Keywords

Flocking, Thermodynamics, GMM, Entropy, ER-GMM, Optimization, Modelling, Leadership

Abstract

This thesis presents a unified framework for modeling, optimizing, and analyzing collective behavior in flocking systems through a thermodynamic and statistical lens. In the first study, we employ entropy-regularized Gaussian Mixture Models (GMM) to identify distinct thermodynamic regimes—ordered, transitional, and disordered—based on the energy and entropy profiles of simulated agent-based flocks. The second study introduces the Organic Adaptive Flocking Algorithm (OAFA), a biologically inspired, entropy-aware extension of the Boids model. OAFA dynamically adjusts rule weights based on local entropy signals and is optimized via a multi-objective NSGAII framework to balance cohesion, separation, energy efficiency, and disorder. Results show that entropy minimization leads to more stable and efficient flock formations with only marginal loss in Pareto hypervolume. The third study explores emergent leadership by quantifying directional influence via transfer entropy between agents’ velocity time series. Influence networks are constructed and analyzed using graph centrality metrics, revealing that leaders consistently occupy high-information central positions and dynamically reconfigure based on thermodynamic state. Together, these studies advance the modeling of decentralized coordination by integrating probabilistic learning, adaptive control, and information flow analysis within a thermodynamic context.

Completion Date

2025

Semester

Summer

Committee Chair

Mantzaris, Alexander

Degree

Doctor of Philosophy (Ph.D.)

College

College of Sciences

Department

Statistics and Data Science

Format

PDF

Identifier

DP0029599

Language

English

Document Type

Thesis

Campus Location

Orlando (Main) Campus

Download

Share

COinS