Sharp cluster-size decay and logarithmic phase boundaries in Pareto-weighted spatial random graphs
Year: 2026
Type: preprint
Abstract: We study supercritical kernel-based spatial random graphs with Pareto vertex weights on Poisson and lattice vertex sets, a framework containing continuum scale-free percolation, geometric inhomogeneous random graphs and the age-dependent random connection model. When long edges strictly dominate surface order, we determine the sharp decay of the finite-cluster size distribution and the order of the second-largest component in a box. In the high–high regime we prove the conjectured exponent ζ_hh on both vertex sets, including the range τ < σ + 1 where the previously known upper exponent did not match; at (d, α, τ, σ) = (2, 2, 5/2, 2) it improves from 4/9 to 2/3. At ties between long-edge mechanisms we identify the logarithmic corrections; on the high–low/high–high tie the correction is (log k)^(b/α) with b/α < 1, so the number of tied mechanisms does not determine it. At high density, deterministic threshold connections satisfy matching rooted and free-box surface-order laws, and randomly thinned ones matching rooted laws. On the lattice, local seeds exist at every strictly supercritical parameter with a positive long-edge exponent, including retention one and dimension one.
Author: Mishra, Shardul
Language: English
Subfield: Mathematical Physics
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7228195777
