Geometry of Clusters

Anders Malthe‐Sørenssen, Anders Malthe-SorenssenLecture Notes in Physics 2024 · DOI · OpenAlex

TL;DR

This work explores the fractal geometry of clusters in percolation theory, moving beyond simple mass distributions to characterize their spatial extent. By defining a non-integer "fractal dimension" ($D \approx 1.89$ for 2D systems), the authors demonstrate how cluster density and connectivity are governed by universal scaling laws and a characteristic "correlation length" ($\xi$) that diverges at the critical threshold.

What Problem Does This Paper Address?

In percolation theory—the study of how clusters form and connect in random media—researchers often focus on the size (mass) of clusters. However, mass alone doesn't describe a cluster's shape or how it fills space. Standard Euclidean geometry fails here: a percolation cluster isn't a solid disk or a simple line; it’s a "holey," branching structure.

The authors address the challenge of quantifying this "holey" geometry. They seek to define how the linear size of a cluster relates to its mass and how the system transitions from a fractal, self-similar state at the critical point ($p_c$) to a homogeneous state further away from it. Understanding this geometry is crucial for predicting physical properties like fluid flow through porous rocks or electrical conductivity in disordered materials.

Methods

The paper combines analytical derivations with computational simulations implemented in Python. Key methodological components include:

  • Radius of Gyration ($R_s$): Used as the primary measure of a cluster's spatial extent, calculated as the standard deviation of the positions of all sites within a cluster.
  • Monte Carlo Simulations: Generating random $L \times L$ lattices to observe cluster formation at various occupation probabilities ($p$).
  • Computational Tools: The authors utilize scipy.ndimage for cluster labeling and numba (specifically the @jit decorator) to optimize the computationally expensive calculation of the pair correlation function.
  • Scaling Arguments: The authors apply "scaling theory" to relate different physical quantities (mass, length, and probability) through power laws, focusing on exponents rather than prefactors.
  • Box Counting: A technique introduced to measure the dimension of fractal objects by seeing how many "boxes" of size $\delta$ are needed to cover the cluster.

Datasets

The study relies on synthetically generated data from Monte Carlo simulations rather than external datasets.

  • Lattice Sizes: Simulations are performed on square lattices ranging from $64 \times 64$ to $800 \times 800$.
  • Parameters: The occupation probability $p$ is varied around the critical threshold ($p_c \approx 0.5927$ for square lattices) to observe the divergence of characteristic lengths.

Key Insights and Ideas

  • Fractal Dimension ($D$): Clusters at the critical point are fractals. In 2D, they have a dimension of $D \approx 1.89$. This means if you double the linear size of a cluster, its mass increases by a factor of $2^{1.89}$ (about 3.7) rather than $2^2$ (4) for a solid object.
  • The Correlation Length ($\xi$): This is the "natural ruler" of the system.
  • At scales smaller than $\xi$, the system looks fractal and "critical."
    • At scales larger than $\xi$, the system looks homogeneous (like a uniform "Swiss cheese").
  • The Density Paradox: For a spanning cluster at $p_c$, the density actually decreases as the system size $L$ increases. This is a hallmark of fractal objects—they become "airier" as they grow.
  • Universality: The critical exponents (like $\nu$ for the correlation length and $D$ for the fractal dimension) are universal. They depend on the dimensionality of the space (2D vs. 3D) but not on the specific lattice geometry (square vs. triangular).

Key Results

  • Mass-Radius Scaling: The relationship $s \propto R_s^D$ was confirmed, with $D = 91/48 \approx 1.89$ for 2D percolation.
  • Divergence of $\xi$: The correlation length was shown to diverge as $p$ approaches $p_c$ following the power law $\xi \propto |p - p_c|^{-\nu}$, where $\nu = 4/3$ in 2D.
  • Correlation Function: The pair correlation function $g(r, p)$—the probability that two sites at distance $r$ belong to the same cluster—follows a power-law decay $r^{-\eta}$ at the critical point, with $\eta \approx 0.208$ in 2D.
  • Spanning Cluster Mass: The mass of the spanning cluster $M(p, L)$ was shown to follow a unified scaling form $M(p, L) = L^D Y(L/\xi)$, which accounts for both finite-size effects and the transition to homogeneity.

Limitations and Future Work

  • Finite Size Effects: In simulations, the system size $L$ is always finite. This "cuts off" the true fractal behavior, as clusters cannot grow larger than the box. The authors note that for $p$ very close to $p_c$, $\xi > L$, making it impossible to distinguish if the system is truly at the critical point.
  • Computational Intensity: Calculating the correlation function $g(r)$ is an $O(N^2)$ operation. While Numba helps, large-scale 3D simulations remain a significant computational challenge.
  • Future Directions: The chapter lays the groundwork for studying the topology of the spanning cluster (e.g., identifying "red links" vs. "blobs") and how this complex geometry dictates the transport of fluids or electricity in the next stages of percolation theory.

Summary updated

Abstract

Abstract We have seen how we can characterize clusters by their mass, s . As p approaches $$p:c$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msub> <mml:mrow> <mml:mi>p</mml:mi> </mml:mrow> <mml:mrow> <mml:mi>c</mml:mi> </mml:mrow> </mml:msub> </mml:math> , the typical cluster size s increases as well as the characteristic cluster diameter. In this chapter we will discuss the geometry of clusters, and by geometry we will mean how the number of sites in a cluster is related to the linear size of the cluster. We will introduce several measures to characterize the spatial extent, the characteristic radius $$R:s$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msub> <mml:mrow> <mml:mi>R</mml:mi> </mml:mrow> <mml:mrow> <mml:mi>s</mml:mi> </mml:mrow> </mml:msub> </mml:math> , of clusters of size s . We will measure $$R:s$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msub> <mml:mrow> <mml:mi>R</mml:mi> </mml:mrow> <mml:mrow> <mml:mi>s</mml:mi> </mml:mrow> </mml:msub> </mml:math> to motivate that it is proportional to $$s^1/D$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msup> <mml:mrow> <mml:mi>s</mml:mi> </mml:mrow> <mml:mrow> <mml:mn>1</mml:mn> </mml:mrow> </mml:msup> <mml:mo>∕</mml:mo> <mml:mi>D</mml:mi> </mml:math> , where D is a new exponent characterizing the dimension of clusters. We will demonstrate that the percolation system is characterized by two lengths, the system size L and a characteristic cluster size $$\xi $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>ξ</mml:mi> </mml:math> , and that the system shows fractal, self-similar scaling when the characteristic length diverges. We develop scaling theories for $$P(s,L)$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>P</mml:mi> <mml:mo>(</mml:mo> <mml:mi>s</mml:mi> <mml:mo>,</mml:mo> <mml:mi>L</mml:mi> <mml:mo>)</mml:mo> </mml:math> for $$p&gt;p_c$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>p</mml:mi> <mml:mo>&gt;</mml:mo> <mml:msub> <mml:mrow> <mml:mi>p</mml:mi> </mml:mrow> <mml:mrow> <mml:mi>c</mml:mi> </mml:mrow> </mml:msub> </mml:math> and lay the foundations for a geometrical understanding and description of the spanning cluster.
BibTeX
@inproceedings{malthesrenssen2024geometry,
  author = {Anders Malthe‐Sørenssen and Anders Malthe-Sorenssen},
  title = {{Geometry of Clusters}},
  booktitle = {Lecture Notes in Physics},
  year = {2024},
  doi = {10.1007/978-3-031-59900-2\_5},
}
Introduction to PercolationLecture Notes in Physics · 2024
Subset GeometryLecture Notes in Physics · 2024
Percolation Theory Using PythonLecture Notes in Physics · 2024
Finite-Dimensional PercolationLecture Notes in Physics · 2024
One-Dimensional PercolationLecture Notes in Physics · 2024
Diffusion in Disordered MediaLecture Notes in Physics · 2024
Figure