Abstract
This study applies scaling theories developed for one-dimensional and infinite-dimensional systems to finite-dimensional percolation, where two dimensions represent the lowest dimension with non-trivial behavior. We introduce effective numerical methods to measure the cluster number density $n(s,p)$ and develop a scaling theory to characterize its behavior near the critical threshold $p_c$. Through data-collapse plots, we demonstrate an efficient method for measuring critical exponents and derive expressions for the spanning cluster density $P$ and the average cluster size $S$. Finally, we establish scaling relations between exponents and define theoretical bounds for their values.
Introduction
Percolation theory examines the connectivity of networks, such as fluids moving through porous media. While one-dimensional chains and infinite-dimensional Bethe lattices allow for exact solutions due to the absence of loops, physical systems in 2D and 3D are significantly more complex. In these finite dimensions, clusters contain loops and multiple redundant paths, making exact mathematical solutions impossible. This research addresses the challenge of characterizing these systems near their critical point $p_c$—the threshold where a globally connected network emerges—using universal scaling laws rather than exhaustive enumeration.
Methodology: Numerical Simulation and Binning
To analyze 2D percolation, we utilize Monte Carlo simulations on $L \times L$ square lattices. Clusters are identified and labeled using standard algorithms. A significant challenge in this analysis is the rarity of large clusters, which are vital for understanding the physics near the critical point.
Standard linear histograms often fail because bins for large cluster sizes remain mostly empty, leading to high statistical noise. To overcome this, we implement logarithmic binning. By increasing bin widths exponentially, we ensure that each bin contains enough samples to provide a reliable estimate of the cluster number density $n(s,p)$ across several orders of magnitude.
Scaling Theory and Data Collapse
The core of our theoretical approach is the scaling ansatz for the cluster number density: $$n(s, p) = s^{-\tau} F((p - p_c)s^\sigma)$$ This model assumes that near $p_c$, the system's behavior is dominated by power laws and a single characteristic cluster size $s_\xi$. To verify this, we use the "data collapse" method: plotting $s^\tau n(s, p)$ against the scaled variable $(p - p_c)s^\sigma$ should cause curves from different occupation probabilities $p$ to fall onto a single master curve if the scaling theory is correct.
Results and Discussion
Our simulations confirm the validity of the scaling ansatz in 2D. As the occupation probability $p$ approaches $p_c$, the distribution of cluster sizes follows a power law, and the characteristic size $s_\xi$ diverges, signaling the onset of global connectivity.
We determined the universal critical exponents for 2D percolation to be approximately $\tau \approx 2.05$ and $\sigma \approx 0.4$. These values satisfy the theoretical bounds derived from the requirement that physical quantities like average cluster size $S$ diverge at the critical point while the spanning cluster density $P$ remains finite. Specifically, we find that $2 \le \tau < 3$ must hold. Furthermore, these exponents are linked by scaling relations such as $\gamma = (3 - \tau) / \sigma$ and $\beta = (\tau - 2) / \sigma$.
Conclusion
By bridging the gap between simplified models and finite-dimensional reality, we demonstrate that scaling theory provides a robust framework for understanding percolation. While finite-size effects on $L=200$ or $512$ lattices introduce some numerical limitations, the fundamental universal behavior is clearly recovered. Future work will explore more advanced finite-size scaling techniques to refine exponent estimates and test the robustness of these results across different lattice geometries and boundary conditions.
Summary updated