Flocking denotes the spontaneous onset of collective motion in systems of self-propelled agents, of which groups of birds are a prototypical example. A pretty coherent corpus of theoretical models has been introduced over the past three decades to explain how this collective behavior arises from microscopic interactions, revealing that the emergence of polar order is a manifestation of the non-equilibrium character of the dynamics. The confront with experimental data allowed for the validation and refinement of those models, and, in some cases, even for the application of quantitative inference approaches. In this thesis we employ standard methods from stochastic calculus to study problems related to the microscopic dynamics of such systems. Motivated by the availability of data collected by the CoBBS team, we firstly derive a novel Bayesian inference method for the inertial dynamics of flocks. Our inference scheme is based on a previously introduced model (the Inertial Spin Model), which is non-Markovian in the observed variables’ space. This feature raises serious technical problems, when combined with discrete-time recordings, and is common to many stochastic dynamic systems. The analytical method we propose for the Inertial Spin Model applies in fact to a larger class of processes; examples are illustrated. We also exploit an analogy between the Renormalization Group and augmentation techniques used to infer partially observed SDEs to provide an alternative proof of the lack of finite-dimensional delay vector embeddings for stochastic dynamical systems. The second focus of this thesis concerns the investigation of non-equilibrium effects in simple models for polar active matter. It is known that the emergence of polar order in systems of aligning self-propelled particles is due to the non-equilibrium character of the dynamics. We quantify the distance from equilibrium through the entropy production rate, which we measure from numerical simulations of interacting active Brownian particles. We investigate two kinds of short-ranged interaction rules, based on different notions of metrics. We find that the entropy production rate is maximal at the transition, while two equilibrium limits are reached in the deeply ordered (perfect flock) or completely disordered (ideal active gas) phase. We pivot on the entropy production rate to study how irreversibility constrains asymmetries in the steady state distribution of microstates. In the presence of pairwise forces, robust signatures of irreversibility are visible in the two-particle density, as confirmed by numerical simulations. On the contrary, in the presence of multi-particle interactions, irreversibility directly constrains only correlations among a higher number of particles. All these correlations are typically neglected in the derivation of hydrodynamic equations for polar active matter through kinetic approaches.

Con flocking si intende il fenomeno per cui sistemi di agenti autopropulsi si muovono spontaneamente nella stessa direzione: gli stormi di uccelli (flocks) sono un esempio tipico di tali sistemi. Molti modelli teorici si sono affastellati nell’utlimo trentennio per spiegare come questo comportamento collettivo possa emergere dalle interazioni microscopiche tra gli individui che compongono il gruppo, rivelando che esso è una specifica manifestazione del carattere di non-equilibrio di questi sistemi attivi. Il confronto coi dati sperimentali ha permesso di convalidare e raffinare questi modelli e, in alcuni casi, persino di tentare approcci quantitativi di inferenza statistica. In questa tesi sono impiegati metodi standard del calcolo stocastico per studiare problemi legati alla dinamica microscopica di tali sistemi. Motivati dalla disponibilità di dati raccolti dal gruppo CoBBS, si è derivato in primis un nuovo metodo di inferenza Bayesiana per la dinamica inerziale degli stormi di uccelli. Il nostro schema di inferenza si basa su un modello precedentemente introdotto (Inertial Spin Model), che, a causa del suo carattere non-Markoviano, solleva difficoltà tecniche quando viene combinato con un’osservazione a tempi discreti. Questo fatto è comune a molti sistemi dinamici stocastici, e il metodo proposto si applica in realtà a una classe di processi più ampia, di cui sono illustrati degli esempi. Inoltre, è possibile sfruttare un’analogia tra alcune tecniche usate nei problemi di inferenza e il Gruppo di Rinormalizzazione, che mostra intuitivamente l’assenza di delay vector embeddings per osservazioni parziali di processi stocastici. Il secondo argomento di ricerca di questo lavoro di tesi riguarda lo studio di effetti di non-equilibrio in semplici modelli di materia attiva polare. È noto che l’emergere di ordine polare in sistemi di particelle autoproulse è dovuto a una violazione della simmetria sotto inversioni temporali. Questa violazione è quantificata dal tasso di produzione di entropia, il quale è stato misurato in simulazioni numeriche di sistemi di particelle Browniane attive. Sono stati investigati due tipi di interazioni locali, basate su diverse nozioni di metrica. In entrambi i casi, la produzione di entropia è massima alla transizione, dove la motilità ha un impatto massimo sull’interazione tra le particelle, mentre due limiti di equilibrio sono raggiunti nella fase fortemente polarizzata o totalmente disordinata. Si è studiato inoltre l’effetto dell’irreversibilità sulla distribuzione di probabilità stazionaria dei microstati del sistema. Una produzione di entropia non nulla impone delle robuste asimmetrie, visibili nelle distribuzioni di due o più particelle. Si noti che le correlazioni associate a tali di- stribuzioni vengono tipicamente trascurate nelle teorie cinetiche usate per derivare equazioni idrodinamiche per materia attiva polare.

Microscopic dynamics of polar active systems: inference methods and signatures of irreversibility for stochastic models / Ferretti, Federica. - (2022 Feb 11).

Microscopic dynamics of polar active systems: inference methods and signatures of irreversibility for stochastic models

FERRETTI, FEDERICA
11/02/2022

Abstract

Flocking denotes the spontaneous onset of collective motion in systems of self-propelled agents, of which groups of birds are a prototypical example. A pretty coherent corpus of theoretical models has been introduced over the past three decades to explain how this collective behavior arises from microscopic interactions, revealing that the emergence of polar order is a manifestation of the non-equilibrium character of the dynamics. The confront with experimental data allowed for the validation and refinement of those models, and, in some cases, even for the application of quantitative inference approaches. In this thesis we employ standard methods from stochastic calculus to study problems related to the microscopic dynamics of such systems. Motivated by the availability of data collected by the CoBBS team, we firstly derive a novel Bayesian inference method for the inertial dynamics of flocks. Our inference scheme is based on a previously introduced model (the Inertial Spin Model), which is non-Markovian in the observed variables’ space. This feature raises serious technical problems, when combined with discrete-time recordings, and is common to many stochastic dynamic systems. The analytical method we propose for the Inertial Spin Model applies in fact to a larger class of processes; examples are illustrated. We also exploit an analogy between the Renormalization Group and augmentation techniques used to infer partially observed SDEs to provide an alternative proof of the lack of finite-dimensional delay vector embeddings for stochastic dynamical systems. The second focus of this thesis concerns the investigation of non-equilibrium effects in simple models for polar active matter. It is known that the emergence of polar order in systems of aligning self-propelled particles is due to the non-equilibrium character of the dynamics. We quantify the distance from equilibrium through the entropy production rate, which we measure from numerical simulations of interacting active Brownian particles. We investigate two kinds of short-ranged interaction rules, based on different notions of metrics. We find that the entropy production rate is maximal at the transition, while two equilibrium limits are reached in the deeply ordered (perfect flock) or completely disordered (ideal active gas) phase. We pivot on the entropy production rate to study how irreversibility constrains asymmetries in the steady state distribution of microstates. In the presence of pairwise forces, robust signatures of irreversibility are visible in the two-particle density, as confirmed by numerical simulations. On the contrary, in the presence of multi-particle interactions, irreversibility directly constrains only correlations among a higher number of particles. All these correlations are typically neglected in the derivation of hydrodynamic equations for polar active matter through kinetic approaches.
11-feb-2022
File allegati a questo prodotto
File Dimensione Formato  
Tesi_dottorato_Ferretti.pdf

accesso aperto

Tipologia: Tesi di dottorato
Licenza: Creative commons
Dimensione 10.22 MB
Formato Adobe PDF
10.22 MB Adobe PDF

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1610853
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact