$K$-selective percolation: A simple model leading to a rich repertoire of phase transitions

TL;DR

The paper introduces the $K$-selective percolation model, revealing complex phase transitions and robustness phenomena in networks through theoretical analysis and finite-size scaling.

Contribution

It presents a novel percolation process with diverse phase transitions, including hybrid, reentrant, and tricritical points, expanding understanding of network robustness and cascade effects.

Findings

Non-monotonic giant component response in networks

Presence of hybrid, continuous, and reentrant phase transitions

Identification of a tricritical-like point on Erdős-Rényi networks

Abstract

We propose the $K$-selective percolation process as a model for the iterative removals of nodes with the specific intermediate degree in complex networks. In the model, a random node with degree $K$ is deactivated one by one until no more nodes with degree $K$ remain. The non-monotonic response of the giant component size on various synthetic and real-world networks implies a conclusion that a network can be more robust against such selective attack by removing further edges. In the theoretical perspective, the $K$-selective percolation process exhibits a rich repertoire of phase transitions, including double transitions of hybrid and continuous, as well as reentrant transitions. Notably, we observe a tricritical-like point on Erd\H{o}s-R\'enyi networks. We also examine a discontinuous transition with unusual order parameter fluctuation and distribution on simple cubic lattices, which does not appear in other percolation models with cascade processes. Finally, we perform finite-size scaling analysis to obtain critical exponents on various transition points, including those exotic ones.