[Paper Review] Evolving network - simulation study. From regular lattice to scale free network

Danuta Makowiec|arXiv (Cornell University)|Oct 31, 2005

Complex Network Analysis Techniques28 references12 citations

TL;DR

This paper proposes a two-step preferential rewiring algorithm that evolves a regular lattice into a stationary network ensemble, demonstrating a topological transition between a 'leafy phase' (with hubs and leaves) and a 'tangling phase' (with frequent edge circulation among low-degree vertices). The key finding is that with optimized parameters, the network self-organizes into a scale-free state with power-law degree distribution (exponent ≈ 2), driven by preferential attachment and synchronization effects.

ABSTRACT

The Watts-Strogatz algorithm of transferring the square lattice to a small world network is modified by introducing preferential rewiring constrained by connectivity demand. The evolution of the network is two-step: sequential preferential rewiring of edges controlled by $p$ and updating the information about changes done. The evolving system self-organizes into stationary states. The topological transition in the graph structure is noticed with respect to $p$. Leafy phase - a graph formed by multiple connected vertices (graph skeleton) with plenty of leaves attached to each skeleton vertex emerges when $p$ is small enough to pretend asynchronous evolution. Tangling phase where edges of a graph circulate frequently among low degree vertices occurs when $p$ is large. There exist conditions at which the resulting stationary network ensemble provides networks which degree distribution exhibit power-law decay in large interval of degrees.

Motivation & Objective

• To model the self-organization of complex networks through a stochastic, dynamic rewiring process.
• To investigate how microscopic rewiring rules influence macroscopic network properties such as degree distribution and clustering.
• To explore the emergence of stationary network ensembles with specific topological features under different evolution parameters.
• To identify conditions under which scale-free networks with power-law degree distributions arise from static lattice evolution.
• To analyze the role of synchronization and information update in shaping network structure and phase transitions.

Proposed method

• The network evolves via a two-step process: preferential rewiring of edges based on vertex degree, followed by updating of connectivity information.
• The algorithm uses a parameter $ p $ to control the fraction of edges rewired per time step, with $ T $ determining the update frequency of connectivity data.
• Preferential rewiring favors high-degree vertices, mimicking preferential attachment, while delayed information update introduces a synchronization effect.
• The system is initialized as a regular square lattice and evolves over discrete time steps until a stationary state is reached.
• Simulations track vertex degree distribution, second moment, maximal degree, and structural features like graph skeleton and leaf formation.
• Phase transitions are identified by analyzing changes in degree distribution shape and structural organization across parameter space.

Experimental results

Research questions

• RQ1 How does preferential rewiring with delayed information update affect the emergence of stationary network ensembles?
• RQ2 What conditions lead to the formation of scale-free networks with power-law degree distributions in a static lattice evolution model?
• RQ3 How do the parameters $ p $ and $ T $ influence the topological transition between leafy and tangling phases?
• RQ4 What is the role of synchronization in the self-organization of network structure and degree distribution?
• RQ5 Can a mechanism resembling preferential attachment emerge in a non-growing network through dynamic rewiring?

Key findings

• For $ p \leq 0.01 $, the network self-organizes into a 'leafy phase' characterized by a dense graph skeleton with high-degree hubs and numerous attached leaves.
• For $ p > 1.0 $, the system enters a 'tangling phase' where edges frequently circulate among low-degree vertices, resulting in exponential degree distributions.
• The second moment of the degree distribution and the maximal vertex degree peak at $ p \approx 0.2 $, indicating a critical transition point between phases.
• When $ p $ is tuned appropriately, the degree distribution exhibits power-law decay over a wide range of degrees with exponent $ \gamma \approx 2 $, indicating scale-free structure.
• The transition between phases is driven by the breakdown of multi-connections in the graph skeleton, which reduces degeneracy and enables effective edge accumulation in hubs.
• The emergence of power-law behavior suggests that a rich-get-richer mechanism operates via effective edge accumulation, even in a non-growing network framework.