Researchers study how random connections affect graph connectivity and movement

A study examines how retaining edges with probability pp affects the geometry and transport of band random graphs.

Researchers studied the connectivity, geometry, and transport of band random graphs. These graphs are random subgraphs of the cycle power CnmC_n^m.

The model retains each edge independently with probability pp. The study looks at the regime where log⁡n≪m<n/2\log n \ll m < n/2.

A sharp threshold for connectivity occurs when p∼(log⁡n)/(2m)p \sim (\log n)/(2m). If n(1−p)2m→λ∈(0,∞)n(1-p)^{2m} \to \lambda \in (0, \infty), the number of isolated vertices converges to Poisson(λ)\mathrm{Poisson}(\lambda).

With high probability, every other vertex lies in a single component C1C_1. This component C1C_1 contains a cyclic chain of connected sets. The diameter has order n/m+log⁡n/log⁡log⁡nn/m + \log n / \log \log n with high probability.

The mean hitting time of a simple random walk satisfies a specific equation. This equation involves the cyclic distance rxyr_{xy}, the degree dG(y)d_G(y), and κm=(m+1)(2m+1)/6\kappa_m = (m+1)(2m+1)/6.

Connectivity follows Erdős--Rényi asymptotics with 2m2m possible neighbours. However, distances, relaxation, and hitting times retain the geometry of the underlying cycle.

Key facts

  • The connectivity threshold is p∼(log⁡n)/(2m)p \sim (\log n)/(2m).
  • The diameter of the main component is of order n/m+log⁡n/log⁡log⁡nn/m + \log n / \log \log n.
  • The mean hitting time HG(x,y)H_G(x,y) depends on the cyclic distance rxyr_{xy} and the degree dG(y)d_G(y).

All sections

Edition