Quanta Magazine · Wetenschap

‘Stunning’ Percolation Proof Solves Decades-Old Puzzle About Phase Transitions

Mathematicians have developed a groundbreaking proof that resolves a long-standing question in percolation theory, specifically concerning the "supercritical sharpness" conjecture for a broad class of networks. This breakthrough, achieved by a team of five mathematicians, clarifies how a network's behavior abruptly shifts past a critical point, akin to a phase transition.

The proof, developed by Sahar Diskin, Philip Easo, Ritvik Ramanan Radhakrishnan, Benny Sudakov, and Vincent Tassion, addresses how quickly a percolation network floods as the probability of connections increases. Percolation theory, which studies flow in networks, has applications ranging from fluid seepage to virus spread. The core problem involved understanding how large connected areas, or "seas" of fluid, form in graphs (networks of points and lines).

The sharpness conjecture predicted that below a critical probability, fluid pools would be small and isolated, while above it, a single, vast "ocean" would dominate the network. While sharpness was proven for lattices in the 1980s, extending this to more general "transitive graphs" remained a significant challenge for decades. This problem was particularly difficult for the "supercritical" case, dealing with probabilities above the critical threshold.

The team's breakthrough came from an unexpected pivot during their research. Initially focused on a different aspect of percolation, they turned their attention to sharpness. A key insight from Vincent Tassion, combined with a novel application of probability techniques like "sprinkling," allowed them to construct a surprisingly simple yet powerful argument. Their proof demonstrates that for any infinite transitive graph, if the probability of connection is even slightly above the critical threshold, the fluid will cover nearly the entire graph.

This proof is considered "stunning" and "a gem" by experts in the field. It not only solves a decades-old puzzle but also opens doors for applying similar techniques to other complex percolation models and physical systems. However, a major question remains regarding the exact behavior at the critical probability on three-dimensional lattices, which most closely model physical reality.

AI-samenvatting op basis van de bron.