Materials when subjected to deformation exhibit unstable plastic flow beyond the elastic limit. In certain range of temperature and strain rates many solid state solutions, both interstitial as well as substitutional, exhibit the phenomenon of serrated yielding which also goes by the name, the Portevin - Le Chatelier (PLC) effect. The origin of this plastic instability is due to the interaction of dislocations with solute atoms. The objective of the thesis is to provide a dynamical systems approach to the study of this plastic flow instability. The thesis work discusses, within the framework of a model, the connection between microscopic dislocation mechanisms and macroscopic mechanical response of the specimen as stress drops in stress-strain curves. An extension of the model to the associated deformation bands is also considered. The emphasis is on the dynamical aspects of the instability. The methods of nonlinear dynamics like geometrical slow manifold and Poincare map formalism are applied for the first time to study the PLC effect. However, the approach and techniques transcend this particular application as the techniques are equally well applicable for many other physical systems as well, in particular, systems involving multiple time scales. The material covered should be of interest to investigators in the materials science, in particular, those, involved in the dislocation patterning and self organization of dislocations. Many theoretical models for the PLC effect exist in literature. Although the physical phenomenon is inherently dynamic, the conventional theoretical models do not involve any dynamical aspect. A dynamical model for this effect, due to Ananthakrishna, Sahoo and Valsakumar provides an explanation in terms of the dynamic interactions between different dislocation species and evolution of densities of these dislocation species. This model is known to reproduce several of the experimental results. It is within the perspective of this model and its extensions we analyze the PLC effect. The macroscopic manifestation of the PLC effect is the repeated load drops or serration in stress-strain curves (beyond the yield point). Each of the load drop is associated with the formation of a spatial dislocation band and its subsequent propagation. From the perspective of a dynamical system, the changeover from the stress-strain curve with single yield drop to repeated yield drops (the PLC effect) corresponds to a Hopf bifurcation wherein equilibrium state changes over to a periodic steady state. These repeated load drops correspond to auto oscillations of the applied stress (in the absence of any periodic driving force). In particular, as implied by the slow loading and sudden load drops, these oscillations are classified as relaxation oscillations. Relaxation oscillations are a result of disparate time scales of dynamics of the participating modes. Within the context of the model, this refers to very different time scales of evolution of densities of mobile (fast), immobile (slow) dislocations and those with a cloud of solute atoms (not too slow). The focus of attention in the thesis work is on these auto relaxation oscillations. There are several methodologies in nonlinear dynamical systems to study the oscillatory behavior of multidimensional systems with multiple time scales. An effective way is to study the reduced dynamical system in an appropriate space without sacrificing the required dynamical information. To this end, we discuss two techniques which compliment each other. 1.Slow manifold approach: This method utilizes the presence of multiple time scales dynamics. Advantage is that the information on the nature of evolution of the periodic orbit is retained. The limitation is that the transition from one stable state to another as parameter is varied cannot be dealt with. 2.Poincare maps:This approach utilizes the recurrent behavior of the period orbit. This is a convenient methodology to study the nature of stability of periodic orbits. However, in this, the information about the nature of evolution is lost. Both the above techniques provide good description in the presence of high dissipation or larger separation of time scales of the participating modes. For slow manifold analysis, this leads to exact slow manifold structure while in the case of Poincare maps, it leads to simpler, lower dimensional attractors. Specific issues that are dealt with using these approaches and others in this thesis are the following. To start with, we first provide a comprehensive overview of the dynamical behavior as envisaged by the model system in physically relevant two parameter space. The existence of relaxation oscillations bounded by back-to-back Hopf bifurcation is a good representation of the fact that the PLC effect manifests only in a window of strain rates. Within this boundary of Hopf bifurcations relaxation oscillations destabilize to give rise to new states of order, including the chaotic states. The changes in the nature of these oscillations with control parameters is projected through the bifurcation diagrams and analyzed using techniques like Floquet multipliers, Lyapunovs exponents etc. After the identification of the relevant parameter space for the monoperiodic relaxation oscillations, we focus our attention on the time scales involved in these relaxation oscillations and its connection to the time scales apparent in serrations of the stress-strain curve of the PLC effect. This characteristic feature of the PLC effect, the stick-slip nature of stress-strain curves, is believed to result from the negative strain rate dependence of the flow stress. The latter is assumed to arise from a competition of the relevant time scales involved in the phenomenon. However, in the previous works, the identification and the role of the time scales in the dynamical phenomenon is not clear. The motivation of this part of the work is to identify the time scales involved in the stress drops of the time series and their origin. Since the dynamics involves distinct time scales, in the long time limit, the evolution is controlled only by the slow modes. Hence, the adiabatic elimination or quasi-steady state approximation of the fast modes leads to an invariant manifold, the slow manifold which is useful for the analysis of time scales. The geometry of the slow manifold which is atypical with two connected pieces is shown to be at the root of the relaxation oscillations. The analysis of the slow manifold structure helps to understand the time scales of the dynamics operating in different regions of the slow manifold. The analysis also helps us to provide a proper dynamical interpretation for the negative branch of the strain rate sensitivity of the flow stress. The slow-fast dynamical nature manifests itself through multiperiodic oscillations also, in the form of mixed mode oscillations (MMOs), which are oscillations with both large amplitude excursions as well as small amplitude loops. In MMOs, the small amplitude oscillatory loops are confined to one part of the slow manifold (around the fixed point) and the large amplitude excursions arise as jumps from one piece of the slow manifold to the other. More generally, MMOs are a characteristic feature of a family of dynamical systems which also exhibit alternate periodic-chaotic sequences in bifurcation portraits. Usually, the origin of these features is explained in terms of either the approach to a homoclinic bifurcation duo to a saddle fixed point (Shilnikov scenario) or a saddle orbit (Gavrilov-Shilnikov scenario). However, the dynamical model exhibits features from both the above scenarios. The emphasis of this study is on explaining the origin of the incomplete approach to a global bifurcation in the dynamical model. Apart from attempting to understand the complex bifurcation sequences, an additional motivation for this study is the apparent lack of systematic investigation into the incomplete approach to global bifurcation exhibited by a variety of physical systems. The method of the analysis is general and applicable to the family of MMO systems. In the model, using the structure of the bifurcation sequences, and the equilibrium fixed point, a local analysis shows that the approach to homoclinicity is asymptotic at best, and is a result of the ‘softening' of eigenvalues of the saddle equilibrium point. This softening, in turn, is a consequence of back-to-back Hopf bifurcation which reflects the constraint of the physical phenomenon, namely, the occurrence of the multiple stress drops only in an interval of the strain rates. The characteristic features, namely, MMOs, alternate periodic-chaotic sequences, and incomplete approach to homoclinicity are related to each other and arise as a consequence of the atypical slow manifold structure. The slow manifold structure analysis assumes that the evolution of the system is constrained within the neighborhood of the slow manifold which also implies that the dynamical system involves high dissipation. Hence, the dimension of the effective dynamics in the long time limit is reduced. The analysis reveals information regarding the structure of the periodic orbit for a given set of parameter values but does not provide any information regarding the nature of stability of the periodic orbits. However, any insight into the mechanism of the instability of the periodic orbits in the model may lead to a better understanding of the underlying physical phenomenon. Poincare maps and equivalent discrete dynamical systems provide a convenient means to obtain such an insight on the nature of the periodic solutions of the dynamical system. This methodology compliments the invariant slow manifold analysis, since in Poincare maps, the nature of the stability information is preserved at the expense of the structure of the periodic orbit. However, these two methodologies are not exclusive to each other, since the slow manifold structure as well as Poincare maps may be constructed using a common factor, namely, extremal values of the fast variable of the dynamical system. The methodologies adopted for the analysis assumes large dissipation arising out of the multiple time scale behavior such that the next maximal amplitude (NMA) maps can be modeled by one dimensional discrete dynamical systems. The dynamical portrait of the model shows differing nature of dynamics and consequently Poincare maps with different geometrical shapes in the {m,c) plane. Within the framework of one dimensional maps, these shapes can be schematically reconstructed using minimal information regarding the principal periodic orbit embedded in higher dimension and its nature of stability. This suggests that one dimensional maps might be sufficient to represent the higher dimensional dynamical system. For most of the parameter space, the NMA maps of the dynamical model possess characteristic features of a locally smooth maximum and asymptotically long tail. These features have been observed in many other physical systems, both experimental and model systems. Hence, this analysis is focused on a broader issue of Poincare maps in a family of dynamical systems with multiple time scale dynamics and mixed mode oscillations. Here, the dynamical model has been used as a representative dynamical system for this family. The scope of the study is to understand the dynamical features of the MMO systems within the framework of one dimensional systems. Specifically, by using some general constraints on the one dimensional map, we first analyze the basic mechanism that is responsible for the reversal of periodic sequences of RLk type which corresponds to the dominant periodic states of the MMO systems. This in turn allows us to understand the period adding sequences as well. The analysis also helps to demonstrate that the width of the periodic states contained within the chaotic regions bounded by two successive periodic states of the form RLk is smaller than that for RLk .To this end, we first construct a model map which mimics the dominant bifurcation sequences of MMO systems. This map is utilized to verify the analytical results for the parameter width of the periodic windows. This analysis also throws light on the origin of the ordered structure of the isolas of RLk periodic orbits, in MMO systems, which was shown to be the result of a back-to-back Hopf bifurcation. The results indicate the ubiquity in the qualitative dynamical features of physical systems from widely differing origin, exhibiting alternate periodic-chaotic sequences. Although the model for the PLC effect is successful in describing the features of the phenomenon, a shortcoming of the dynamical model has been the absence of the spatial aspect. A dominant process in the PLC effect is the movement of dislocations (mainly through cross glide) which is essentially nonlocal. This feature has been incorporated into the dynamical model through a 'diffusive' term for the mobile dislocations. Preliminary results indicate that various types of band propagation, as seen in experiments, are recovered. It is known that the solute atmosphere aggregation occurs primarily during the waiting time of the mobile dislocations after its arrest. As another extension, the present model has been revised to incorporate these aging effects also. An outline of the thesis is as follows. Focus of this thesis work is on the dynamical aspects of the PLC effect. The phenomenology and few techniques in nonlinear dynamics are introduced in Chapters 1 and 2. Chapter 3 provides a comprehensive tour of dynamical behavior of the model in physically relevant two-parameter space. The rest of the work is presented in three parts (six chapters). In the first part of the thesis, the structure of the relaxation oscillations in the phase space is analyzed using the topology of the slow manifold. A connection between the slow manifold structure and the negative strain rate sensitivity of the flow stress is attempted using this analysis (Chapter 4). As a natural extension, the approach is utilized for the analysis of multiperiodic relaxation oscillations also. The emphasis is on the connection between the dynamical behavior of the model and incomplete approach to a global bifurcation (Chapter 5). In the second part of the thesis, the stability properties of periodic orbits are analyzed in detail using the Poincare map formalism, complimenting the study on the structure of periodic orbits using slow manifold. The structure and gross features of the Poincare map are reproduced utilizing only minimum information regarding the principal periodic orbit in the multidimensional space (Chapter 6). Within the framework of one dimensional systems, we analyze the mechanisms responsible for the structure of bifurcation portraits of MMO systems (Chapter 7). Third and the last part, of work focuses on modeling the spatial aspect of the PLC effect and refinement of the dynamical model (Chapters). The last chapter, Chapter9, is devoted for discussion of the results and scope for future work.

La dynamique des systèmes hyperboliques est considérée bien comprise du point de vue topologique aussi bien que du point de vue stochastique. S. Smale et R. Abraham ont donné un exemple montrant que, en général, les systèmes hyperboliques ne sont pas denses parmi tous les systèmes diffélrentiables. Dans les années 1970, M. Brin et Y. Pesin ont proposé une nouvelle notion: hyperbolicité partielle pour affaiblir la notion d’hyperbolicité. Un but de cette thèse est de comprendre la dynamique de certains systèmes partiellement hyperboliques du point de vue stochastique aussi bien que du point de vue topologique. Du point de vue stochastique, nous démontrons les résultats suivants: — Il existe un sous-ensemble U ouvert et dense de difféomorphismes non hyperboliques robustement transitifs loin de tangences homocliniques, tels que pour tout f ∈ U, il existe des mesures ergodiques non hyperboliques qui sont limite faible des mesures périodiques, avec un seul exposant de Lyapunov nul, et dont les supports sont la variété entière; — Il existe un sous-ensemble ouvert et dense de l’ensemble des difféomorphismes partiellement hyperboliques (mais non hyperboliques) de dimension centrale un dont les feuilletages forts sont robustement minimaux, de sorte que la fermeture de l’ensemble des mesures ergodiques est l’union de deux convexes qui sont la fermeture des ensembles de mesures ergodiques hyperboliques de deux s-indices différents respectivement; ces deux ensembles convexes se coupent le long de la fermeture de l’ensemble des mesures ergodiques non hyperboliques. Par conséquent, toute mesure ergodique non hyperbolique est approchée par des mesures périodiques. C’est le cas pour une perturbation robustement transitive du temps un d’un flot d’Anosov transitif, ou du produit fibré d’un difféomorphisme d’Anosov sur le tore par une rotation du cercle. Ces résultats sont basés sur des résultats locaux dont les démonstrations impliquent beaucoup de définitions techniques. Du point de vue topologique, pour tout flot d’Anosov non transitif sur des variétés de dimension 3 orientables, nous construisons de nouveaux difféomorphismes partiellement hyperboliques en composant le temps t des flots d’Anosov (pour t > 0 large) avec des twists de Dehn le long des tores transversaux. Ces nouveaux difféomorphismes partiellement hyperboliques sont robustement dynamiquement cohérents. Cela généralise dans un cas général le processus spécial dans [BPP] pour construire de nouveaux difféomorphismes partiellement hyperboliques. De plus, nous démontrons que pour les nouveaux difféomorphismes partiellement hyperboliques que nous avons construits, leurs feuilletages centraux sont topologiquement équivalentes aux flots d’Anosov utilisés pour les construire. En conséquence, la structure des feuilles centrales des nouveaux difféomorphismes partiellement hyperboliques est la même que la structure des orbites d’un flot d’Anosov. La présence de mesures ergodiques non hyperboliques montre la non hyperbolicité des systémes. Dans cette thése, nous cherchons également à comprendre: dans quelle mesure la présence de mesures ergodiques non hyperboliques peut-elle caractériser le degré de non-hyperbolicité des systèmes? Nous démontrons que, pour les difféomorphismes génériques, si une classe homoclinique contient des orbites périodiques d’indices différents et sans certaines dominations, il existe une mesure ergodique non hyperbolique avec plus d’un exposant de Lyapunov qui s’annule et dont le support est la classe homoclinique entière. Le nombre d’exposants de Lyapunov nuls montre combien d’hyperbolicité a été perdue dans un tel type de systèmes. / The dynamics of hyperbolic systems is considered well understood from topological point of view as well as from stochastic point of view. S. Smale and R. Abraham gave an example showing that, in general, the hyperbolic systems are not dense among all differentiable systems. In 1970s, M. Brin and Y. Pesin proposed a new notion: partial hyperbolicity to release the notion of hyperbolicity. One aim of this thesis is to understand the dynamics of certain partially hyperbolic systems from stochastic point of view as well as from topological point of view. From stochastic point of view, we prove the following results: — There exists an open and dense subset U of robustly transitive nonhyperbolic diffeomorphisms far from homoclinic tangency, such that forany f ∈ U, there exist non-hyperbolic ergodic measures as the weak*- limit of periodic measures, with only one vanishing Lyapunov exponent, and whose supports are the whole manifold; — There exists an open and dense subset of partially hyperbolic (but nonhyperbolic) diffeomorphisms with center dimension one whose strong foliations are robustly minimal, such that the closure of the set of ergodic measures is the union of two convex sets which are the closure of the sets of hyperbolic ergodic measures of two different s-indices respectively; these two convex sets intersect along the closure of the set of nonhyperbolic ergodic measures. As a consequence, every non-hyperbolic ergodic measure is approximated by periodic measures. That is the case for robustly transitive perturbation of the time one map of a transitive Anosov flow, or of the skew product of an Anosov torus diffeomorphism by a rotation of the circle. These results are based on some local results whose statements involve in lots of technical definitions. From topological point of view, for any non-transitive Anosov flow on orientable 3-manifolds, we build new partially hyperbolic diffeomorphisms by composing the time t-map of the Anosov flow (for t > 0 large) with Dehn twists along transverse tori. These new partially hyperbolic diffeomorphisms are robustly dynamically coherent. This generalizes the special process in [BPP] for constructing new partially hyperbolic diffeomorphisms to a general case. Furthermore, we prove that for the new partially hyperbolic diffeomorphisms we built, their center foliations are topologically equivalent to the Anosov flows used for building them. As a consequence, one has that the structure of the center leaves of the new partially hyperbolic diffeomorphisms is the same asthe structure of the orbits of an Anosov flow. The presence of non-hyperbolic ergodic measures shows the non-hyperbolicity of the systems. In this thesis, we also attempt to understand: to what extent, can the presence of non-hyperbolic ergodic measures character how far from hyperbolicity the systems are? We prove that, for generic diffeomorphisms, if a homoclinic class contains periodic orbits of different indices and without certain dominations, then there exists a non-hyperbolic ergodic measure with more than one vanishing Lyapunov exponents and whose support is the whole homoclinic class. The number of vanishing Lyapunov exponents shows how much hyperbolicity has been lost in such kind of systems.

Submitted by isabela.moljf@hotmail.com (isabela.moljf@hotmail.com) on 2016-08-15T14:53:56Z No. of bitstreams: 1 hugorolandojachohancco.pdf: 1087180 bytes, checksum: 30f544acc78e47c2892563f8ab257478 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-08-19T14:01:42Z (GMT) No. of bitstreams: 1 hugorolandojachohancco.pdf: 1087180 bytes, checksum: 30f544acc78e47c2892563f8ab257478 (MD5) / Made available in DSpace on 2016-08-19T14:01:42Z (GMT). No. of bitstreams: 1 hugorolandojachohancco.pdf: 1087180 bytes, checksum: 30f544acc78e47c2892563f8ab257478 (MD5) Previous issue date: 2016-07-18 / Consideramos os campos vetoriais C1 sobre uma variedade riemanniana compacta, sem bordo, de dimensão finita n, com n ≥3. Uma classe homoclínica de um campo vetorial é o fecho de conjunto de pontos homoclínicos transversais associados com uma órbita periódica hiperbólica. Neste trabalho, provamos que as classes homoclínicas para um conjunto residual de campos vetoriais C1 são conjuntos neutrais, mais ainda, a classe homoclínica é a intersecção dos fechos do conjunto estável e o conjunto instável. Como consequencia das propriedades do conjuntos neutrais, provamos as propriedades genéricas das classes homoclínicas. Assim, provamos que as classes homoclínicas de campos vetoriais C1-genérico X são conjuntos transitivos maximais, saturados e que dependem continuamente da órbita periódica. Também provamos que uma classe homoclínica de X não apresentam ciclos de X formados por classes homoclínicas de X. Além disso, uma classe homoclínica de X é isolado se, e somente se, é Ω-isolado. Mais ainda, é isolado, se a classe homoclínica é hiperbólica. Todas estas propriedades são bem conhecidos para campos vetoriais estruturalmente estáveis e Axioma A. / We consider the vector fields C1 on a compact Riemannian manifold, boundaryless of finite dimension n, with n ≥3. A homoclinic class of a vector field is the closure of the set transverse homoclinic point associated with a hyperbolic periodic orbit. In this work, we prove that the homoclinic classes for a residual set of vector fields C1, are neutral sets, moreover, the homoclinic class is the intersection of the closure the stable set and unstable set. As a consequence of the properties of the neutral sets, we prove the generic properties of homoclinic classes. Thus, we proved that in the homoclinic classes of generic C1 vector fields X are maximal transitive sets, saturated and depend continuously on the periodic orbit. We also proved that a homoclinic class X, does not exhibit cycles of X formed by homoclinic class of X. Furthermore, homoclinic class X is isolated if it only if it is Ω-isolated. But still, it is isolated, the homoclinic class is hyperbolic. All these properties are well known to structurally stable vector fields and Axiom A.

Neuronen sind bei der Informationsübertragung des zentralen Nervensystems von entscheidender Bedeutung. Ihre Aktivität liegt der Signalverarbeitung und höheren kognitiven Prozessen zugrunde. Neuronen sind in den extrazellulären Raum eingebettet, der mehrere Teilchen, darunter auch Ionen, enthält. Ionenkonzentrationen sind nicht statisch. Intensive neuronale Aktivität kann intrazelluläre und extrazelluläre Ionenkonzentrationen verändern. In dieser Arbeit untersuche ich das Wechselspiel zwischen neuronaler Aktivität und der Dynamik der Ionenkonzentrationen. Dabei konzentriere ich mich hauptsächlich auf extrazelluläre Kalium- und intrazelluläre Natriumkonzentrationen. Mit Hilfe der Theorie dynamischer Systeme zeige ich, wie moderate Änderungen dieser Ionenkonzentrationen die neuronale Aktivität qualitativ verändern können, wodurch sich möglicherweise die Signalverarbeitung verändert. Dann modelliere ich ein leitfähigkeitsbasiertes neuronales Netzwerk mit Spikes. Das Modell sagt voraus, dass eine moderate Änderung der Konzentrationen, die einen Mikroschaltkreis von Neuronen umgeben, die Leistungsspektraldichte der Populationsaktivität verändern könnte. Insgesamt unterstreicht diese Arbeit die Bedeutung der Dynamik der Ionenkonzentrationen für das Verständnis neuronaler Aktivität auf langen Zeitskalen und liefert technische Erkenntnisse darüber, wie das Zusammenspiel zwischen ihnen modelliert und analysiert werden kann. / Neurons are essential in the information transfer mechanisms of the central nervous system. Their activity underlies both basic signal processing, and higher cognitive processes. Neurons are embedded in the extracellular space, which contains multiple particles, including ions which are vital to their functioning. Ionic concentrations are not static, intense neuronal activity alters the intracellular and extracellular ionic concentrations which in turn affect neuronal functioning. In this thesis, I study the interplay between neuronal activity and ionic concentration dynamics. I focus specifically on the extracellular potassium and intracellular sodium concentrations. Using dynamical systems theory, I illustrate how moderate changes in these ionic concentrations can qualitatively change neuronal activity, potentially altering signal processing. I then model a conductance-based spiking neural network. The model predicts that a moderate change in the concentrations surrounding a microcircuit of neurons could modify the power spectral density of the population activity. Altogether, this work highlights the need to consider ionic concentration dynamics to understand neuronal activity on long time scales and provides technical insights on how to model and analyze the interplay between them.

[pt] Na primeira parte da tese, consideramos difeomorfismos com ciclos heterodimensionais associados a um par de selas P e Q de co-índice dois. Provamos que difeomorfismos com ciclos que possuem no mínimo um par de autovalores centrais do ciclo não real geram ciclos heterodimensionais robustos. Além disso, quando os autovalores centrais são não-reais, os ciclos robustos estão associados as continuações das selas iniciais (ou seja, os ciclos podem ser estabilizados). Na segunda parte deste trabalho estudamos mapas produto cruzado sobre aplicações deslocamento (do tipo Bernoulli) com fibras contrativas e dependência Holder nos pontos da base. Provamos que sistemas que satisfazem a propriedade de cobertura exibem blender simbólicos. Estes blenders são generalizações do blender usual cuja principal característica é que suas direções centrais podem ter qualquer dimensão d maior ou igual que 1. / [en] In the first part of the thesis, we consider diffeomorphisms having heterodimensional cycles associated with a pair of saddles P and Q of co-index two. We prove that diffeomorphisms with cycles, which have at least one pair of non-real central eigenvalues, generate robust heterodimensional cycles. Moreover, when both central eigenvalues are non-real, the robust cycles are associated with the continuation of the initial saddles (i.e. the cycle can be stabilized). In the second part of this work we study skew product maps over Bernoulli shifts with contracting fibers and Holder dependence on the base points. We prove that systems satisfying the covering property exhibit symbolic blenders. These blenders are generalizations of the usual blenders whose main property is that their central direction may have any dimension d greater than or equal to 1.