Nesta tese, apresenta-se a análise numérica e experimental do comportamento dinâmico não linear de um pórtico excitado por uma fonte não ideal - um motor elétrico de corrente contínua desbalanceado - como exemplo dessa classe de problemas. Elabora-se um modelo matemático com quatro graus de liberdade: dois relacionados com o deslocamento horizontal e vertical do ponto de apoio da máquina e dois com os parâmetros de funcionamento do motor elétrico. Adota-se a formulação Lagrangeana para gerar as equações de movimento contendo termos não lineares até ordem cúbica. A solução numérica é obtida através do método de Runge-Kutta com passo adaptativo. Ensaios dinâmicos e estáticos foram realizados com o motor e com o sistema completo, com vistas a validar o modelo matemático. Um sistema de medição digital armazenou todos os dados obtidos em arquivos de fácil leitura pelos programas atualmente disponíveis. Filmagens de alta velocidade e fotografia estroboscópica registraram movimentos típicos relacionados com comportamentos não lineares. Os resultados numéricos e experimentais mostram boa correlação entre si, além de apresentarem alguns dos fenômenos associados ao se forçar uma estrutura a passar por uma de suas ressonâncias excitando-a com um dispositivo com potência limitada, como o efeito Sommerfeld. Outros fenômenos, devido ao comportamento geometricamente não linear da estrutura, são também detectados, tais como saturação modal e transferência de energia. / Numerical and experimental analysis of the nonlinear dynamics of a portal frame excited by a non-ideal source - an unbalanced direct current motor - is presented in this thesis as an example of this class of problems. A four degree of freedom model is elaborated: two of them related to the horizontal and vertical structural displacements and two others to the functioning parameters of the motor. A Lagrangian approach for deducing the equations of motion, up to cubic non-linear terms is followed. The numerical solution is obtained through Runge-Kutta algorithm with adaptive step. Static and dynamic tests were performed with the motor and with the complete system, in order to validate the mathematical model. A digital acquisition system recorded all data in computer files, ready to be read by available commercial programs. High speed filming and stroboscopic photography were used to register typical movements related to nonlinear behavior. Numerical and mathematical results show good correlation, as well as present some phenomena related to passage through resonance with limited power excitation such as the Sommerfeld effect. Some other phenomena, related to the nonlinear behavior of the structure are also detected, as modal saturation and energy transference.

Das Studium des derzeitigen Klimawandels ist von hoher Relevanz da viele Menschen dessen direkten Folgen ausgesetzt sind. Das Verständnis der Vergangenheit ist der Schlüssel zu Fragen zum Klima als auch zur Vegetation, dem Leben, der Evolution und der Natur allgemein. Diese Dissertation beschäftigt sich damit Übergange zwischen dynamischen Regimen in nichtlinearen Systemen zu detektieren. Den Anfang macht eine kurze Einführung in die Methodik mit der Einführung der Rekurrenzplots (RP) als wichtigem Werkzeug für die folgende Analyse. Die genaue Gestalt eines RP hängt dabei von einem freien Parameter ab, dem Distanzschwellwert. Hier schlage ich eine neue Variante vor, den gewichteten Rekurrenzplot (wRP), welcher ohne Abhängigkeit von diesem Parameter auskommt. Darüberhinaus schlage ich eine neue Möglichkeit vor den Schwellwert für Rekurrenznetzwerke (RN) für eine gegebene Zeitreihe auszuwählen. Einen optimalen Schwellwert auszuwählen ist ausschlaggebend für das Ergebnis der Zeitreihenanalyse. In den folgenden theoretischen Ausführungen beschreibe ich daher eine neuartige Vorbehandlung von heterogenen (insbesondere irregulär gesampelten) Zeitreihen. Diese treten oft in Proxyda- ten, etwa Speläothemen, auf und lassen sich nicht direkt mit RP analysieren. Die vorgestellten Ansätze bieten geeignete Methoden um dynamische Übergänge zu untersuchen. Sie können unterschiedlichste Anwendungen in einer Vielzahl von Zeitreihenanalysen finden. Beispielsweise wird generell die Untersuchung irregulär gesampelter Datensätze ermöglicht. Letztlich, da die Zusammenwirkung der Proxydaten aus Nordwestaustralien und Südchina besser verstanden ist, stellt sich die Aufgabe solche Analysen mit Hilfe der hier vorgestellten Methoden auf Paläoklimadaten der gesamten Erde auszuweiten, um ein umfassendes Netzwerk der paarweisen Wechselbeziehungen zu erstellen, das für ein besseres Verständnis des Erdsystems hilfreich sein kann. / The study of climate change is a very important field of science, since life is directly affected by these changes. Investigating the past climate changes leads to the forecasts of future possibilities. Therefore, enlightening the past is a keystone to answer questions about climate as well as vegetation, life, evolution, and nature. This Thesis focuses on detecting dynamical regime transitions in nonlinear dynamical systems as well as in climate proxies where they mark previous critical climate changes. The thesis begins with a brief methodological background and brings the recurrence plot (RP) to forefront as the main tool for the further analyses. The thesis consists of three main studies: (i) The formation of RPs naturally depends on a free parameter in the analysis given by the distance threshold. I propose an alternative definition by using a weighted variant of the RP, called weighted recurrence plot, which removes dependence on this free parameter. (ii) Furthermore, I suggest a novel way to select the threshold for a recurrence network for a specific time series. Selecting the optimization parameters for a specific time series is very important for the performance of the analysis. (iii) Next, I introduce a new preprocessing technique to deal with the heterogeneousness of time series, since the RP is not directly applicable on such data sets and the proxies from speleothems, in general, are irregularly sampled. Among these presented approaches are suitable methods to investigate the dynamical transition and they can be used for different purposes in a large variety of time series analyses. These techniques can be used in several different disciplines which have heterogeneousness in their sources. After the relationship between the two proxies from Australia and China has been uncovered, in the future, it should be achievable to extend the study to create a large paleoclimate relationship network for the entire Earth by using the methods given in this Thesis.

O trabalho se baseia em uma formulação numérica (Método dos Elementos Finitos Posicional) que é combinada a um modelo viscoelástico adequado (Kelvin-Voigt adaptado), o que direciona para o cumprimento do objetivo: a simulação de sólidos viscoelásticos em deformações finitas e de fluidos viscosos. A formulação desenvolvida é Lagrangeana total descrita para posições, permitindo aplicações em não linearidade dinâmica (com a utilização do método de Newton-Raphson para solução do sistema de equações não lineares e integração temporal via algoritmo implícito de Newmark) e sua combinação com um modelo viscoelástico coerente é deduzida neste trabalho. Inicialmente, são resolvidos problemas com elemento de chapa bidimensional, porém o elemento finito final utilizado é de sólido prismático de base triangular. Dois modelos são adotados para consideração do comportamento viscoelástico, (i) um modelo modificado de Kelvin-Voigt associado ao modelo constitutivo de Saint-Venant-Kirchhoff e (ii) um modelo visco-hiperelástico completo coerente para deformações finitas desenvolvido a partir da decomposição multiplicativa sobre o gradiente da função mudança de configuração em uma parcela volumétrica e duas isocóricas. Foram selecionados e comentados 15 exemplos em detalhe, abrangendo todas as etapas desta pesquisa, com problemas elásticos, dinâmicos, viscoelásticos em pequenas e grandes deformações, de flexão, de impacto e de fluidos viscosos. Os resultados obtidos para os exemplos de validação foram satisfatórios, coerentes com as referências, e o conjunto das análises conduzidas mostram a potencialidade da formulação alternativa desenvolvida neste trabalho. / The work is based on a numerical formulation (Positional Finite Element Method) combined with a suitable viscoelastic model (adapted Kelvin-Voigt), what directs to achieving its main goal: the simulation of viscoelastic solids in finite strain and of viscous fluids. The developed formulation is total Lagrangian described for positions, allowing applications in nonlinear dynamics (using the Newton-Raphson method for solution of the system of nonlinear equations, and performing time integration via an implicit Newmark algorithm); its combination with adequate viscoelastic model is shown step-by-step in this work. Initially, problems are solved using two-dimensional plate element, but the final finite element is a triangular-based prismatic solid. Two models are adopted in order to consider the viscoelastic behavior, (i) a modified Kelvin-Voigt model associated with the Saint-Venant-Kirchhoff constitutive model, and (ii) a coherent visco-hyperelastic model for finite deformations developed from the multiplicative decomposition over the deformation gradient in one volumetric and two isochoric parts. 15 examples were selected and commented in detail, comprehending all stages of this research, solving problems that are elastic, dynamic, viscoelastic under small and large strain, under flexural behavior, submitted to impact, and of viscous fluids problems. The results obtained for the validation examples were satisfactory, consistent with the references, and the whole of the conducted analysis shows the potentials of the alternative formulation developed in this work.

Dynamische Effekte in Halbleiterlasern aufgrund aktiver optischer Rückkopplung werden untersucht. Basierend auf Laserstrukturen mit verteilter Rückkopplung wird ein neuer Lasertyp entwickelt. Durch Experimente und ergänzende Simulationen wird ein umfassender Überblick der Dynamik dieses neuen Lasertyps gegeben. Zuerst werden die grundsätzlichen Veränderungen der Laserdynamik durch verzögerte optische Rückkopplung diskutiert. Es zeigen sich Hysterese- und Pulsationsphänomene. Ursprung der Pulsationen sind entdämpfte Relaxationsoszillationen und Modenschwebungen. Die Pulsationsphänomene werden anhand von integrierten Mehrsektionslasern mit passiver und aktiver optischer Rückkopplung untersucht. Es zeigt sich, daß nur Laser mit aktiver Rückkopplung die vollständige Kontrolle von Rückkoppelphase und -stärke erlauben, und damit den Zugang zum gesamten Spektrum der Laserdynamik mit sehr kurzer Verzögerungszeit bieten. Anhand des Lasertyps mit aktiver Rückkopplung wird eine umfassende Bifurkationsanalyse durchgeführt. Die Ergebnisse einer numerischen Untersuchung werden durch umfangreiche Experimente verifiziert. Sattel-Knoten-, Hopf- und Torus-Bifurkationen organisieren das Bauelementverhalten. Die Koexistenz von entdämpften Relaxationsoszillationen und Schwebungspulsationen erlaubt die Untersuchung von internen Synchronisationsphänomenen, Resonanzen und chaotischer Dynamik. Der Einfluß spontaner Emission auf die Laserdynamik zeigt sich am Beispiel rauschinduzierter Vorläufer einzelner Bifurkationen sowie am Übergang zwischen den beiden Pulsationsphänomenen. Abschließend wird die Anwendung des Lasers mit integriert-aktiver Rückkopplung als optischer Taktregenerator bei einer Datenrate von 40 Gbit/s demonstriert. Arbeitspunkte mit koexistierenden Oszillatoren und ganzzahligem Frequenzverhältnis eignen sich darüber hinaus zur rein optischen Frequenzteilung im GHz-Bereich. / Dynamical effects in semiconductor lasers due to active optical feedback are investigated. A novel laser type based on distributed feedback structures is developed and realized. Experiments as well as simulations give a comprehensive overview on the nonlinear dynamics of this laser type. First, the fundamental modifications of the solitary laser dynamics due to delayed optical feedback in the very short feedback cavity regime are discussed. Hysteresis effects and pulsation phenomena due to undamped relaxation oscillations and mode beating are identified. These oscillation types are experimentally confirmed by multi-section lasers with integrated passive and active feedback. It turns out that only the active feedback laser (AFL) allows for the full control of feedback phase and strength, enabling the access to the whole spectrum of laser dynamics in the very short delay limit. A complete bifurcation analysis for the AFL is presented. Results obtained in a numerical path-following study are verified by extensive experiments. Saddle-node, Hopf and torus bifurcations are identified to organize the device behavior. The coexistence of undamped relaxation oscillations and mode beating in the strong feedback regime allows for the experimental investigation of resonances, mutual locking and chaotic behavior. The modification of the laser dynamics by spontaneous emission noise is shown with the example of noisy precursors and irregular dynamics at the transition between the two oscillation types. Finally, the applicability of the AFL for optical clock recovery is demonstrated at a data rate of 40 Gbit/s. Operation points with coexisting oscillations and rational frequencies are shown to be suited for all-optical frequency division in the GHz range.

In dieser Arbeit werden zwei nichtlineare Phänomene untersucht, Multistabilität durch verzögerte Rückkopplung und Synchronisation von quasiperiodischen Oszillationen. Dies geschieht mit Hilfe von Halbleiterlasern und auf dem selben Chip wie der Laser integrierter ultrakurzer optischer Rückkopplung. Verzögerte Rückkopplung ist unter anderem die Ursache für das Phänomen der Faltung von Lasermoden, und damit für das Auftreten von mehreren möglichen Laserzuständen für die selben Parameter. Ein tristabiles Regime von Dauerstrichzuständen kann im Experiment für mehrere breite Parameterbereiche der Rückkopplung beobachtet werden. Sehr nahe der Laserschwelle wird einer der Laserzustände durch den stabilen ``aus''''-Zustand ersetzt. Theoretische Betrachtungen im Rahmen des paradigmatischen Lang-Kobayashi Models verzögerter Rückkopplung ermöglichen eine in sich konsistente Interpretation der experimentellen Ergebnisse. Neben der Beeinflussung des stationären Verhaltens eines Halbleiterlasers kann verzögerte Rückkopplung Instabilitäten in der Laseremission hervorrufen. Abhängig von Rückkoppelstärke und -phase werden zwei verschiedene Intensitätspulsationen des emittierten Lichtes beobachtet. Synchronisationsprozesse solcher Pulsationen wurden von mir in einem System von zwei verschiedenen gekoppelten Multisektionslasern untersucht. Periodische Selbstpulsationen von Laser 1 werden hierfür in Laser 2 injiziert, welcher sich in einem Regime quasiperiodischer Intentensitätspulsationen mit zwei fundamentalen Frequenzen befindet. Das Experiment zeigt eine neue Art von Übergang zu synchronem Verhalten, welche kürzlich mit Hilfe von gekoppelten generischen Phasen- und van der Pol Oszillatormodellen aufgedeckt wurde. Desweiteren konnten bislang unerforschte Prozesse des Kohärenzübertrags auch zu nichtsynchronisierten Oszillationen beobachtet werden. / In this work two nonlinear phenomena are investigated, multistability due to delayed feedback and synchronization of quasiperiodic oscillations. The experimental devices are semiconductor lasers with ultra-short optical feedback, which is integrated on the same chip as the laser. Delayed feedback causes the folding of lasing modes, leading to hysteresis effects and even the coexistence of several laser states for the same parameters. A regime of tristability of continuous-wave (cw) states is found for multiple ranges of applied currents. Very close to threshold, one of the lasing states may be replaced by the stable ``off''''-state. Theoretical investigations in the framework of the paradigmatic Lang-Kobayashi model provide a consistent understanding of the experimental findings. Besides modifying the stationary behavior of a semiconductor laser, delayed feedback can cause instabilities of the laser output. Depending on strength and phase of the feedback, two types of self-sustaining pulsations of the emitted light intensity are found in our devices. Synchronization processes of such pulsations are studied in a system of two coupled multisection lasers. Periodic self-pulsations of laser 1 are injected into laser 2, which is operating in a regime with two-frequency quasiperiodic self-pulsations. The experimental system demonstrates the new type of transitions to synchrony between three frequencies which has been recently revealed using generic coupled phase and van der Pol oscillator models. Moreover, carefully determining the coherence of the noisy oscillations, so far unexplored processes of coherence transfer to nonsynchronized oscillations are revealed.

Dynamik auf Netzwerken ist ein mathematisches Feld, das in den letzten Jahrzehnten schnell gewachsen ist und Anwendungen in zahlreichen Disziplinen wie z.B. Physik, Biologie und Soziologie findet. Die Funktion vieler Netzwerke hängt von der Fähigkeit ab, die Elemente des Netzwerkes zu synchronisieren. Mit anderen Worten, die Existenz und die transversale Stabilität der synchronen Mannigfaltigkeit sind zentrale Eigenschaften. Erst seit einigen Jahren wird versucht, den verwickelten Zusammenhang zwischen der Kopplungsstruktur und den Stabilitätseigenschaften synchroner Zustände zu verstehen. Genau das ist das zentrale Thema dieser Arbeit. Zunächst präsentiere ich erste Ergebnisse zur Klassifizierung der Kanten eines gerichteten Netzwerks bezüglich ihrer Bedeutung für die Stabilität des synchronen Zustands. Folgend untersuche ich ein komplexes Verzweigungsszenario in einem gerichteten Ring von Stuart-Landau Oszillatoren und zeige, dass das Szenario persistent ist, wenn dem Netzwerk eine schwach gewichtete Kante hinzugefügt wird. Daraufhin untersuche ich synchrone Zustände in Ringen von Phasenoszillatoren die mit Zeitverzögerung gekoppelt sind. Ich bespreche die Koexistenz synchroner Lösungen und analysiere deren Stabilität und Verzweigungen. Weiter zeige ich, dass eine Zeitverschiebung genutzt werden kann, um Muster im Ring zu speichern und wiederzuerkennen. Diese Zeitverschiebung untersuche ich daraufhin für beliebige Kopplungsstrukturen. Ich zeige, dass invariante Mannigfaltigkeiten des Flusses sowie ihre Stabilität unter der Zeitverschiebung erhalten bleiben. Darüber hinaus bestimme ich die minimale Anzahl von Zeitverzögerungen, die gebraucht werden, um das System äquivalent zu beschreiben. Schließlich untersuche ich das auffällige Phänomen eines nichtstetigen Übergangs zu Synchronizität in Klassen großer Zufallsnetzwerke indem ich einen kürzlich eingeführten Zugang zur Beschreibung großer Zufallsnetzwerke auf den Fall zeitverzögerter Kopplungen verallgemeinere. / Since a couple of decades, dynamics on networks is a rapidly growing branch of mathematics with applications in various disciplines such as physics, biology or sociology. The functioning of many networks heavily relies on the ability to synchronize the network’s nodes. More precisely, the existence and the transverse stability of the synchronous manifold are essential properties. It was only in the last few years that people tried to understand the entangled relation between the coupling structure of a network, given by a (di-)graph, and the stability properties of synchronous states. This is the central theme of this dissertation. I first present results towards a classification of the links in a directed, diffusive network according to their impact on the stability of synchronization. Then I investigate a complex bifurcation scenario observed in a directed ring of Stuart-Landau oscillators. I show that under the addition of a single weak link, this scenario is persistent. Subsequently, I investigate synchronous patterns in a directed ring of phase oscillators coupled with time delay. I discuss the coexistence of multiple of synchronous solutions and investigate their stability and bifurcations. I apply these results by showing that a certain time-shift transformation can be used in order to employ the ring as a pattern recognition device. Next, I investigate the same time-shift transformation for arbitrary coupling structures in a very general setting. I show that invariant manifolds of the flow together with their stability properties are conserved under the time-shift transformation. Furthermore, I determine the minimal number of delays needed to equivalently describe the system’s dynamics. Finally, I investigate a peculiar phenomenon of non-continuous transition to synchrony observed in certain classes of large random networks, generalizing a recently introduced approach for the description of large random networks to the case of delayed couplings.

Die Atmung und das Herz-Kreislauf-System interagieren als zwei in Reihe angeordnete funktionelle Einheiten. Die gleichsinnige Kontrolle beider Systeme bildet die Grundlage homöostatischer Bedingungen im Organismus. Neurophysiologische Studien geben Hinweise auf die Existenz eines gemeinsamen neuronalen kardiorespiratorischen Netzwerkes, welches im ventrolateralen Teil der Medulla oblongata gelegen ist. Da zentrale Mechanismen der Regulation einer direkten Untersuchung nicht zugänglich sind, erweisen sich die linearen und nichtlinearen Verfahren der Zeitreihenanalyse als hilfreich, um Erkenntnisse von der Arbeitsweise des kardiorespiratorischen Kontrollsystems zu gewinnen. Grundlage der Studie bildet eine Datenbank polygraphischer Messungen (einschließlich EKG, thorakales und abdominales Respirogramm, Elektrookulogramm und Aktogramm), die an 42 gesunden Kindern, 11 Mädchen und 31 Knaben im Alter von 12 bis 15 Jahren erhoben wurde. Die Messungen erfolgten über 24 Stunden hinweg, während folgender Vigilanzstadien: ruhiger Wachzustand, REM- und nonREM-Schlaf. Die spektralen Charakteristika der Herzfrequenzvariabilität wurden berechnet, um die sympatho-vagale Einflußnahme auf den Nodus sinusoidalis kennzeichnen zu können. Die lineare Intensität der kardiorespiratorischen Beziehung wurde aus den Kohärenzspektren abgeleitet. Um nichtlineares Verhalten erfassen zu können, wurden der größte Lyapunov-Exponent und die Korrelationsdimension der Herzfrequenz, sowie die Korrelationsdimension des Atemsignals bestimmt. Die Analyse der Herzfrequenzvariabilität ergab für die Gesamtleistung die höchsten Werte innerhalb der REM-Phasen, im Wachzustand lagen diese deutlich niedriger und während des nonREM-Schlafes waren sie am kleinsten. Dieses Verhalten wurde im wesentlichen bestimmt von Änderungen der Spektralleistung im niederfrequenten Bereich. Die Komplexität der Herzfrequenz, die sich mit der Korrelationsdimension schätzen läßt, zeigte eine deutliche Abnahme im Schlaf. Dagegen erwies sich der Lyapunov-Exponent als weniger sensitiv bezüglich der Vigilanz. Die kardiorespiratorische Kohärenz ließ eine strenge Abhängigkeit vom Vigilanzstadium erkennen mit hohen Werten im nonREM-Schlaf und dem Minimum innerhalb der REM-Phasen. Im Gegensatz zur Komplexität der Herzfrequenz erreichte die Komplexität der Atmung die niedrigsten Werte in den REM-Phasen. Mit den Ergebnissen der Spektralanalyse lassen sich vigilanzstadienspezifische Einstellungen in der vegetativen Kontrolle der Herzfrequenz abgrenzen. Die nichtlinearen Verfahren offenbaren niederdimensionale deterministisch-chaotische Strukturen der Herzfrequenz. Die Zahl unabhängiger Mechanismen, die Anteil an der kardiorespiratorischen Regulation haben, ist im Wachzustand am größten. Diese Änderungen lassen das Gesamtsystem in Abhängigkeit von der Vigilanz verschiedene Arbeitspunkte einnehmen. / Breathing and blood flow interact as two, in series coupled units. To adapt heart beat and oxygen supply, a common coordination is required. Concluded from neurophysiological investigations, there is evidence for the existence of one cardiorespiratory network located in the ventrolateral part of the medulla. Since the physiological mechanisms inside the complex regulatory network are not readily accessible, linear and non-linear methods of time series analysis are a useful approach to investigate cardiorespiratory control. To study normal regulation, 42 healthy children, 11 girls and 31 boys (12-15 yr.), were investigated throughout 24 hours under different states of vigilance: wakefulness at rest, REM, and nonREM-sleep. All participants underwent polygraphic measurements, including ECG, thoracic and abdominal respirograms, electrooculogram, and actogram. To estimate the sympatho-vagal drive to the sinus node, the parameters of heart rate power spectra were calculated. The linear intensity of cardiorespiratory coupling was concluded from the coherence spectra. As to non-linear properties of heart rate, the largest Lyapunov exponents as well as the correlation dimension were determined. Similarly, the correlation dimension of the respiratory signals was evaluated. The total power of the heart rate spectrum was found to be greatest during REM, it decreased during wakefulness and was low in nonREM-sleep. These variations are mainly accounted for by low frequency power. The "complexity" of heart rate, as indicated by the correlation dimension, is diminished during sleep phases, whereas the Lyapunov exponents are less affected. The cardiorespiratory coherence is strongly modulated by vigilance with an increase during nonREM and lowest values during REM. The complexity of respiration was also affected by vigilance. A different behavior of heart rate complexity was found during REM-phases. Concluded from spectral analysis, a specific setting of autonomic heart rate regulation for each vigilance stage can be suggested. A low dimensional deterministic chaos is present in heart rate time series. More independent control loops were found to be active during wakefulness. Revealed by parameters of the non-linear dynamics, different stages of vigilance determine different operating points in the cardiorespiratory coordination.

Die Untersuchung vergangener Klimavariabilität ist ein einzigartiger Schlüssel zum Verständnis zukünftigen Verhaltens des Erdsystems unter anthropogener Einwirkung. Dies ist von besonderer Wichtigkeit, da es die einzige Realisierung des „Erdsystemexperiments“ ist, die für uns zugänglich ist. Paleoklimaarchive, wie Bäume, Stalagmiten oder Gletscher stellen in ihrer Struktur und Zusammensetzung zeitabhängige Aufzeichungen früherer Klimavariabilität dar. Die statistische Analyse von Zusammenhängen zwischen solchen Zeitreihen kann helfen, die den Paläoklimaproxies zugrundeliegenden Klimaprozesse und, letztlich, der Erdsystemdynamik zu verstehen. Drei Hauptherausforderungen müssen gemeistert werden, um dies möglich zu machen: die Zeitreihen sind unregelmäßig aufgelöst in (i) Zeit, (ii) Raum und die Zeit selbst ist eine Variable die rekonstruiert werden muss, was (iii) zusätzliche Unsicherheiten mit sich bringt. Dazu habe ich den Paläoklimanetzwerkansatz entwickelt, inspiriert von der zunehmenden Anwendung von Methoden aus dem Bereich der komplexen Netzwerke in der Klimatologie. Ich habe Schätzer für Pearson-Korrelation, Transinformation (Mutual Information) und Ereignissynchronisation (Event Synchronization) eingeführt, die keine Zeitreihen mit regelmäßigen Beobachtungsintervallen benötigen. Der Einfluß von Altersunsicherheiten auf Schätzungen solcher Ähnlichkeitsmaße wird numerisch durch Ensembles von möglichen Akkumulationsverläufen abgeschätzt. Ein einfaches Modell für Informationsflüsse im Asiatischen Sommermonsun (ASM) ermöglicht den Test der Fähigkeiten von (Paläoklima-)Netzwerkmaßen, räumlich-zeitliche Klimaänderungen von Zeitreihen räumlich heterogen verteilter Orte zu detektieren. / Investigating past climate changes offers a unique key to understanding the future behavior of the Earth system under anthropogenic perturbation, because it is the only realization of the “Earth system experiment” accessible. Paleoclimate archives such as trees, stalagmites, or glacial deposits provide in their structure and composition time-dependent records of earlier climatic variability. Statistical analysis of dependencies amongst such time series helps to infer on the climatic processes reflected in the paleoclimate proxy data and then, ultimately, on the dynamics of the Earth system. Three inherent technical challenges need to be met: the datasets are heterogeneously sampled in (i) time and (ii) space, and time itself is a variable that needs to be reconstructed which (iii) introduces additional uncertainties. To address these issues I developed the paleoclimate network framework, inspired by the increasing application of complex networks methodology in climate. I introduced estimators for Pearson correlation, mutual information and event synchronization that do not require time series sampled at regular intervals. The impacts of age uncertainty on such similarity estimates was assessed numerically, using ensembles of possible accumulation histories. A simple model for information flow in the Asian summer monsoon (ASM) was used to test the ability of (paleoclimate) network measures to detect spatio-temporal transitions from time series observed at heterogeneous locations in space.

Inicialmente, apresentam-se características de dispositivos eletrônicos conhecidos como PLLs (phase-locked loops). PLLs são amplamente empregados para se extrair sinais de tempo em canais de comunicação e em aplicações nas quais se deseja controle automático de freqüência. O objeto principal é estudar PLLs analógicos descritos por uma equação diferencial ordinária de terceira ordem. Assim, deduzem-se condições de estabilidade assintótica e identifica-se um regime de caos conservativo, que ocorre sob certas combinações de valores de parâmetros. Três métodos de controle não-linear/caótico são então apresentados e aplicados. Os métodos são os seguintes: o Método de Pyragas via realimentação de variável de estado; o Método de Pyragas com atraso temporal na realimentação; e o Método de Sinha, o qual efetua o controle perturbando um parâmetro do sistema. Simulações numéricas são levadas a cabo a fim de ilustrar o comportamento dinâmico do sistema quando sujeito à ação desses métodos. Este trabalho termina com um estudo de uma rede formada por uma cadeia de PLLs. Condições para soluções síncronas, periódicas e caóticas (dissipativas e conservativas) são deduzidas para tal rede. / Firstly, features of electronic devices known as PLLs (Phase-Locked Loops) are presented. PLLs are widely employed to extract time signals in communication channels and in applications where automatic control of frequency is desired. The main goal is to study analog PLLs described by a third-order nonlinear ordinary differential equation. Thus, conditions for asymptotic stability are derived and a regime of conservative chaos occurring under certain combinations of parameter values is identified. Then, three methods of control of nonlinear/ chaotic dynamics are presented and applied. The methods are the following: the Pyragas method via feedback of state variable; the Pyragas method with time delay in the feedback; and the Sinhas method, which performs the control by disturbing a parameter of the system. Numerical simulations are accomplished in order to illustrate the dynamical behavior of the system when subjected to the action of these methods. This work ends with a study of a single-chain PLL network. Conditions for synchronous, periodic and chaotic (dissipative and conservative) solutions are derived for such a network.

Esta tese trata o escoamento ao redor de um cilindro logo após a sua primeira instabilidade, dentro do seu regime bidimensional periódico. A abordagem é principalmente teórica, passa por experimentos e culmina em uma importante parte numérica que complementa a teoria com evidências e ilustrações. As principais contribuições são a análise sobre a composição modal da solução dentro do regime periódico e o método desenvolvido para identificar autovetores de uma linearização da equação de Navier-Stokes presentes em uma dada solução. As bases compostas pelos autovetores identificados servem para a projeção da equação de Navier-Stokes e dão a essência dos modelos reduzidos deste estudo. A aplicação numérica apresentada para Re = 60 traz duas iterações do processo, com duas bases de autovetores de dimensões 12 e 24. Os modelos reduzidos são numericamente estáveis e a sua integração apresenta custo várias ordens mais baixo que o da simulação numérica completa. As séries temporais das coordenadas e as bases de autovetores possibilitam a recomposição do escoamento e a sua comparação com a simulação numérica de referência. A análise de aderência foi baseada nas médias temporais, nos valores de Strouhal e na estrutura dos harmônicos. Ambos modelos reduzidos têm correspondência próxima com o comportamento assintótico do escoamento e a tendência convergente das iterações é clara. As simetrias espaciais e temporais dos harmônicos são facilmente identificadas na estrutura dos modelos, de forma que as bases construídas podem ser entendidas como conjuntos de estruturas coerentes do fenômeno. / This thesis concerns the flow past a cylinder just after its first bifurcation, within its two-dimensional periodic regime. The approach is mainly theoretical, goes through experiments and is concluded by an important numerical part which complements the theory with evidences and illustrations. The main contributions are the analysis concerning the modal composition of the solution within the periodic regime and a method to identify eigenvectors of some linearizaton of the Navier-Stokes equation participating on a given solution. The bases spanned by the identified eigenvectors are employed in the projection of the Navier-Stokes equation and are central to the reduced models of this study. The numerical results for Re = 60 present two iterations of the process, with two bases of dimensions 12 e 24. The reduced models are numerically stable and their integration is many orders less costly than that of the full simulation. The time series of the modal coordinates and the eigenvectors bases allow the recomposition of the flow and its comparison with the full simulation results. The convergence analysis was based on the time averages, the Strouhal number values and the harmonic structure. Both reduced models have close correspondence with the asymptotic behavior of the flow and the convergent trend of the iterations is clear. The space and time symmetries of the harmonics have a simple representation within the structure of the models, therefore the identified bases can be understood as sets of coherent structures of the phenomenon.