A universally applicable approach to connectivity percolation

TL;DR

This paper introduces a universal, parameter-free method to estimate percolation thresholds across various systems by mapping percolation problems onto branching processes, providing rigorous lower bounds and accurate predictions.

Contribution

It presents a general, first-principles approach to determine percolation thresholds applicable to diverse systems, improving upon system-specific and simulation-based methods.

Findings

Provides rigorous lower bounds for percolation thresholds

Achieves accurate predictions with minimal effort

Applicable to continuum and network percolation problems

Abstract

Percolation problems appear in a large variety of different contexts ranging from the design of composite materials to vaccination strategies on community networks. The key observable for many applications is the percolation threshold. Unlike the universal critical exponents, the percolation threshold depends explicitly on the specific system properties. As a consequence, theoretical approaches to the percolation threshold are rare and generally tailored to the specific application. Yet, any percolating cluster forms a discrete network the emergence of which can be cast as a graph problem and analyzed using branching processes. We propose a general mapping of any kind of percolation problem onto a branching process which provides rigorous lower bounds of the percolation threshold. These bounds progressively tighten as we incorporate more information into the theory. We showcase our approach for different continuum problems finding accurate predictions with almost no effort. Our approach is based on first principles and does not require fitting parameters. As such it offers an important theoretical reference in a field that is dominated by simulation studies and heuristic fit functions.