京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第18783号 / 理博第4041号 / 新制||理||1582(附属図書館) / 31734 / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)教授 佐々 真一, 准教授 篠本 滋, 准教授 武末 真二 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

In this thesis, I study fluctuations and rare events of time-additive observables of discrete-time Markov chains on finite state spaces. The observable of interest is the mean node connectivity visited by a random walk running on instances of an Erdős-Rényi (ER) random graph. I implement and analyze the Adaptive Power Method (APM) which converges to the driven process, a biased random walk defined through a control parameter that simulates trajectories corresponding to rare events of the observable in the original dynamics. The APM demonstrates good convergence and accurately produces the desired quantities from a single trajectory. Due to the bulk-dangling-chain structure in the ER graph, the driven process seems to undergo a dynamical phase transition (DPT) for infinitely large graphs, meaning the behavior of the trajectories changes abruptly as the control parameter is varied. Observations show that the random walk visits two distinct phases, being de-localized in the bulk or localized in the chain. Through two simpler models capturing the bulk-dangling-chain property of the ER graph I study how the DPT occurs as the graph size increases. I observe that the trajectories of the driven process near the transition show intermittent behavior between the two phases. The diverging time scale of the DPT is found to be the average time that the random walk spends in a phase before it transitions to the other one. On the ER graph the trajectories are also intermittent but the form of the time scaling remains open due to computational limits on the graph size.

No description available.

There are some new challenges in system administration and design to optimize the resource management for a cloud based application. Some applications demand stringent performance requirements (e.g. delay and jitter bounds), while some applications exhibit bursty (volatile) workloads. This thesis proposes an epidemic model inspired (and continuous time Markov Chain based) framework, which can reproduce workload volatility namely the "buzz effects" (when there is a sudden increase of a content popularity) of a Video on Demand (VoD) system. Two estimation procedures (heuristic and a Markov Chain Monte Carlo (MCMC) based approach) have also been proposed in this work to calibrate the model against workload traces. Obtained model parameters from the calibration procedures reveal some interesting property of the model. Based on numerical simulations, precisions of both procedures have been analyzed, which show that both of them perform reasonably. However, the MCMC procedure outperforms the heuristic approach. This thesis also compares the proposed model with other existing models examining the goodness-of-fit of some statistical properties of real workload traces. Finally this work suggests a probabilistic resource provisioning approach based on a Large Deviation Principle (LDP). LDP statistically characterizes the buzz effects that causeextreme workload volatility. This analysis exploits the information obtained using the LDP of the VoD system for defining resource management policies. These policies may be of some interest to all stakeholders in the emerging context of cloud networking.

Esta tese trata a existência de Princpios de Grandes Desvios (PGD), no âmbito de sistemas fermiônicos em equilbrio. A motivação fsica detrás de nossos estudos são medidas experimentais de resistência elétrica de nanofios de silcio dopados com átomos de fósforo. Estas medidas mostram que efeitos quânticos no transporte de carga elétrica quase desaparecem para nanofios de comprimentos maiores que alguns nanômetros, mesmo para temperaturas muito baixas (4.2°K). A fim de provar matematicamente tal efeito, dividimos nosso trabalho em diversos passos: 1. No primeiro passo, para férmions não interagentes numa rede com desordem, mostramos que a incerteza quântica da densidade da corrente elétrica microscópica, em torno de seus valores macroscópicos(clássicos), é suprimida exponencialmente rápido em relação ao volume da região da rede onde um campo elétrico externo é aplicado. A desordem é modelada como um potencial elétrico aleatório juntamente com amplitudes aleatórias de saltos com valores complexos. O célebre modelo de Anderson de tight-binding é um exemplo particular do caso geral considerado aqui. Nossa análise matemática é baseada em estimativas de Combes-Thomas, o Teorema Ergódico de Akcoglu-Krengel e no formalismo de Grandes Desvios, em particular o Teorema de Gärtner-Ellis. 2. Em segundo lugar, provamos que, para férmions interagindo fracamente na rede, as funções geradoras J(s), s R de cumulantes de distribuições de probabilidades associadas com estados KMS pode ser escrito como o limite de logartmos de integrais gaussianas de Berezin. Mostramos que os determinantes das covariáncias associadas às integrais gaussianas são majorados uniformemente (via desigualdades de Hölder para normas Schatten). Tais covariâncias são também somáveis, em casos gerais de interesse, incluindo assim, sistemas que não são invariantes por translação. 3. No terceiro passo, analisamos expansões de logartmos de integrais gaussianas de Berezin, e assim combinando com métodos construtivos de teoria quântica de campos, mostramos a analiticidade de J(s) para s nas vizinhanças de 0. Finalmente, discutimos como combinar os passos 2-3, a fim de provar (matematicamente falando) os resultados experimentais mencionados acima para férmions interagindo em equilbrio. De fato, os resultados encontrados nesta tese, generalizam trabalhos prévios no âmbito do PGD usado para o estudo de sistemas quânticos. / This Thesis deals with the existence of Large Deviation Principles (LDP) in the scope of fermionic systems at equilibrium. The physical motivation beyond our studies are experimental measures of electric resistance of nanowires in silicon doped with phosphorus atoms. The latter demonstrate that quantum effects on charge transport almost disappear for nanowires of lengths larger than a few nanometers, even at very low temperature (4.2°K). In order to mathematically prove the latter, we divide our work in several steps: 1. In the first step, for noninteracting lattice fermions with disorder, we show that quantum uncertainty of microscopic electric current density around their (classical) macroscopic values is suppressed, exponentially fast with respect to the volume of the region of the lattice where an external electric field is applied. Disorder is modeled by a random external potential along with random, complex-valued, hopping amplitudes. The celebrated tight-binding Anderson model is one particular example of the general case considered here. Our mathematical analysis is based on Combes-Thomas estimates, the Akcoglu-Krengel ergodic theorem, and the large deviation formalism, in particular the Gärtner-Ellis theorem. 2. Secondly, we prove that for weakly interacting fermions on the lattice, the logarithm moment generating function J(s), s R of probability distributions associated with KMS states can be written as the limit of logarithms of Gaussian Berezin integrals. The covariances of the Gaussian integrals are shown to have a uniform determinant bound (via Hölder inequalities for Schatten norms) and to be summable in general cases of interest, including systems that are not translation invariant. 3. In the third step we analyze expansions of logarithms of Gaussian Berezin integrals, which combined with constructive methods of quantum field theory is useful to show the analyticity of J(s) for s in a neighborhood of 0. We finally discuss how to combine steps 2-3 in order to prove (mathematically speaking) for interacting fermions in equilibrium the experimental results above mentioned. In fact, the found results in this Thesis generalize previous works in the scope of LDP used to study quantum systems.

Cette thèse porte sur l'étude numérique des fluctuations extrêmes de la force de traînée exercée par un écoulement turbulent sur un corps immergé.Ce type d'évènement, très rare, est difficile à caractériser par le biais d'un échantillonnage direct, puisqu'il est alors nécessaire de simuler l'écoulement sur des durées extrêmement longues. Cette thèse propose une approche différente, basée sur l'application d'algorithmes d'échantillonnage d'événements rares. L'objectif de ces algorithmes, issus de la physique statistique, est de modifier la statistique d'échantillonnage des trajectoires d'un système dynamique, de manière à favoriser l'occurrence d'événements rares. Si ces techniques ont été appliquées avec succès dans le cas de dynamiques relativement simples, l'intérêt de ces algorithmes n'est à ce jour pas clair pour des dynamiques déterministes extrêmement complexes, comme c'est le cas pour les écoulement turbulents.Cette thèse présente tout d'abord une étude de la dynamique et de la statistique associée aux fluctuations extrêmes de la force de traînée sur un obstacle carré fixe immergé dans un écoulement turbulent à deux dimensions. Ce cadre simplifié permet de simuler la dynamique sur des durées très longues, permettant d'échantillonner un grand nombre de fluctuations dont l'amplitude est assez élevée pour être qualifiée d'extrême.Dans un second temps, l'application de deux algorithmes d’échantillonnage est présentée et discutée.Dans un premier cas, il est illustré qu'une réduction significative du temps de calcul d'extrêmes peut être obtenue. En outre, des difficultés liées à la dynamique de l'écoulement sont mises en lumière, ouvrant la voie au développement de nouveaux algorithmes spécifiques aux écoulements turbulents. / This thesis discusses the numerical simulation of extreme fluctuations of the drag force acting on an object immersed in a turbulent medium.Because such fluctuations are rare events, they are particularly difficult to investigate by means of direct sampling. Indeed, such approach requires to simulate the dynamics over extremely long durations.In this work an alternative route is introduced, based on rare events algorithms.The underlying idea of such algorithms is to modify the sampling statistics so as to favour rare trajectories of the dynamical system of interest.These techniques recently led to impressive results for relatively simple dynamics. However, it is not clear yet if such algorithms are useful for complex deterministic dynamics, such as turbulent flows.This thesis focuses on the study of both the dynamics and statistics of extreme fluctuations of the drag experienced by a square cylinder mounted in a two-dimensional channel flow.This simple framework allows for very long simulations of the dynamics, thus leading to the sampling of a large number of events with an amplitude large enough so as they can be considered extreme.Subsequently, the application of two different rare events algorithms is presented and discussed.In the first case, a drastic reduction of the computational cost required to sample configurations resulting in extreme fluctuations is achieved.Furthermore, several difficulties related to the flow dynamics are highlighted, paving the way to novel approaches specifically designed to turbulent flows.

Il y a de nouveaux défis dans l'administration et dans la conception des systèmes pour optimiser la gestion des ressources des applications basées en nuage Cloud Computing. Certaines applications demandent des performances rigoureuses (par exemple, par rapport aux retards et aux limites de la gigue), tandis que d'autres applications présentent des charges de travail en rafale (volatiles). Cette thèse propose un framework inspiré dans un modèle épidémique (et basé sur des Chaînes de Markov à Temps Continu), qui peut reproduire la volatilité de la charge de travail, à savoir les effets de buzz (quand il y a une augmentation soudaine de la popularité d'un contenu) d'un système de Vidéo à la Demande (VoD). Deux méthodes d'estimation (basés sur des heuristiques et des Chaînes de Markov Monte Carlo - MCMC) ont été également proposées dans ce travail, de façon à ajuster le modèle selon les comportements de la charge de travail. Les paramètres du modèle obtenus à partir des procédures d'étalonnage révèlent des propriétés intéressantes du modèle. Basé sur des simulations numériques, la précision des deux procédures a été analysée, en montrant que les deux présentent des performances raisonnables. Toutefois, la méthode MCMC dépasse la performance de l'approche heuristique. Cette thèse compare également le modèle proposé avec d'autres modèles existants, tout en examinant la qualité de l'ajustement de certaines propriétés statistiques sur des traces réelles de la charge de travail. Finalement, ce travail propose une approche probabiliste de provisionnement des ressources, basée sur le Principe de Grandes Déviations (LDP). LDP caractérise statistiquement les effets de buzz, qui causent de la volatilité extrême de la charge de travail. Cette analyse exploite les informations obtenues en utilisant le LPD du système VoD pour la définition des politiques de gestion des ressources. Ces politiques peuvent être intéressantes pour toutes les acteurs dans le nouveau contexte de l'informatique en nuage. / There are some new challenges in system administration and design to optimize the resource management for a cloud based application. Some applications demand stringent performance requirements (e.g. delay and jitter bounds), while some applications exhibit bursty (volatile) workloads. This thesis proposes an epidemic model inspired (and continuous time Markov Chain based) framework, which can reproduce workload volatility namely the "buzz effects" (when there is a sudden increase of a content popularity) of a Video on Demand (VoD) system. Two estimation procedures (heuristic and a Markov Chain Monte Carlo (MCMC) based approach) have also been proposed in this work to calibrate the model against workload traces. Obtained model parameters from the calibration procedures reveal some interesting property of the model. Based on numerical simulations, precisions of both procedures have been analyzed, which show that both of them perform reasonably. However, the MCMC procedure outperforms the heuristic approach. This thesis also compares the proposed model with other existing models examining the goodness-of-fit of some statistical properties of real workload traces. Finally this work suggests a probabilistic resource provisioning approach based on a Large Deviation Principle (LDP). LDP statistically characterizes the buzz effects that causeextreme workload volatility. This analysis exploits the information obtained using the LDP of the VoD system for defining resource management policies. These policies may be of some interest to all stakeholders in the emerging context of cloud networking.

Esta tese trata a existência de Princpios de Grandes Desvios (PGD), no âmbito de sistemas fermiônicos em equilbrio. A motivação fsica detrás de nossos estudos são medidas experimentais de resistência elétrica de nanofios de silcio dopados com átomos de fósforo. Estas medidas mostram que efeitos quânticos no transporte de carga elétrica quase desaparecem para nanofios de comprimentos maiores que alguns nanômetros, mesmo para temperaturas muito baixas (4.2°K). A fim de provar matematicamente tal efeito, dividimos nosso trabalho em diversos passos: 1. No primeiro passo, para férmions não interagentes numa rede com desordem, mostramos que a incerteza quântica da densidade da corrente elétrica microscópica, em torno de seus valores macroscópicos(clássicos), é suprimida exponencialmente rápido em relação ao volume da região da rede onde um campo elétrico externo é aplicado. A desordem é modelada como um potencial elétrico aleatório juntamente com amplitudes aleatórias de saltos com valores complexos. O célebre modelo de Anderson de tight-binding é um exemplo particular do caso geral considerado aqui. Nossa análise matemática é baseada em estimativas de Combes-Thomas, o Teorema Ergódico de Akcoglu-Krengel e no formalismo de Grandes Desvios, em particular o Teorema de Gärtner-Ellis. 2. Em segundo lugar, provamos que, para férmions interagindo fracamente na rede, as funções geradoras J(s), s R de cumulantes de distribuições de probabilidades associadas com estados KMS pode ser escrito como o limite de logartmos de integrais gaussianas de Berezin. Mostramos que os determinantes das covariáncias associadas às integrais gaussianas são majorados uniformemente (via desigualdades de Hölder para normas Schatten). Tais covariâncias são também somáveis, em casos gerais de interesse, incluindo assim, sistemas que não são invariantes por translação. 3. No terceiro passo, analisamos expansões de logartmos de integrais gaussianas de Berezin, e assim combinando com métodos construtivos de teoria quântica de campos, mostramos a analiticidade de J(s) para s nas vizinhanças de 0. Finalmente, discutimos como combinar os passos 2-3, a fim de provar (matematicamente falando) os resultados experimentais mencionados acima para férmions interagindo em equilbrio. De fato, os resultados encontrados nesta tese, generalizam trabalhos prévios no âmbito do PGD usado para o estudo de sistemas quânticos. / This Thesis deals with the existence of Large Deviation Principles (LDP) in the scope of fermionic systems at equilibrium. The physical motivation beyond our studies are experimental measures of electric resistance of nanowires in silicon doped with phosphorus atoms. The latter demonstrate that quantum effects on charge transport almost disappear for nanowires of lengths larger than a few nanometers, even at very low temperature (4.2°K). In order to mathematically prove the latter, we divide our work in several steps: 1. In the first step, for noninteracting lattice fermions with disorder, we show that quantum uncertainty of microscopic electric current density around their (classical) macroscopic values is suppressed, exponentially fast with respect to the volume of the region of the lattice where an external electric field is applied. Disorder is modeled by a random external potential along with random, complex-valued, hopping amplitudes. The celebrated tight-binding Anderson model is one particular example of the general case considered here. Our mathematical analysis is based on Combes-Thomas estimates, the Akcoglu-Krengel ergodic theorem, and the large deviation formalism, in particular the Gärtner-Ellis theorem. 2. Secondly, we prove that for weakly interacting fermions on the lattice, the logarithm moment generating function J(s), s R of probability distributions associated with KMS states can be written as the limit of logarithms of Gaussian Berezin integrals. The covariances of the Gaussian integrals are shown to have a uniform determinant bound (via Hölder inequalities for Schatten norms) and to be summable in general cases of interest, including systems that are not translation invariant. 3. In the third step we analyze expansions of logarithms of Gaussian Berezin integrals, which combined with constructive methods of quantum field theory is useful to show the analyticity of J(s) for s in a neighborhood of 0. We finally discuss how to combine steps 2-3 in order to prove (mathematically speaking) for interacting fermions in equilibrium the experimental results above mentioned. In fact, the found results in this Thesis generalize previous works in the scope of LDP used to study quantum systems.

Cette thèse se consacre à l'étude du modèle de dénaturation de l'ADN introduit par Poland et Scheraga dans les années soixante. Les modèles de dépiégeage en milieu aléatoire, avec lesquels la correspondance a été établie, sont également traités. Dans le cas où les interactions entre le système et l’environnement sont homogènes, le problème a été résolu : selon la valeur d'un paramètre géométrique, une transition de phase d'ordre un ou deux se produit. En revanche, lorsque les interactions sont prises aléatoires (on parle d'un système en présence de désordre), nous ne connaissons ni le point critique, ni l'ordre de la transition en régime defort désordre. Pour simplifier le problème, de nombreux auteurs font usage d'une représentation hiérarchique grâce à laquelle une renormalisation exacte de la fonction de partition peut être écrite. Mais à nouveau, la question du point critique et de l'ordre de la transition n'a pas été résolue. Nous avons introduit un nouveau système (Toymodel) plus simple que la version hiérarchique en changeant la forme de la renormalisation. Le problème, ainsi posé, permet de mettre en évidence une famille de distributions qui ne varient presque pas lors d'une renormalisation, avec lesquelles nous avons pu dériver des équations du type Berezinskii-Kosterlitz- Thouless. Aussi, en présence de désordre, la transition de phase n'admet pas de point fixe critique. Ces deux éléments, en accord avec nos résultats numériques, nous poussent à croire que nous sommes en présence d'une transition de phase d'ordre infini. La seconde partie de la thèse rapporte un travail sur le processus simple d'exclusion symétrique, qui est l'un des modèles les plus simples de physique statistique hors d'équilibre pour lequel un état stationnaire est connu. La fonction de grandes déviations a été calculée dans le passé par les approches microscopiques et macroscopiques et ici, nous en avons calculé la première correction de taille finie. Le résultat a ensuite été comparé aux corrections similaires pour des systèmes à l'équilibre. / This thesis is a study of a DNA denaturation model, introduced by Poland and Scheraga during the 1960s. The depinning models with random environment, with which the similarity has been made, are also concerned. If the interactions between the system and the environment are homogeneous, the problem has been solved: depending on the value of a geometrical parameter, a first or a second order phase transition happens. On the other hand, when the interactions are random, we know neither the critical point nor the phase transition order in the case of strong disorder. In order to simplify the problem, some authors have used a hierarchical representation through which an exact renormalization can be written. Despite this simplification, the critical point and the transition order have not been found. By changing the renormalization relation, we introduced a Toy-model which is simpler than the hierarchical version. The new problem leaded us to a family of distributions, which stay almost the same under renormalization, and allow us to derive the Berezinskii-Kosterlitz- Thouless equations. Also, with strong disorder, the phase transition does not have a critical fixed point. These two elements, according to our numerical results, predict that the order transition is infinite. The second part of this thesis reports on a work about the simple symmetric exclusion process, which is one of the simplest out of equilibrium models for which a stationary state is known. The large deviation function has been calculated in the past through microscopic and macroscopic approaches. Here, we calculated the leading finite-size correction. Then the result has been compared to similar corrections for equilibrium systems.

Après une introduction générale et la présentation d'un exemple explicite dans le chapitre 1, nous exposons certains outils et techniques généraux dans le chapitre 2.- dans le chapitre 3, nous démontrons l'existence d'un principe de grandes déviations (PGD) pour les composantes de Cartan le long des marches aléatoires sur les groupes linéaires semi -simples G. L'hypothèse principale porte sur le support S de la mesure de la probabilité en question et demande que S engendre un semi-groupe Zariski dense. - Dans le chapitre 4, nous introduisons un objet limite (une partie de la chambre de Weyl) que l'on associe à une partie bornée S de G et que nous appelons le spectre joint J(S) de S. Nous étudions ses propriétés et démontrons que J(S) est une partie convexe compacte d'intérieur non-vide dès que S engendre un semi -groupe Zariski dense. Nous relions le spectre joint avec la notion classique du rayon spectral joint et la fonction de taux du PGD pour les marches aléatoires. - Dans le chapitre 5, nous introduisons une fonction de comptage exponentiel pour un S fini dans G, nous étudions ses propriétés que nous relions avec J(S) et démontrons un théorème de croissance exponentielle dense. - Dans le chapitre 6, nous démontrons le PGD pour les composantes d'Iwasawa le long des marches aléatoires sur G. L'hypothèse principale demande l'absolue continuité de la mesure de probabilité par rapport à la mesure de Haar.- Dans le chapitre 7, nous développons des outils pour aborder une question de Breuillard sur la rigidité du rayon spectral d'une marche aléatoire sur le groupe libre. Nous y démontrons un résultat de rigidité géométrique. / After giving a detailed introduction andthe presentation of an explicit example to illustrateour study in Chapter 1, we exhibit some general toolsand techniques in Chapter 2. Subsequently,- In Chapter 3, we prove the existence of a large deviationprinciple (LDP) with a convex rate function, forthe Cartan components of the random walks on linearsemisimple groups G. The main hypothesis is onthe support S of the probability measure in question,and asks S to generate a Zariski dense semigroup.- In Chapter 4, we introduce a limit object (a subsetof the Weyl chamber) that we associate to a boundedsubset S of G. We call this the joint spectrum J(S)of S. We study its properties and show that for asubset S generating a Zariski dense semigroup, J(S)is convex body, i.e. a convex compact subset of nonemptyinterior. We relate the joint spectrum withthe classical notion of joint spectral radius and therate function of LDP for random walks on G.- In Chapter 5, we introduce an exponential countingfunction for a nite S in G. We study its properties,relate it to joint spectrum of S and prove a denseexponential growth theorem.- In Chapter 6, we prove the existence of an LDPfor Iwasawa components of random walks on G. Thehypothesis asks for a condition of absolute continuityof the probability measure with respect to the Haarmeasure.- In Chapter 7, we develop some tools to tackle aquestion of Breuillard on the rigidity of spectral radiusof a random walk on a free group. We prove aweaker geometric rigidity result.