Network dismantling based on subgraph growth

Network dismantling is critical for reliability of critical infrastructure such as power grid, communication network and transportation system. Existing researches follow two primary directions: node centrality-based methods and optimization-based frameworks. Node centrality based methods focus on sequentially removing nodes according to their importances in topology or dynamic processes. Network dismantling optimization minimizes macroscopic robustness metrics through node removal combinations. However, integration of these two strategies remains unstudied, which may promote this direction. In this paper, the network dismantling based on subgraph growth (NDSG) is proposed. It models network dismantling as multistage meso‑scale subgraph growth, integrating critical structure destruction and graph partitioning through subgraph boundaries removal. Network dismantling experimental results show that NDSG surpasses GND by 19.2 % and CI by 11.7 % on synthetic networks and exceeds GND by 9.7 % and CI by 73 % on real networks. Our study highlights the importance of subgraph structure in network dismantling and provides a novel perspective on network reliability.