This thesis presents various kinetic models (mainly of Boltzmann-type) for multi-agent systems over social networks. Each chapter is largely based on original work either previously published in peer-reviewed journals or as preprint article still under review at the time of writing. The thesis is divided into two parts. The first one is devoted to diffusion processes within a population, either of a canonically intended infectious disease or of a more abstract - but still extremely concerning - one such as misinformation. The second part deals with networks more directly: we focus either on dynamics where networks are the underlying structure within the population and we primarily study multi-agent systems that evolve over this structure (either static or dynamic), or on dynamics where the network itself is modeled after a multi-agent system and our primary focus is on its topology. Our main research effort is to provide models that can be applied to a real-world scenario: consequently, we complete our study of the analytical properties of the proposed models by obtaining surrogate approximations via suitable mean-field limit procedures. This allows us to accurately describe the large-time behavior of the systems, the evolution of macroscopic quantities of interest and, finally, to effectively calibrate the corresponding model to available data.
Questa tesi presenta vari modelli cinetici (principalmente di tipo Boltzmann) per sistemi multi-agente su reti sociali. Ogni capitolo si basa in gran parte su lavori originali pubblicati in precedenza su riviste peer-reviewed o come articolo preprint in fase di revisione al momento della stesura. La tesi è divisa in due parti. La prima è dedicata ai processi di diffusione all'interno di una popolazione, sia di una malattia infettiva canonicamente intesa, sia di una più astratta - ma comunque allarmante - come la disinformazione. La seconda parte si occupa direttamente delle reti: ci concentriamo sia sulle dinamiche in cui le reti sono la struttura sottostante all'interno della popolazione e studiamo principalmente i sistemi multi-agente che evolvono su questa struttura (statica o dinamica), sia sulle dinamiche in cui la rete stessa è modellata come un sistema multi-agente e studiamo soprattutto la sua topologia. Il nostro principale sforzo di ricerca è quello di fornire modelli che possano essere applicati a uno scenario reale: di conseguenza, completiamo il nostro studio delle proprietà analitiche dei modelli proposti ottenendo approssimazioni surrogate attraverso opportune procedure di limite di campo medio. Questo ci permette di descrivere accuratamente il comportamento dei sistemi nel tempo, l'evoluzione delle grandezze macroscopiche di interesse e, infine, di calibrare efficacemente il modello corrispondente ai dati disponibili.
Kinetic Equations for Many-Agent Systems On Large Networks: Emerging Patterns and Data-Oriented Approaches
FRANCESCHI, JONATHAN
2025
Abstract
This thesis presents various kinetic models (mainly of Boltzmann-type) for multi-agent systems over social networks. Each chapter is largely based on original work either previously published in peer-reviewed journals or as preprint article still under review at the time of writing. The thesis is divided into two parts. The first one is devoted to diffusion processes within a population, either of a canonically intended infectious disease or of a more abstract - but still extremely concerning - one such as misinformation. The second part deals with networks more directly: we focus either on dynamics where networks are the underlying structure within the population and we primarily study multi-agent systems that evolve over this structure (either static or dynamic), or on dynamics where the network itself is modeled after a multi-agent system and our primary focus is on its topology. Our main research effort is to provide models that can be applied to a real-world scenario: consequently, we complete our study of the analytical properties of the proposed models by obtaining surrogate approximations via suitable mean-field limit procedures. This allows us to accurately describe the large-time behavior of the systems, the evolution of macroscopic quantities of interest and, finally, to effectively calibrate the corresponding model to available data.File | Dimensione | Formato | |
---|---|---|---|
franceschi-phdthesis.pdf
accesso aperto
Dimensione
22.64 MB
Formato
Adobe PDF
|
22.64 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/192550
URN:NBN:IT:UNIPV-192550