L'objet de cette thèse est l'étude de l'auto-organisation dans les automates cellulaires unidimensionnels.Les automates cellulaires sont un système dynamique discret ainsi qu'un modèle de calcul massivement parallèle, ces deux aspects s'influençant mutuellement. L'auto-organisation est un phénomène où un comportement organisé est observé asymptotiquement, indépendamment de la configuration initiale. Typiquement, nous considérons que le point initial est tiré aléatoirement: étant donnée une mesure de probabilité décrivant une distribution de configurations initiales, nous étudions son évolution sous l'action de l'automate, le comportement asymptotique étant décrit par la(les) mesure(s) limite(s).Notre étude présente deux aspects. D'abord, nous caractérisons les mesures qui peuvent être atteintes à la limite par les automates cellulaires; ceci correspond aux différents comportements asymptotiques pouvant apparaître en simulation. Cette approche rejoint divers résultats récents caractérisant des paramètres de systèmes dynamiques par des conditions de calculabilité, utilisant des outils d'analyse calculable. Il s'agit également d'une description de la puissance de calcul des automates cellulaires sur les mesures.Ensuite, nous proposons des outils pour létude de l'auto-organisation dans des classes restreintes. Nous introduisons un cadre d'étude d'automates pouvant être vus comme un ensemble de particules en interaction, afin d'en déduire des propriétés sur leur comportement asymptotique. Une dernière direction de recherche concerne les automates convergeant vers la mesure uniforme sur une large classe de mesures initiales (phénomène de randomisation). / The subject of this thesis is the study of self-organization in one-dimensional cellular automata.Cellular automata are a discrete dynamical system as well as a massively parallel model of computation, both theseaspects influencing each other. Self-organisation is a phenomenon where an organised behaviour is observed asymptotically, regardless of the initial configuration. Typically, we consider that the initial point is sampled at random; that is, we consider a probability measure describing the distribution of theinitial configurations, and we study its evolution under the action of the automaton, the asymptoticbehaviour being described by the limit measure(s).Our work is two-sided. On the one hand, we characterise measures that can bereached as limit measures by cellular automata; this corresponds to the possible kinds of asymptoticbehaviours that can arise in simulations. This approach is similar to several recent results characterising someparameters of dynamical systems by computability conditions, using tools from computable analysis. Thisresult is also a description of the measure-theoretical computational power of cellular automata.On the other hand, we provided tools for the practical study of self-organization in restricted classes of cellularautomata. We introduced a frameworkfor cellular automata that can be seen as a set of interacting particles, in order todeduce properties concerning their asymptotic behaviour. Another ongoing research direction focus on cellular automata that converge to the uniform measurefor a wide class of initial measures (randomization phenomenon).

[pt] Provamos que campos de vetores C1- genéricos em uma variedade compacta não possuem probabilidades invariantes absolutamente contínuas em relação a uma medida de volume. Este trabalho estende ao caso de tempo contínuo um resultado de Avila e Bochi. / [en] We prove that C1-generic vector fields in a compact manifold do not have absolutely continuous invariant probabilities. This extends a result of Avila and Bochi to the continuous time case.

[pt] Os expoentes de Lyapunov são uma ferramenta bastante utilizada quando busca-se entender o comportamento de sistemas dinâmicos, em particular de cociclos lineares. De fato, concentramo-nos no expoente maximal, pois este determina o comportamento geral do sistema, de modo que sua positividade pode ser um indicativo de que estamos lidando com um sistema caótico. Nesse sentido estudamos um teorema provado por Michael Herman, que fornece uma cota inferior para o expoente de Lyapunov maximal de uma classe de cociclos lineares definidos por rotações no círculo. A prova deste resultado utiliza um processo de complexificação do cociclo e um argumento de subharmonicidade. Surpreendentemente, essa cota inferior é na verdade uma identidade, o que foi provado posteriormente por Avila e Bochi. Como será mostrado nesta dissertação, o argumento para obter a identidade depende crucialmente da harmonicidade, e não da mera subharmonicidade de certas funções associadas às iterações do cociclo. / [en] Lyapunov exponents are a widely used tool when trying to understand the behavior of dynamical systems in general, and in particular that of linear cocycles. We focus on the maximal exponent, as it determines the general behavior of the system, in that its positivity can be an indication that we are dealing with a chaotic system. In this sense, we study a theorem obtained by Michael Herman, providing a lower bound on the maximal Lyapunov exponent of a class of linear cocycles defined by circle rotations. The proof of this result employs the complexification of the cocycle and an argument based on subharmonicity. Surprisingly, this lower bound is in fact an identity, which was proven later by Avila and Bochi. As it will be shown in this dissertation, the argument for obtaining this identity depends crucially on the harmonicity, as opposed to the mere subharmonicity of certain functions associated with the iterates of the cocycle.

The demand for automated decision making, learning and inference in uncertain, risk sensitive and dynamically changing situations presents a challenge: to design computational approaches that promise to be widely deployable and flexible to adapt on the one hand, while offering reliable guarantees on safety on the other. The tension between these desiderata has created a gap that, in spite of intensive research and contributions made from a wide range of communities, remains to be filled. This represents an intriguing challenge that provided motivation for much of the work presented in this thesis. With these desiderata in mind, this thesis makes a number of contributions towards the development of algorithms for automated decision-making and inference under uncertainty. To facilitate inference over unobserved effects of actions, we develop machine learning approaches that are suitable for the construction of models over dynamical laws that provide uncertainty bounds around their predictions. As an example application for conservative decision-making, we apply our learning and inference methods to control in uncertain dynamical systems. Owing to the uncertainty bounds, we can derive performance guarantees of the resulting learning-based controllers. Furthermore, our simulations demonstrate that the resulting decision-making algorithms are effective in learning and controlling under uncertain dynamics and can outperform alternative methods. Another set of contributions is made in multi-agent decision-making which we cast in the general framework of optimisation with interaction constraints. The constraints necessitate coordination, for which we develop several methods. As a particularly challenging application domain, our exposition focusses on collision avoidance. Here we consider coordination both in discrete-time and continuous-time dynamical systems. In the continuous-time case, inference is required to ensure that decisions are made that avoid collisions with adjustably high certainty even when computation is inevitably finite. In both discrete-time and finite-time settings, we introduce conservative decision-making. That is, even with finite computation, a coordination outcome is guaranteed to satisfy collision-avoidance constraints with adjustably high confidence relative to the current uncertain model. Our methods are illustrated in simulations in the context of collision avoidance in graphs, multi-commodity flow problems, distributed stochastic model-predictive control, as well as in collision-prediction and avoidance in stochastic differential systems. Finally, we provide an example of how to combine some of our different methods into a multi-agent predictive controller that coordinates learning agents with uncertain beliefs over their dynamics. Utilising the guarantees established for our learning algorithms, the resulting mechanism can provide collision avoidance guarantees relative to the a posteriori epistemic beliefs over the agents' dynamics.

The thermally-driven rotating annulus is a laboratory experiment used to study the dynamics of planetary atmospheres under controlled and reproducible conditions. The predictability of this experiment is studied by applying the same principles used to predict the atmosphere. A forecasting system for the annulus is built using the analysis correction method for data assimilation and the breeding method for ensemble generation. The results show that a range of flow regimes with varying complexity can be accurately assimilated, predicted, and studied in this experiment. This framework is also intended to demonstrate a proof-of-concept: that the annulus could be used as a testbed for meteorological techniques under laboratory conditions. First, a regime diagram is created using numerical simulations in order to select points in parameter space to forecast, and a new chaotic flow regime is discovered within it. The two components of the framework are then used as standalone algorithms to measure predictability in the perfect model scenario and to demonstrate data assimilation. With a perfect model, regular flow regimes are found to be predictable until the end of the forecasts, and chaotic regimes are predictable over hundreds of seconds. There is a difference in the way predictability is lost between low-order chaotic regimes and high-order chaos. Analysis correction is shown to be accurate in both regular and chaotic regimes, with residual velocity errors about 3-8 times the observational error. Specific assimilation scenarios studied include information propagation from data-rich to data-poor areas, assimilation of vortex shedding observations, and assimilation over regime and rotation rate transitions. The full framework is used to predict regular and chaotic flow, verifying the forecasts against laboratory data. The steady wave forecasts perform well, and are predictable until the end of the available data. The amplitude and structural vacillation forecasts lose quality and skill by a combination of wave drift and wavenumber transition. Amplitude vacillation is predictable up to several hundred seconds ahead, and structural vacillation is predictable for a few hundred seconds.

This thesis contributes to two completely unrelated debates in the economic literature, similar only in the relatively high degree of controversy characterizing each one. <p>The first part is methodological and macroeconomic in nature, addressing the question of whether the distribution of income across countries is converging (i.e. are the poor catching up to the rich?) or diverging (i.e. are we witnessing the formation of two exclusive clubs, one for poor countries and another one for rich countries?). Applications of the simple Markov model to this question have generated evidence in favor of the divergence hypothesis. In the first chapter, I critically review these results. I use statistical inference to show that the divergence results are not statistically robust, and I explain that this instability of the results comes from the application of a model for discrete data to data that is actually continuous. In the second chapter, I reposition the whole convergence-divergence debate by placing it in the context of Silverman’s classic survey of non-parametric density estimation techniques. This allows me to use the basic notions of fuzzy logic to adapt the simple Markov chain model to continuous data. When I apply the newly adapted Markov chain model to the cross-country distribution question, I find evidence against the divergence hypothesis, and this evidence is statistically robust. <p>The second part of the thesis is empirical and microeconomic in nature. I question whether observed differences between husbands’ and wives’ participation in labor markets are due to different preferences or to different constraints. My identification strategy is based on the idea that the more power an individual has relative to his/her partner, the more his/her actions will reflect his/her preferences. I use 2001 PSID data on cohabiting couples to estimate a simultaneous equations model of the spousal time allocation decision. My results confirm the stylized fact that specialization and trade does not explain time allocation for couples in which the wife is the primary breadwinner, and suggest that power could provide a more general explanation of the observations. My results show that wives with relatively more power choose to work more on the labor market and less at home, whereas husbands with more power choose to do the opposite. Since women start out from a lower level of labor market participation than men do, it would seem that spouses’ agree that the ideal mix of market work and housework lies somewhere between the husbands’ and the wives’ current positions. / Doctorat en sciences économiques, Orientation économie / info:eu-repo/semantics/nonPublished

We investigate the biological constraints determined by the complex relationships between ecological and immunological processes of host-pathogen interactions, with emphasis on influenza viruses in human, which are responsible for a number of pandemics in the last 150 years. We begin by discussing prolegomenous reviews of historical perspectives on the use of theoretical modelling as a complementary tool in public health and epidemiology, current biological background motivating the objective of the thesis, and derivations of mathematical models of multi-locus-allele systems for infectious diseases with co-circulating serotypes. We provide detailed analysis of the multi-locus-allele model and its age-specific extension. In particular, we establish the necessary conditions for the local asymptotic stability of the steady states and the existence of oscillatory behaviours. For the age-structured model, results on the existence of a mild solution and stability conditions are presented. Numerical studies of various strain spaces show that the dynamic features are preserved. Specifically, we demonstrate that discrete antigenic forms of pathogens can exhibit three distinct dynamic features, where antigenic variants (i) fully self-organize and co-exist with no strain structure (NSS), (ii) sort themselves into discrete strain structure (DSS) with non-overlapping or minimally overlapping clusters under the principle of competitive exclusion, or (iii) exhibit cyclical strain structure (CSS) where dominant antigenic types are cyclically replaced with sharp epidemics dominated by (1) a single strain dominance with irregular emergence and re-emergence of certain pathogenic forms, (2) ordered alternating appearance of a single antigenic type in periodic or quasi-periodic form similar to periodic travelling waves, (3) erratic appearance and disappearance of synchrony between discrete antigenic types, and (4) phase-synchronization with uncorrelated amplitudes. These analyses allow us to gain insight into the age-specific immunological profile in order to untangle the effects of strain structures as captured by the clustering behaviours, and to provide public health implications. The age-structured model can be used to investigate the effect of age-specific targeting for public health purposes.

[pt] Uma medida de probabilidade com suporte compacto em um grupo de matrizes determina uma sequência de matrizes aleatórias i.i.d. Considere o processo multiplicativo correspondente e suas médias geométricas. O teorema de Furstenberg-Kesten, análogo da lei dos grandes números neste cenário, garante que as médias geométricas desse processo multiplicativo convergem quase certamente para uma constante, chamada de expoente de Lyapunov maximal da medida dada. Este conceito pode ser reformulado no contexto mais geral da teoria ergódica usando cociclos lineares aleatórios sobre o shift de Bernoulli. Uma questão natural diz respeito às propriedades de regularidade do expoente de Lyapunov como uma função dos seus dados. Sob uma condição de irredutibilidade e em um cenário específico (que foi posteriormente generalizado por vários autores) Le Page estabeleceu a continuidade de Holder do expoente de Lyapunov. Recentemente, Baraviera e Duarte obtiveram uma prova direta e elegante deste tipo de resultado. Seu argumento usa a fórmula de Furstenberg e as propriedades de regularidade da medida estacionária. Seguindo sua abordagem, neste trabalho obtemos um novo resultado mostrando que, sob a mesma hipótese de irredutibilidade, o expoente de Lyapunov depende Hölder continuamente da medida, relativamente à métrica de Wasserstein, generalizando assim o resultado de Baraviera e Duarte. / [en] A compactly supported probability measure on a group of matrices determines a sequence of i.i.d. random matrices. Consider the corresponding multiplicative process and its geometric averages. Furstenberg-Kesten s theorem, the analogue of the law of large numbers in this setting, ensures that the geometric averages of this multiplicative process converge almost surely to a constant, called the maximal Lyapunov exponent of the given measure. This concept can be reformulated in the more general context of ergodic theory using random linear cocycles over the Bernoulli shift. A natural question concerns the regularity properties of the Lyapunov exponent as a function of the data. Under an irreducibility condition and in a specific setting (which was later generalized by various authors) Le Page established the Holder continuity of the Lyapunov exponent. Recently, Baraviera and Duarte obtained a direct and elegant proof of this type of result. Their argument uses Furstenberg s formula and the regularity properties of the stationary measure. Following their approach, in this work we obtain a new result showing that under the same irreducibility hypothesis, the Lyapunov exponent depends Holder continuously on the measure, relative to the Wasserstein metric, thus generalizing the result of Baraviera and Duarte.

Diese Dissertation untersucht Aspekte des Zusammenspiels von ergodischem Langzeitver- halten und der Glättungseigenschaft dynamischer Systeme, die von stochastischen Differen- tialgleichungen (SDEs) mit Sprüngen erzeugt sind. Im Speziellen werden SDEs getrieben von Lévy-Prozessen und der Marcusschen kanonischen Gleichung untersucht. Ein vari- ationeller Ansatz für den Malliavin-Kalkül liefert eine partielle Integration, sodass eine Variation im Raum in eine Variation im Wahrscheinlichkeitsmaß überführt werden kann. Damit lässt sich die starke Feller-Eigenschaft und die Existenz glatter Dichten der zuge- hörigen Markov-Halbgruppe aus einer nichtstandard Elliptizitätsbedingung an eine Kom- bination aus Gaußscher und Sprung-Kovarianz ableiten. Resultate für Sprungdiffusionen auf Untermannigfaltigkeiten werden aus dem umgebenden Euklidischen Raum hergeleitet. Diese Resultate werden dann auf zufällige dynamische Systeme angewandt, die von lin- earen stochastischen Differentialgleichungen erzeugt sind. Ruelles Integrierbarkeitsbedin- gung entspricht einer Integrierbarkeitsbedingung an das Lévy-Maß und gewährleistet die Gültigkeit von Oseledets multiplikativem Ergodentheorem. Damit folgt die Existenz eines Lyapunov-Spektrums. Schließlich wird der top Lyapunov-Exponent über eine Formel der Art von Furstenberg–Khasminsikii als ein ergodisches Mittel der infinitesimalen Wachs- tumsrate über die Einheitssphäre dargestellt. / The present thesis investigates certain aspects of the interplay between the ergodic long time behavior and the smoothing property of dynamical systems generated by stochastic differential equations (SDEs) with jumps, in particular SDEs driven by Lévy processes and the Marcus’ canonical equation. A variational approach to the Malliavin calculus generates an integration-by-parts formula that allows to transfer spatial variation to variation in the probability measure. The strong Feller property of the associated Markov semigroup and the existence of smooth transition densities are deduced from a non-standard ellipticity condition on a combination of the Gaussian and a jump covariance. Similar results on submanifolds are inferred from the ambient Euclidean space. These results are then applied to random dynamical systems generated by linear stochas- tic differential equations. Ruelle’s integrability condition translates into an integrability condition for the Lévy measure and ensures the validity of the multiplicative ergodic theo- rem (MET) of Oseledets. Hence the exponential growth rate is governed by the Lyapunov spectrum. Finally the top Lyapunov exponent is represented by a formula of Furstenberg– Khasminskii–type as an ergodic average of the infinitesimal growth rate over the unit sphere.