Finite Size Scaling

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

Abstract

In this chapter, we introduce the theory of finite size scaling (FSS) and demonstrate how we can apply the theory to improve our measurements of the properties of percolation clusters. Usually, we attempt to measure properties of percolation systems in the largest possible system we can simulate. Here, we demonstrate that if the system behaves according to simple scaling relations, it is instead much better to systematically vary the system size and interpolate to infinite system sizes. This approach is generally called finite size scaling. We provide a thorough introduction to the theory and its applications to understand the scaling of the density of the spanning cluster, $P(p,L)$, the average cluster size, $S(p,L)$, and the percolation probability $\Pi(p,L)$.

Introduction

In computational physics, the ultimate goal is often to understand the behavior of systems in the "thermodynamic limit"—that is, for an infinitely large system. However, numerical simulations are fundamentally limited by finite memory and processing time. In the study of percolation, critical properties like the spanning cluster density $P$ or the average cluster size $S$ follow power laws near a critical threshold $p_c$. In an infinite system, these properties exhibit sharp transitions or divergences. In a finite system of size $L$, these transitions are "smeared out" because the correlation length $\xi$—the characteristic distance over which the system's components are connected—is constrained by the system boundaries.

Finite Size Scaling (FSS) is a powerful technique that turns this limitation into an advantage. By systematically studying how system properties change with $L$, we can accurately extrapolate infinite-system behavior and extract critical exponents.

Methodology

The FSS workflow involves simulating site percolation across multiple grid sizes (e.g., $L = 25, 50, 100, 200$). The core of the method is the scaling ansatz, a mathematical hypothesis suggesting that a physical property $M$ near the critical point depends on the ratio of the system size to the correlation length:

$$M(p, L) = L^{x/\nu} f(L/\xi)$$

where $x$ is a critical exponent, $\nu$ is the correlation length exponent, and $f$ is a universal scaling function. To validate this ansatz, researchers use data collapse. By plotting rescaled variables—such as $L^{\beta/\nu}P$ against $L^{1/\nu}(p-p_c)$—data from different system sizes should fall onto a single universal curve. If the curves collapse, the scaling theory and the chosen exponents are confirmed.

Results and Discussion

The spanning cluster density $P(p, L)$ clearly demonstrates the impact of finite system size. As shown in Figure 1, the transition becomes sharper as $L$ increases. At the critical point $p_c$, the density follows a power-law decay $P(p_c, L) \propto L^{-\beta/\nu}$.

Spanning cluster density $P(p, L)$ for different system sizes (a) and its power-law scaling at the critical point (b).

The "aha moment" of FSS occurs when we apply the scaling transformation. When the raw data from Figure 1 is rescaled according to the scaling ansatz, the disparate curves for different $L$ values collapse onto a single, universal scaling function. This collapse, shown in Figure 2, proves that the underlying physics is consistent across scales.

Data collapse of the spanning cluster density $P(p, L)$ onto a universal curve using the scaling ansatz.

Furthermore, FSS provides a robust method for estimating the percolation threshold $p_c$. By identifying an effective threshold $p_x(L)$ for each size and plotting it against $L^{-1/\nu}$, the intercept at $L \to \infty$ yields a high-precision estimate of the true $p_c$. Even with relatively small and noisy simulations, this method can produce results remarkably close to theoretical values (e.g., $p_c \approx 0.5935$ for 2D site percolation).

Conclusion

Finite Size Scaling is more than a workaround for finite computational resources; it is a fundamental tool for identifying universal behavior in complex systems. By shifting focus from a single "large" simulation to a suite of multi-scale simulations, researchers can extract precise critical parameters and confirm the validity of scaling theories. While the examples here focus on 2D site percolation, the principles of FSS are broadly applicable across statistical mechanics and condensed matter physics.

Summary updated

Abstract

Abstract In this chapter we will introduce the theory of finite size scaling and demonstrate how we can apply the theory to improve our measurements of the properties of percolation clusters. Usually, we attempt to measure properties of percolation system in the largest possible system we can simulate. Here, we demonstrate that if the system behaves according to simple scaling relations, it is instead much better to systematically vary the system size and the interpolate to infinite system sizes. This approach is generally called finite size scaling and we provide a thorough introduction to the theory and its applications to understand the scaling of the density of the spanning cluster, $$P(p,L)$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>P</mml:mi> <mml:mo>(</mml:mo> <mml:mi>p</mml:mi> <mml:mo>,</mml:mo> <mml:mi>L</mml:mi> <mml:mo>)</mml:mo> </mml:math> , the average cluster size, $$S(p,L)$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>S</mml:mi> <mml:mo>(</mml:mo> <mml:mi>p</mml:mi> <mml:mo>,</mml:mo> <mml:mi>L</mml:mi> <mml:mo>)</mml:mo> </mml:math> , and the percolation probability $$\varPi (p,L)$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>Π</mml:mi> <mml:mo>(</mml:mo> <mml:mi>p</mml:mi> <mml:mo>,</mml:mo> <mml:mi>L</mml:mi> <mml:mo>)</mml:mo> </mml:math> .
BibTeX
@inproceedings{malthesrenssen2024finite,
  author = {Anders Malthe‐Sørenssen and Anders Malthe-Sorenssen},
  title = {{Finite Size Scaling}},
  booktitle = {Lecture Notes in Physics},
  year = {2024},
  doi = {10.1007/978-3-031-59900-2\_6},
}
Finite-Dimensional PercolationLecture Notes in Physics · 2024
Percolation Theory Using PythonLecture Notes in Physics · 2024
One-Dimensional PercolationLecture Notes in Physics · 2024
Introduction to PercolationLecture Notes in Physics · 2024
RenormalizationLecture Notes in Physics · 2024
Geometry of ClustersLecture Notes in Physics · 2024
Figure