The theory of percolation on hypergraphs

Ginestra Bianconi,Sergey N. Dorogovtsev
2023-11-11
Abstract:Hypergraphs capture the higher-order interactions in complex systems and always admit a factor graph representation, consisting of a bipartite network of nodes and hyperedges. As hypegraphs are ubiquitous, investigating hypergraph robustness is a problem of major research interest. In the literature the robustness of hypergraphs as been so far only treated adopting factor-graph percolation which describe well higher-order interactions which remain functional even after the removal of one of more of their nodes. This approach, however, fall short to describe situations in which higher-order interactions fail when anyone of their nodes is removed, this latter scenario applying for instance to supply chains, catalytic networks, protein-interaction networks, networks of chemical reactions, etc. Here we show that in these cases the correct process to investigate is hypergraph percolation with is distinct from factor graph percolation. We build a message-passing theory of hypergraph percolation and we investigate its critical behavior using generating function formalism supported by Monte Carlo simulations on random graph and real data. Notably, we show that the node percolation threshold on hypergraphs exceeds node percolation threshold on factor graphs. Furthermore we show that differently from what happens in ordinary graphs, on hypergraphs the node percolation threshold and hyperedge percolation threshold do not coincide, with the node percolation threshold exceeding the hyperedge percolation threshold. These results demonstrate that any fat-tailed cardinality distribution of hyperedges cannot lead to the hyper-resilience phenomenon in hypergraphs in contrast to their factor graphs, where the divergent second moment of a cardinality distribution guarantees zero percolation threshold.
Physics and Society,Disordered Systems and Neural Networks,Statistical Mechanics,Social and Information Networks
What problem does this paper attempt to address?
The paper is primarily dedicated to addressing the percolation theory problem on hypergraphs and exploring the similarities and differences between it and the percolation theory on factor graphs. Specifically: 1. **Research Background**: - Hypergraphs can capture higher-order interactions in complex systems and can be represented as a bipartite network—factor graphs. - The percolation theory on factor graphs has a certain research foundation in describing higher-order interactions, but this theory does not apply to all situations when nodes are removed. 2. **Core Issues**: - When a node in a hypergraph is removed, all hyperedges connected to it will fail. However, in a factor graph, removing a node does not cause the factor nodes connected to it to completely fail. - The paper focuses on higher-order interactions that fail when any one node is damaged, such as supply chains, catalytic networks, protein interaction networks, etc. 3. **Research Objectives**: - Develop a hypergraph percolation theory suitable for the aforementioned second type of higher-order interactions. - Demonstrate the differences between hypergraph percolation and factor graph percolation and analyze these differences through message-passing algorithms. 4. **Main Findings**: - The node percolation threshold is higher on hypergraphs than on factor graphs. - On hypergraphs, the node percolation threshold is not the same as the hyperedge percolation threshold, and the node percolation threshold is greater than the hyperedge percolation threshold. - These results indicate that any hyperedge with a long-tail distribution cannot lead to the hyper-resilience phenomenon in hypergraphs, whereas in factor graphs, a long-tail distribution guarantees a zero percolation threshold. Through this research, the paper aims to fill the gap in the existing theory regarding the understanding of percolation behavior for specific types of higher-order interactions.