnostiquer des cellules intersectorielles dont le rang excède 2 dans certains cas et peut croître indéfiniment avec N. Pour les blocs de Jordan intrasectoriels, nous montrons que les représentations de connectivités sur le cylindre et celles du modèle XXZ sont isomorphes sauf pour certaines valeurs précises de q et du paramètre de torsion v. En utilisant le comportement de la transformation i_N^d dans un voisinage des valeurs critiques (q_c, v_c), nous construisons explicitement des vecteurs généralisés de Jordan de rang 2 et discutons l’existence de blocs de Jordan intrasectoriels de plus haut rang. / Lattice models such as percolation, the Ising model and the Potts model are useful for the description of phase transitions in two dimensions. Finding analytical solutions is done by calculating the partition function, which in turn requires finding eigenvalues of transfer matrices. At the critical point, the two dimensional statistical models are invariant under conformal transformations and the construction of rational conformal field theories, as the continuum limit of these lattice models, allows one to compute the partition function at the critical point. Many researchers think however that the paradigm of rational conformal conformal field theories can be extended to include models with non diagonalizable transfer matrices. These models would then be described, in the scaling limit, by logarithmic conformal field theories and the representations of the Virasoro algebra coming into play would be indecomposable. We recall the construction of the double-row transfer matrix D_N(λ, u) of the Fortuin-Kasteleyn model, seen as an element of the Temperley-Lieb algebra. This transfer matrix comes into play in physical theories through its representation in link modules (or standard modules). The vector space on which this representation acts decomposes into sectors labelled by a physical parameter d, the number of defects, which remains constant or decreases in the link representations. This thesis is devoted to the identification of the Jordan structure of D_N(λ, u) in the link representations. The parameter β = 2 cos λ = −(q + 1/q) fixes the theory : for instance β = 1 for percolation and √2 for the Ising model. On the geometry of the strip with open boundary conditions, we show that D_N(λ, u) has the same Jordan blocks as its highest Fourier coefficient, F_N. We study the non-diagonalizability of F_N through the divergences of some of the eigenstates of ρ(F_N) that appear at the critical values of λ. The Jordan cells we find in ρ(D_N(λ, u)) have rank 2 and couple sectors d and d′ when specific constraints on λ, d, d′ and N are satisfied. For the model of critical dense polymers (β = 0) on the strip, the eigenvalues of ρ(D_N(λ, u)) were known, but their degeneracies only conjectured. By constructing an isomorphism between the link modules on the strip and a subspace of spin modules of the XXZ model at q = i, we prove this conjecture. We also show that the restriction of the Hamiltonian to any sector d is diagonalizable, and that the XX Hamiltonian has rank 2 Jordan cells when N is even. Finally, we study the Jordan structure of the transfer matrix T_N(λ, ν) for periodic boundary conditions. When λ = πa/b and a, b ∈ Z×, the matrix T_N(λ, ν) has Jordan blocks between sectors, but also within sectors. The approach using F_N admits a generalization to the present case and allows us to probe the Jordan cells that tie different sectors. The rank of these cells exceeds 2 in some cases and can grow indefinitely with N. For the Jordan blocks within a sector, we show that the link modules on the cylinder and the XXZ spin modules are isomorphic except for specific curves in the (q, v) plane. By using the behavior of the transformation i_N^d in a neighborhood of the critical values (q_c, v_c), we explicitly build Jordan partners of rank 2 and discuss the existence of Jordan cells with higher rank.