|
|
| Line 1: |
Line 1: |
| '''Percolation theory''' is the mathematical study of how connected clusters form in random networks as the probability of link formation increases. At a critical probability — the '''[[Percolation Threshold|percolation threshold]]''' — the system undergoes a [[Phase Transition|phase transition]]: isolated clusters suddenly merge into a single giant component that spans the entire network, and global communication becomes possible.
| | - |
| | |
| The canonical model is bond percolation on a lattice: each edge is present with independent probability p. Below the threshold p_c, only finite clusters exist. Above p_c, an infinite cluster appears with probability 1. The transition is sharp: the fraction of nodes in the giant component jumps from zero to a finite value discontinuously in the thermodynamic limit. The mathematics was developed by Broadbent and Hammersley (1957) and has since been extended to continuum percolation, directed percolation, and percolation on complex networks. The [[Erdős–Rényi Model|Erdős–Rényi random graph]] G(n, p) can be viewed as bond percolation on the complete graph, with its giant component threshold at p = 1/n serving as a percolation threshold. Near this critical point, component growth is described by a [[Branching Process|branching process]], making the Erdős–Rényi model one of the most analytically tractable settings for studying percolation phenomena.
| |
| | |
| In [[Network Science|network science]], percolation provides the baseline model for robustness: random node removal is equivalent to site percolation, and the percolation threshold tells us how much damage a network can sustain before it fragments. But real-world failures are rarely random — they target hubs, and targeted attack destroys connectivity far more efficiently than random percolation. The gap between percolation predictions and cascade dynamics reveals that real networks fail through load redistribution, not random deletion.
| |
| | |
| [[Category:Systems]]
| |
| [[Category:Mathematics]]
| |
