Abstract
The percolation problem can be solved exactly in two limits: in the one-dimensional and the infinite dimensional cases. Here, we will first address the one-dimensional system. While the one-dimensional system does not allow us to study the full complexity of the percolation problem, many of the concepts and measures introduced to study the one-dimensional problem can be generalized to higher dimensions.
Introduction
Percolation theory explores the behavior of connected clusters within a random lattice, providing a framework to understand phenomena ranging from material conductivity to the spread of epidemics. While finding exact solutions in two or three dimensions is mathematically challenging, the one-dimensional (1D) case serves as a vital pedagogical tool. In 1D, the system is simplified to its barest essentials, allowing for an exact analytical treatment that introduces universal concepts such as cluster number density, correlation length, and scaling functions.
Methodology: Numerical and Analytical Approaches
The study of 1D percolation employs a dual approach. Analytically, researchers use combinatorics to derive the probability of cluster formation. Numerically, Python-based simulations allow for the generation of random lattices and the empirical measurement of cluster statistics.
A critical step in numerical analysis is the identification of clusters—sequences of adjacent occupied sites. Using tools like scipy.ndimage.measurements, each cluster is labeled and its size is calculated to build a statistical distribution.
The Cluster Number Density
A fundamental measure in percolation is the Cluster Number Density, $n(s, p)$, defined as the probability that a site is a specific member (typically the leftmost) of a cluster of size $s$. In a 1D lattice with occupation probability $p$, this is derived as: $$n(s, p) = (1 - p)^2 p^s$$ This formula reflects the requirement that a cluster of size $s$ must consist of $s$ occupied sites bounded by two empty sites.
Criticality and Scaling
In 1D, the percolation threshold $p_c$ is exactly 1, meaning a spanning cluster only exists when the lattice is entirely full. As $p$ approaches this critical point, the system exhibits power-law behavior characterized by universal exponents. In the 1D case, exponents such as $\sigma, \gamma,$ and $\nu$ are all equal to 1.
One of the most powerful concepts established in this model is data collapse. By rescaling the cluster size $s$ by a characteristic cut-off size $s_\xi$, the distribution curves for various occupation probabilities $p$ all fall onto a single universal scaling function: $$n(s, p) = s^{-\tau} F(s/s_\xi)$$
Key Results
The analytical derivations lead to several key results regarding the system's behavior near the threshold:
- Average Cluster Size ($S$): Diverges as $p \to 1$ following $S = (1+p)/(1-p)$, or $S \propto |p - p_c|^{-1}$.
- Correlation Length ($\xi$): The distance over which sites are likely to belong to the same cluster, derived as $\xi = -1/\ln p$. This also diverges as $|p - p_c|^{-1}$ near the threshold.
Conclusion
While 1D percolation is "anomalous" due to its threshold at $p_c = 1$, it provides the mathematical foundation for understanding phase transitions. The measures and scaling laws developed here are directly applicable—though with greater geometric complexity—to higher-dimensional systems where fractal clusters and non-trivial thresholds emerge.
Summary updated