[Paper Review] Ultra Small World in Scale-Free Networks

Reuven Cohen, Shlomo Havlin|arXiv (Cornell University)|May 23, 2002

Stochastic processes and statistical mechanics36 citations

TL;DR

This paper analytically demonstrates that scale-free networks with a degree exponent a > 3 exhibit a diameter d ~ ln N, significantly smaller than the d ~ ln N scaling in regular small-world networks. For a > 2, it constructs a deterministic scale-free network with the minimal possible diameter d ~ ln ln N, revealing ultra-small-world behavior in scale-free topology.

ABSTRACT

We study the diameter, or the mean distance between sites, in a scale-free network, having N sites and degree distribution p(k) ~ k^-a, i.e. the probability of having k links outgoing from a site. In contrast to the diameter of regular random networks or small world networks which is known to be d ~ lnN, we show, using analytical arguments, that scale free networks with 2 3, d ~ lnN. We also show that, for any a>2, one can construct a deterministic scale free network with d ~ lnlnN, and this construction yields the lowest possible diameter.

Motivation & Objective

• To understand the scaling of network diameter in scale-free networks with power-law degree distributions.
• To investigate whether scale-free networks can achieve smaller diameters than traditional small-world networks.
• To construct a deterministic scale-free network with the minimal possible diameter for a > 2.
• To establish the theoretical lower bound on diameter in scale-free networks.

Proposed method

• Analytical derivation of diameter scaling using degree distribution p(k) ~ k^(-a) for scale-free networks.
• Application of mean-field and branching process approximations to estimate typical path lengths.
• Construction of a deterministic scale-free network model with controlled degree sequences to minimize diameter.
• Proof that for a > 2, the diameter scales as d ~ ln ln N, achieving the theoretical minimum.

Experimental results

Research questions

• RQ1 How does the diameter of a scale-free network scale with system size N for different values of the degree exponent a?
• RQ2 Can scale-free networks achieve a smaller diameter than classical small-world networks?
• RQ3 What is the minimal possible diameter achievable in a deterministic scale-free network for a > 2?
• RQ4 Is the ln ln N scaling of diameter in scale-free networks a fundamental lower bound?

Key findings

• For scale-free networks with a > 3, the diameter scales as d ~ ln N, similar to classical small-world networks.
• For scale-free networks with 2 < a ≤ 3, the diameter scales as d ~ ln N, indicating a slower growth than in regular random networks.
• For any a > 2, a deterministic scale-free network can be constructed with diameter d ~ ln ln N, achieving the minimal possible diameter.
• The ln ln N scaling represents the theoretical lower bound on diameter in scale-free networks, making them ultra-small-world networks.