Codebase for my MSc thesis in Computer Science (joint UvA-VU, 2025), supervised by Daan van den Berg and Cees van Leeuwen.
Simulates adaptively rewiring networks with chaotic node dynamics (logistic map) across the full density spectrum to study how connection density drives structural transitions. Uses a synchrony-based rewiring rule and stochastic blockmodeling to analyze community structure, hub formation, and global topology.
At specific densities, networks self-organize into evenly partitioned, densely connected modules — "canonical structures." Above a known fragmentation-to-small-world threshold, a stable modular regime emerges with predictable community counts and sizes. As density increases further, modularity collapses and hubs centralize — though this hub formation does not constitute a functional rich-club mediating inter-community communication. A mid-density region exhibits critical slowing and metastability, suggesting transitions between competing structural configurations.
- network/ — Network model definitions
- simulation/ — Core simulation engine and rewiring algorithms
- postrun_output/ — Analysis and community detection via DC-SBM
- runtime_output/ — Raw simulation data
- main.py — Entry point for running experiments
Python, NumPy, graph-tool, DC-SBM (degree-corrected stochastic blockmodel)
~10,000 independent simulations, each for 10 million time steps. Published in Nonlinear Science, 2026.