This thesis investigates how group-level differences in willingness of opinion expression shape the extent to which certain standpoints are visible in public debate online. Against the backdrop of facilitated communication and connection to like-minded others through digital technologies, models and methods are developed and case studies are carried out – by and large from a network perspective. To this end, we first propose a model of opinion dynamics that examines social- structural conditions for public opinion expression or even predominance of different groups. The model focuses not on opinion change, but on the decision of individuals whether to express their opinion publicly or not. Groups of agents with different, fixed opinions interact with each other, changing the willingness to express their opinion according to the feedback they receive from others. We formulate the model as a multi-group game, and subsequently provide a dynamical systems perspective by introducing reinforcement learning dynamics. We show that a minority can dominate public discourse if its internal connections are sufficiently dense. Moreover, increased costs for opinion expression can drive even internally well-connected groups into silence. We then focus on how interaction networks can be used to infer political and social positions. For this purpose, we develop a new type of force-directed network layout algorithm. While being widely used, a rigorous interpretation of the outcomes of existing force-directed algorithms has not been provided yet. We argue that interpretability can be delivered by latent space approaches, which have the goal of embedding a network in an underlying social space. On the basis of such a latent space model, we derive a force-directed layout algorithm that can not only be used for the spatialisation of generic network data – exemplified by Twitter follower and retweet networks, as well as Facebook friendship networks – but also for the visualization of surveys. Comparison to existing layout algorithms (which are not grounded in an interpretable model) reveals that node groups are placed in similar configurations, while said algorithms show a stronger intra-cluster separation of nodes, as well as a tendency to separate clusters more strongly in retweet networks. In two case studies, we observe actual public debate on the social media platform Twitter – topics are the Saxon state elections 2019, and violent riots in the city of Leipzig on New Year’s Eve in the same year. We show that through the interplay of retweet and reply networks, it is possible to identify differences in willingness of opinion expression on the platform between opinion groups. We find that for both events, propensities to get involved in debate are asymmetric. Users retweeting far-right parties and politicians are significantly more active, making their positions disproportionately visible. Said users also act significantly more confrontational in the sense that they reply mostly to users from different groups, while the contrary is not the case. The findings underline that naive reliance on what others express online can be collectively dangerous, especially in an era in which social media shapes public discourse to an unprecedented extent.

Human mobility is a complex phenomenon emerging from the nexus between social, demographic, economic, political and environmental systems. In this thesis we develop novel mathematical models for the study of complex systems, to improve our understanding of mobility patterns and enhance our ability to predict local and global flows for real-world applications.The first and second chapters introduce the concept of human mobility from the point of view of complex systems science, showing the relation between human movements and their predominant drivers. In the second chapter in particular, we will illustrate the state of the art and a summary of our scientific contributions. The rest of the thesis is divided into three parts: structure, causes and effects.The third chapter is about the structure of a complex system: it represents our methodological contribution to Network Science, and in particular to the problem of network reconstruction and topological analysis. We propose a novel methodological framework for the definition of the topological descriptors of a complex network, when the underlying structure is uncertain. The most used topological descriptors are redefined – even at the level of a single node – as probability distributions, thus eluding the reconstruction phase. With this work we have provided a new approach to study the topological characteristics of complex networks from a probabilistic perspective. The forth chapter deals with the effects of human mobility: it represents our scientific contribution to the debate about the COVID-19 pandemic and its consequences. We present a complex-causal analysis to investigate the relationship between environmental conditions and human activity, considered as the components of a complex socio-environmental system. In particular, we derive the network of relations between different flavors of human mobility data and other social and environmental variables. Moreover, we studied the effects of the restrictions imposed on human mobility – and human activities in general – on the environmental system. Our results highlight a statistically significant qualitative improvement in the environmental variable of interest, but this improvement was not caused solely by the restrictions due to COVID-19 pandemic, such as the lockdown.The fifth and sixth chapters deal with the modelling of causes of human mobility: the former is a concise chapter that illustrate the phenomenon of human displacements caused by environmental disasters. Specifically, we analysed data from different sources to understand the factors involved in shaping mobility patterns after tropical cyclones. The latter presents the Feature-Enriched Radiation Model (FERM), our generalization of the Radiation Model which is a state-of-the-art mathematical model for human mobility. While the original Radiation Model considers only the population as a proxy for mobility drivers, the FERM can handle any type of exogenous information that is used to define the attractiveness of different geographical locations. The model exploits this information to divert the mobility flows towards the most attractive locations, balancing the role of the population distribution. The mobility patterns at different scales can be reshaped, following the exogenous drivers encoded in the features, without neglecting the global configuration of the system.

No

In order to determine the dynamic factors inducing the evolution of environmental management strategies in the context of sustainable development, I draw upon theory from the field of general evolutionary sciences (including chaos theory, complexity science, and nonlinear thermodynamics). I translate this emerging theoretical worldview to a general evolutionary methodology for the analysis of sustainable development strategies. Application of the methodology to selected case studies indicates that sustainable development strategies co-evolve in response to organizational values, technology, and organizational structure inside the firm, and to the environmental management field outside the firm. Competing notions of sustainable development influence the politicization process that limits the types of sustainable development strategies chosen and implemented. The evidence also indicates that new pathways of efficiency are emerging in sustainable development based on market driven strategies, institutional partnerships, and the formation of new industries. The dissertation is based on case study analysis of three sustainable development projects of three different environmental organizations, the International Institute for Energy Conservation, the Environmental Law Institute, and Sanders International. The results have implications for sustainable development theory and practical implications for policy analysts and sustainable development advocates, as well as for a subtler and deeper personal understanding of our place in the world. / Ph. D.

Photosynthesis involves the absorption of photons by light-harvesting pigments and the subsequent transfer of excitation from the absorption centre to the reaction centre. This highly efficient phenomenon of excitation transfer has traditionally been explained by the Forster mechanism of incoherent hopping of excitation from one chromophore to another. Recently 2D electronic spectroscopic evidences were gathered by Fleming and coworkers on the photosynthetic Fenna-Matthews-Olson (FMO) complex in green sulfur bacteria [1]. Subsequent simulation studies by the same group [2] led to the proposition of a quantum-mechanical, coherent, wave-like transfer of excitation among the chromophores. However, Fleming's conclusions regarding retention of coherence appeared surprising because, the complex would interact with the numerous degrees of freedom of the protein scaffold surrounding it, leading to decoherence, which is expected to be rapid. Thus, we were interested in proposing an analytical treatment to rationalize the excitation transfer. Traditional approaches employed for studying excitation energy transfer involve the master equation techniques where the system-bath coupling is perturbative and is truncated after a few orders. It is important to note that the system-bath coupling causes both decoherence and population relaxation. Such a perturbative approximation is difficult to justify for the photosystem, as the system-bath coupling and the interchromophoric electronic coupling have comparable values. Also, these treatments are largely numerical studies and demand involved calculations. Thus, exact calculations for such a system (7-level) are very difficult. Consequently, we were interested in developing an analytical approach where the coupling is treated as non-perturbative. We devised a novel analytical treatment which employs a unitary transformation analogous to the one used for the theory of nonadiabatic effects in chemical reactions [3]. Our treatment rests on an adiabatic basis which are eigenstates calculated at each nuclear position (i.e. at each configuration of the bath) bearing a parametric dependence in Qi, where Qi denotes the shift of the exciton at site `i' due to the environment. The treatment is justified because in the case of coherent transfer, the excitation would travel mostly amongst the adiabatic states and the effects of non-adiabaticity are small. We observed that the system-bath coupling, after the unitary transformation, could be decoupled at the lowest order into two parts: a) an adiabatic contribution, which accounts solely for decoherence (this is evaluated almost exactly in our approach) and b) a non-adiabatic contribution which accounts for population relaxation from one adiabatic state to another (treated by a Markovian master equation). When we applied our technique to the FMO complex, our prediction for population evolution at the chromophores showed excellent correspondence with those obtained by Nalbach and coworkers using path-integral calculations [4], which are exact. These were calculations where the environment was modelled using a Drude spectral density. Our method allowed the calculations to be readily performed for different temperatures as well. It should be specifically emphasized that, unlike the involved and cumbersome path-integral calculations by Nalbach and coworkers [4] or the hierarchical equation calculations by Ishizaki et al. [2], our method is simple, easy to apply and computationally expedient. Further it became evident that the ultra-efficiency of energy transfer in photosynthetic complexes is not completely captured by coherence alone but is the result of an interplay of coherence and the dissipative influence of the environment (also known as ENAQT or Environment Assisted Quantum Transport [5]). An added advantage of our analytical treatment was the flexibility it offered. Thus, we could use our formalism to perform expedient analyses on the behavior of the system under various conditions. For example, we may wish to evaluate the consequences of introducing correlations among the bath degrees of freedom on the efficiency of transfer to the reaction centre. To this end, we applied our formalism by introducing correlations among the bath degrees of freedom and then by introducing anticorrelations among the bath degrees of freedom. The conclusions were interesting, for they suggested that the efficiency of transfer to the reaction centre was enhanced by the presence of anti-correlations, when compared with an uncorrelated bath. Uncorrelated baths, in turn, had a higher efficiency of energy transfer than correlated baths [6]. Thus, the population evolution is fastest for the anti-correlated bath, followed by the uncorrelated bath and is slowest for the correlated bath. Similar conclusions have been reached at by Tiwari et al. [7]. We could also extend the formalism for studying the system under different spectral densities for the environment, apart from just the Drude spectral density which is popularly used in literature associated with FMO calculations. For instance, the FMO system could be analyzed for the Adolphs-Renger spectral density [3, 8]. Once again our results showed excellent agreement with those reported by Nalbach. We also analyzed the FMO system under the spectral density proposed by Kleinekathofer and coworkers [9]. It was found that these latter spectral densities had more profound participation from the environment, therefore coherences were destroyed more effectively and population relaxation was faster. The excitation transfer to the final site (site closest to the reaction centre in the FMO complex) was found to be faster for the Adolphs and Renger spectral density and the spectral density proposed by Kleinekathofer and coworkers, when compared to the Drude spectral density. Also, the excitation transfer was fastest when we modelled the environment using the Kleinekathofer spectral density. This reinforced the previous conclusions that the dissipative effects of the environment promote a faster energy transport. Being an almost analytical approach, our technique could be applied to systems with larger number of levels as well. A good example of such a case is the MEH-PPV polymer. 2D electronic-spectroscopic experiments performed on this polymer in solution speculate that the excitation energy transfer might be coherent even at physiological temperatures [10]. A prototype for studying this system might be a conjugated polymer with around 80-100 chromophores. Linewidths and Lineshapes in the vicinity of Graphene It has been reported that a vibrating dipole may de-excite by transferring energy non-radiatively to a neighboring metal surface [11]. It is also understood that due to its delocalized pi-cloud, graphene has a continuum of energy states and can behave like a metal sheet and accept energies. Thus, we proposed that if a vibrationally excited dipole de-excites in the vicinity of a graphene sheet, graphene may get electronically excited and thus serve as an effective quencher for such vibrational excitations. Depending on the distance of the dipole from the graphene sheet, the transfer might be intense enough to be spectroscopically probed. We have investigated the rate of such an energy transfer. We use the Dirac cone approximation for graphene, as this enables us to obtain analyt-ical results. The Fermi Golden rule was used to evaluate the rate of energy transfer from the excited dipole to the graphene sheet [12]. The calculations were performed for both the instances: a) energy transfer from a dipole to undoped graphene and, b) energy trans-fer from a dipole to doped graphene. For undoped graphene, the carrier (electron) charge density in the conduction band is zero and we would only have transitions from the valence band to the conduction band. As a consequence of absence of carrier charge density in CB (conduction band), the screening of Coulombic interactions in the graphene plane is ineffective. Thus, one could use the non-interacting polarizability for undoped graphene in the rate expression [13]. However, when we consider the case of doped graphene where EF is shifted upwards into CB, the conduction band electrons will contribute to screening. In this case, we have two sets of transitions: a) from ki in VB (valence band) to kf in CB and b) ki in CB to kf in CB, where ki and kf are the wavevectors which correspond to the initial and final electronic states in graphene. So we have used the polarizability propagator in the random phase approximation [14] to calculate the rate following the approach of [13]. It is also known that the imaginary part of the frequency domain dipole-dipole corre-lation function is a measure of the lineshape [15]. We were, thus, interested in evaluating the lineshape for these transitions. For evaluating the correlation function, we used the partitioning technique developed by L•owdin [16] and subsequently extracted the lineshape from its imaginary part. Using this method, we calculated lineshape for the vibrational excitation of CO molecule in the vicinity of an undoped graphene lattice. The linewidth for this system also was obtained. It could be seen that the vibrational linewidth for 1 CO in the vicinity (5 A) of undoped graphene (EF = 0:00eV ) is small (0:012 cm ) but could be observed experimentally. The lineshape calculations were also extended to cases where it is possible to have atomic transitions by placing an electronically excited atom in the vicinity of the graphene sheet. We considered the following two cases: a) 3p ! 2s transition in hydrogen atom, at a distance of 12 A from the graphene sheet and, b) 4p ! 3s transition in hydrogen atom, at a distance of 20 A from the graphene sheet. The linewidths for atomic transitions could be easily probed in these cases ( 55 cm 1 for 3p ! 2s and 56 cm 1 for 4p ! 3s). In the preceding calculations, the transi-tion dipoles were considered perpendicular to the graphene surface. It is worthwhile to note that if the transition dipoles are considered parallel to the graphene surface, the respective linewidths would be half of those obtained for the case where the transition dipoles are perpendicular. Another interesting possibility would be to consider a lanthanide metal complex placed within a few nanometers from graphene. Lanthanides are known to have sharp f-f transitions [17] and consequently, one could easily observe the effects of broadening due to energy transfer to the electronic system of graphene. Energy Eigenmodes for arrays of Metal Nanoparticles In the final part of the thesis we consider organized assemblies of metal nanoparti-cles, specifically helical and cylindrical assemblies and investigate the plasmonic excitation transfer across these assemblies. These were motivated by recent studies which reported growth of chiral asymmetric assemblies of nanoparticles on D and L- isomers of dipheny-lalanine peptide nanotubes [18]. The plasmons in the helical/cylindrical assemblies are expected to couple with each other via electromagnetic interactions. We construct the Hamiltonian for such systems and evaluate the eigenmodes and energies pertaining to these modes in the wave vector space. We also perform calculations for the group velocity for each eigenmode as this gives us an idea of which eigenmode transports excitation the fastest.

The Normal Accident Theory (NAT) attempts to understand why accidents occur in systems with high-risk technologies. NAT is characterized by two attributes: complexity and coupling. The combination of these attributes results in unplanned and unintended catastrophic consequences. High-risk technology systems that are complex and tightly coupled have a high probability of experiencing system failures. The mining industry has experienced significant incidents involving haulage operations up to and including severe injuries and fatalities. Although the mining industry has dramatically reduced fatalities and lost time accidents over the last three decades or more, accidents still continue to persist. For example, for the years 1998 - 2002, haulage operations in surface mines alone have accounted for over 40% of all accidents in the mining industry. The systems thinking was applied as an approach to qualitatively and quantitatively evaluate NAT in mining haulage operations. A measurement index was developed to measure this complexity. The results from the index measurements indicated a high degree of complexity that exists in haulage transfer systems than compared to loading and unloading systems. Additionally, several lines of evidence also point to the applicability of NAT in mining systems. They include strong organizational management or safety system does not guarantee zero accidents, complexity is exhibited in mining systems, and they are interactive and tightly coupled systems. Finally, the complexity of these systems were assessed with results indicating that a large number of accidents occur when there are between 4 or 5 causal factors. These factors indicate the degree of complexity necessary before accidents begin to occur.

Ecosystems are often made up of interactions between large numbers of species. They are considered complex systems because the behaviour of the system as a whole is not always obvious from the properties of the individual parts. A complex system can be represented by a network: a set of interconnected objects. In the case of ecological networks and food webs, the objects are species and the connections are interactions between species. Many complex systems are dynamic and exhibit intricate time series. Time series analysis has been developed to understand a wide range of natural phenomena. This thesis deals with the structure, dynamics, and robustness of ecological networks, the spatial dynamics of fluctuations in a social system, and the analysis of cardiac time series. Biodiversity on Earth is decreasing largely due to human-induced causes. My work looks at the effect of anthropogenic change on ecological networks. In Chapter Two, I investigate predator adaptation on food-web robustness following species extinctions. I identify a new theoretical category of species that may buffer ecosystems against environmental change. In Chapter Three, I study changes in parasitoid-host (consumer-resource) interaction frequencies between complex and simple environments. I show that the feeding preferences of parasitoid species actively change in response to habitat modification. Ecological networks are embedded in spatially-heterogeneous landscapes. In Chapter Four, I assess the role of geography on population fluctuations in an analogous social system. I demonstrate that fluctuations in the number of venture capital firms registered in cities in the United States of America are consistent with spatial and temporal contagion. Understanding how physiological signals vary through time is of interest to medical practitioners. In Chapter Five, I present a technique for quickly quantifying disorder in high frequency event series. Applying the algorithm to patient cardiac time series provides a rapid way to detect the onset of heart arrhythmia. Increasingly, answers to scientific questions lie at the intersection of traditional disciplines. This thesis applies techniques developed in physics and mathematics to problems in ecology and medicine.

Dans ce mémoire, différentes architectures pour le contrôle et la surveillance des systèmes techniques complexes de grande dimension (STCGD) sont discutées. Les problématiques de maintenance conditionnelle et d'évaluation de l'état de santé sont définies. Les types de diagnostic et de pronostic sont présentés afin d'aboutir à une évaluation de l'état de santé des STCGD. Les études relatives au diagnostic décentralisé sont discutées puis les apports des NTIC et des technologies distribuées au diagnostic sont présentés. Par la suite, le diagnostic distribué et les travaux relatifs à ce mode de déploiement sont introduits. Les limites des approches centralisées et décentralisées du diagnostic sont présentées et confrontées à l'apport des approches distribuées. Les informations et/ou les connaissances supports aux diagnostic et au pronostic ainsi que leur modélisation afin de les exploiter sont décrites et formalisées. Une caractérisation des statuts que peut prendre un composant est proposée. Il est décrit les pré-requis nécessaires pour la couche de surveillance des STCGD et les principes du diagnostic et du pronostic sont ensuite présentés sous la forme de différents algorithmes. Enfin, une méthode d'évaluation de l'état de santé des STCGD est proposée. Plusieurs déploiements peuvent être envisagés pour l'évaluation de la santé des STCGD. Une plateforme de simulation a été développée pour évaluer les performances des déploiements centralisés et distribués. La plateforme a eu pour but de se comporter comme la couche de surveillance d'un STCGD. Un cas d'étude paramétrable est proposé pour chacun des deux déploiements et leurs performances sont comparées. / In this dissertation, various architectures for the control and the monitoring of Large Scale Complex Technical Systems (LSCTS) are discussed. The problematic of condition-based maintenance and health status assessment is defined. A diagnostic and prognostic typology is presented leading to the assessment of the health status of LSCTSs. Decentralized diagnosis studies are discussed then the contributions of the ICT and of the distributed technologies for the diagnosis are presented. Thereafter, the distributed diagnosis and works relative to this kind of deployments are introduced. The limits of the centralized and decentralized diagnosis approaches are presented. Then the centralized approaches are compared to the distributed ones. Information and/or knowledge that support the diagnosis and the prognosis as well as their modeling in order to exploit them are described and formalized. A characterization is proposed for the different status of a component can be in. Requirements are described for the monitoring layer of the LSCTSs are described in order to implement the proposed diagnosis and prognosis principles that are then specified by the means of algorithms. Eventually, a health assessment method of the LSCTSs is also proposed. Several deployments can be considered to implement the health assessment of the LSCTSs. A simulation platform, which was developed to evaluate the performances of the centralized and the distributed deployments, is presented. Among the purposes of the platform, one is to behave as the monitoring layer of a LSCTS. A use case is proposed for two deployments and their performances are compared.

Cette thèse se propose de montrer comment les interactions vont influer sur le système complexe de communication d'une formation à distance, et en dévoiler les limites. L'objectif général est de montrer que, dans ce système, il est possible d'expliquer l'abandon par les pertes de prises qui se dessinent dans la dynamique entre les représentations et les communications des acteurs, et que certaines limites du système favorisent ces pertes de prise. La finalité de cette thèse est de cerner ces limites pour définir une catégorie d'abandon communicationnel. Cette recherche qualitative s'appuie sur la théorie de la communication pragmatique de Palo Alto, pour analyser les interactions entre les acteurs, et sur la théorie des représentations sociales, pour traduire en images l'évolution des représentations sur une année. Le concept de « prise » relie la situation communicationnelle à l'abandon. Il permet de confronter les repères communs aux acteurs et leurs perceptions personnelles des situations vécues, pour rendre compte de leur moyen d'agir et de s'engager. / The purpose of this thesis is to show how interactions can influence a complex system of distance training communication, and reveal its limits. The main aim is to show that dropping out can be explained by loss of grips from the representations and communication dynamics of involved actors, and how some limits of the system favor this loss of grips. The end purpose of this thesis is to number these limits by defining a category of communicational drop out. This qualitative research is based on the Palo Alto theory of pragmatic communication, to analyze interactions between actors, as well as the theory of social representations by translating the evolution of representations into images, over one year. The concept of “grip” creates a link between communicational events and dropping out. It confronts actors' shared landmarks with their personal perception of lived events, which in turn, accounts for their level of commitment and action.

Les technologies modernes permettent d'avoir des renseignements toujours plus précis sur les interactions entre individus. Dans ce contexte, la collaboration SocioPatterns a permis de développer une infrastructure mesurant, avec une très grande résolution temporelle, la proximité face-à-face d'individus volontaires, portant des badges de radio-identification. Cette infrastructure a été déployée dans divers contextes, tels que des conférences scientifiques, un musée, une école ou encore un service hospitalier. La simple analyse de ces données représente un enjeu majeur pour l'étude de la dynamique humaine et soulève des questions aussi fondamentales que la recherche d'outils et de techniques d'analyse adaptés. Cette thèse présente la caractérisation statistique de la dynamique de proximité physique, mise en relation avec le contexte et les autres métadonnées disponibles, telles que l'âge, le sexe des individus, ou bien la structure de leurs réseaux sociaux virtuels. Si la structure des contacts diffère considérablement selon le contexte, les distributions empiriques des durées des interactions et entre interactions sont très similaires. Un modèle individu-centré, présenté dans cette thèse, propose des règles d'interactions microscopiques simples susceptibles de donner lieu à cette structure macroscopique complexe des temps d'interaction. Enfin, la caractérisation de la dynamique des contacts entre individus constitue une étape cruciale pour comprendre les mécanismes de propagation de maladies telles que la grippe dans une population. / Modern technologies allow to access to more and more detailed information on human interactions. In this context, the SocioPatterns collaboration has allowed to develop an infrastructure based on radio-identification devices, that records human proximity patterns at a fine grained resolution, among voluntary individuals. This infrastructure has been deployed in diverse contexts, such as scientific conferences, a museum, a primary school, or a hospital department. The mere analysis of these data represents a high stake for the study of human dynamics and raises fundamental issues such as the need of adequate tools and analysis techniques. This thesis presents the statistical characterization of physical proximity dynamics, put into relation with the context and other available metadata such as the age, the gender of participants or the structure of their virtual social networks. Although contact patterns considerably differ amongst the various contexts, the empirical distributions of interaction durations and of inter-contact times are very similar. An agent-based model, presented in this thesis, suggests simple microscopic interaction rules able to produce the complex macrostructure of interaction durations. In the last place, the characterization of contact dynamics constitutes a determining step for understanding spreading mechanisms of diseases such as the influenza. The human proximity data have allowed to analyze the level of information needed on contact dynamics for the elaboration of epidemiological models of contagion. Such models allow to better estimate the impact of public health strategies, e.g. the closure of school classes and targeted vaccinations.