1,721,033 research outputs found
Modeling decision-making in social systems: a network dynamical systems approach
Full text linkL'obiettivo di questa tesi è quello di migliorare la nostra abilità di modellare i processi decisionali nei sistemi sociali, con l'obiettivo ultimo di supportare i decisori e gli enti governativi a prendere decisioni supportate da studi quantitativi. In questa direzione, abbiamo adattato l'approccio delle reti complesse, combinando diverse tecniche tipiche della teoria del controllo di sistemi dinamici e dalle scienze sociali per fornire uno strumento efficace e multidisciplinare per affrontare la complessità che caratterizza le dinamiche sociali. Per farlo, in questa tesi esploriamo innanzitutto le condizioni in cui è possibile conferire determinate proprietà di controllabilità e osservabilità a reti complesse di sistemi dinamici, al fine di quantificare la nostra capacità di guidare o monitorare i loro comportamenti collettivi in presenza di vincoli quanto più realistici possibili. Successivamente, dedichiamo la nostra attenzione a determinate reti complesse e cioè i sistemi sociali, in particolare allo studio della dinamica delle opinioni in gruppi di individui interconnessi che discutono su un certo argomento, ossia all'analisi dell'evoluzione nel tempo delle loro opinioni sotto l'effetto dell'influenza dei legami sociali e di altri meccanismi psicologici. Modellare la dinamica delle opinioni ci consente di descrivere i comportamenti collettivi che si verificano nei gruppi sociali del mondo reale e di svelare come le interazioni sociali, vale a dire la pressione dei pari e altri pregiudizi, plasmino la formazione delle opinioni e portino il gruppo a manifestare comportamenti colletivi quali il consenso, il disaccordo o la polarizzazione delle opinioni.
In questo contesto, abbiamo analizzato come influenze esterne, come quelle esercitate da personaggi che godono di una certa visibilità, i cosiddetti influencer, possano indirizzare il profilo delle opinioni di un gruppo sociale verso uno stato desiderato di regime. Ciò è stato fatto attraverso l'utilizzo di strumenti di controllo delle reti, ovvero il controllo pinning, che ci consente di comprendere come agenti con relativamente poche connessioni siano in grado di sfruttare la struttura delle interconnessioni sociali per persuadere il gruppo sociale.The aim of this thesis is to advance our ability to model the decision making-process in social systems, with the overarching goal of helping policymakers and government bodies take mathematically backed decisions. Toward this aim, we adapted a complex network approach, leveraging tools from network science, control theory, and social science which altogether offers a powerful and multidisciplinary approach to tackle the complexity that pervades social dynamics. We explored the conditions under which we can confer given controllability and observability properties to complex networks of dynamical systems, so to have an insight on our ability to steer or monitor their collective behaviors in the presence of realistic constraints. Then, we studied the opinion dynamics in groups of interconnected individuals discussing on a given topic, that is, we analyzed the evolution over time of their opinions under the effect of social ties and other psychological traits, such as stubbornness or the tendency to conform. Opinions dynamics models enable us to describe the collective behaviors that occur in real world social groups and to unveil how the social interactions, namely the peer pressure and other biases, shape our opinion formation and result in the group exhibiting behaviors such as consensus, disagreement, or polarization of opinions.
In particular, we analyzed how external influences, such as the ones exerted by opinion leaders (the so-called influencers) can steer the opinion profile of a social group towards a desired state at steady state. To this aim, we borrowed a tool from network control, namely pinning control, to show how agents with relatively few connections can exploit the structure of the social interconnections to diffuse their influence throughout the social group. Using heuristic approaches and leveraging theoretical and graphical knowledge of the network dynamical systems under investigation, we showed that a smart selection of the individuals to directly influence allows to maximize the effect of persuading actions of opinion leaders. Finally, we illustrated how the model we proposed can provide quantitative predictions on opinions' distribution in a given population, which in turn can be used to gauge the effectiveness of different awareness campaigns strategies aimed at mitigating vaccine hesitancy
Criteria for stochastic pinning control of networks of chaotic maps
No full textThis paper investigates the controllability of discrete-time networks of coupled chaotic maps through stochastic pinning. In this control scheme, the network dynamics are steered towards a desired trajectory through a feedback control input that is applied stochastically to the network nodes. The network controllability is studied by analyzing the local mean square stability of the error dynamics with respect to the desired trajectory. Through the analysis of the spectral properties of salient matrices, a toolbox of conditions for controllability are obtained, in terms of the dynamics of the individual maps, algebraic properties of the network, and the probability distribution of the pinning control. We demonstrate the use of these conditions in the design of a stochastic pinning control strategy for networks of Chirikov standard maps. To elucidate the applicability of the approach, we consider different network topologies and compare five different stochastic pinning strategies through extensive numerical simulations
On adaptive bounded synchronization in Power Network models
No full textIn this paper, we discuss a generalized model for studying bounded synchronization in complex networks. Namely, a novel adaptive strategy is proposed to control the network nodes. The aim of this strategy is to asymptotically make the mismatch between the nodes' trajectories lower than a prescribed threshold. The proposed strategy is then illustrated with reference to the Power Network model recently derived in [7]. After showing simulations that confirm the effectiveness of the approach, its limitations in view of possible implementations are discussed
Stochastic Pinning Controllability of Noisy Complex Networks
Full text linkOur ability to coordinate the behavior in networks of complex dynamical systems is often challenged by the presence of noise affecting the individual dynamics and the communication links. In the literature, conservative global conditions guaranteeing the almost sure convergence toward the desired trajectory of a virtual node, the pinner, have been derived. In this article, we identify the minimal conditions on the individual dynamics, interconnection topology, and noise intensities, so that the network exponentially converges onto the pinner's trajectory. Specifically, we broaden the master stability function approach to deal with networks of coupled stochastic differential equations, and provide necessary and sufficient conditions for local exponential pinning controllability of networks of stochastic systems. Interestingly, our analyses show that noise can be either beneficial or detrimental for pinning controllability, depending on how it diffuses in each node. Our analytical findings are illustrated with representative numerical examples
Self-tuning proportional integral control for consensus in heterogeneous multi-agent systems
No full textIn this paper, we present a distributed Proportional-Integral (PI) strategy with self-tuning adaptive gains for reaching asymptotic consensus in networks of non-identical linear agents under constant disturbances. Alternative adaptive strategies are presented, based on global or local measures of the agents' disagreement. The proposed approaches are validated on a representative numerical example. Preliminary analytical results further confirm the viability of the self-tuning strategies
Formation Control of Multi-agent Systems in an Uncertain Environment
No full textSteering the dynamics of a multi-agent system towards a balanced circular formation is a paradigmatic ex-ample of formation control that attracted extensive research interest from the engineering community. The traditional estimation and control strategies typically relied on both accurate directional measurements and a thorough knowledge of the environment, which are seldom available in applications. Here, we consider a scenario where the information exchanged by the agents is limited, the measurement being uncertain and intermittent. Additionally, we do not assume that the number of agents composing the formation is known. In this uncertain environment, we propose an estimation and control strategy that leverages interval analysis to (i) estimate the size of the formation, (ii) estimate the relative distance among the agents, and (iii) achieve a balanced circular formation. The effectiveness of the strategy is demonstrated through extensive numerical simulations
Evolving topologies for network synchronization
No full textABSTRACT
In this paper we describe possible decentralized strategies to adaptively synchronize complex networks of oscillators. Firstly, we review two adaptation strategies for the coupling gains in the network, and then we describe a completely decentralized strategy to select adaptively the network edges. The report is complemented by a representative numerical example
Topological analysis of group fragmentation in multiagent systems
No full textIn social animals, the presence of conflicts of interest or multiple leaders can promote the emergence of two or more subgroups. Such subgroups are easily recognizable by human observers, yet a quantitative and objective measure of group fragmentation is currently lacking. In this paper, we explore the feasibility of detecting group fragmentation by embedding the raw data from the individuals’ motions on a low-dimensional manifold and analyzing the topological features of this manifold. To perform the embedding, we employ the ISOMAP algorithm, which is a data-driven machine learning tool extensively used in computer vision.We implement this procedure
on a data set generated by a modified `a la Vicsek model, where agents are partitioned into two or more subsets
and an independent leader is assigned to each subset. The dimensionality of the embedding manifold is shown
to be a measure of the number of emerging subgroups in the selected observation window and a cluster analysis
is proposed to aid the interpretation of these findings. To explore the feasibility of using this approach to
characterize group fragmentation in real time and thus reduce the computational cost in data processing and
storage, we propose an interpolation method based on an inverse mapping from the embedding space to the
original space. The effectiveness of the interpolation technique is illustrated on a test-bed example with potential
impact on the regulation of collective behavior of animal groups using robotic stimuli
Synchronization and pinning control of networks via adaptation and edge snapping
No full textIn this paper, we propose novel adaptive pinning control strategies for synchronization of complex networks. The novelty of theses approaches is the adaptive selection of pinned nodes along with the fully decentralized adaptation of the coupling and control gains. The effectiveness of the proposed strategies is validated with numerical simulations on a testbed example
- …
