🔢 mathematics

Universal correlations in the Abelian sandpile model

This paper presents a large-scale numerical study of bulk correlation functions in the two-dimensional Abelian sandpile model across square, honeycomb, and kagome lattices using Wilson's algorithm and the Majumdar–Dhar bijection, confirming logarithmic conformal field theory predictions and providing the first systematic analysis for the kagome lattice alongside a new analytical derivation for its height-1 probability.

Original authors: Qiyu Liu, Jan-Niklas Herre, Prajit Baruah, Sebastian Dreizler, Christoph Karrasch, Wioletta Ruszel, Dirk Schuricht

Published 2026-09-10

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Qiyu Liu, Jan-Niklas Herre, Prajit Baruah, Sebastian Dreizler, Christoph Karrasch, Wioletta Ruszel, Dirk Schuricht

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quiet mathematics of nature, there is a phenomenon where systems seem to find their own critical point without anyone telling them to. Imagine a pile of sand. If you keep dropping grains on it, the pile grows until it becomes unstable, triggering an avalanche that reshapes the whole structure. In the real world, this happens with earthquakes, forest fires, and even solar flares. Scientists call this self-organized criticality, a state where a system naturally settles into a delicate balance where small changes can cause effects of any size. To study this, researchers use a simplified computer model called the Abelian sandpile. It is a grid of sites, each holding a certain number of grains. When a site gets too full, it topples, sending grains to its neighbors, which might then topple in turn. This chain reaction continues until the system stabilizes. The question that has long fascinated physicists is how the height of the sand at one spot is related to the height at another spot far away. In a critical system, these connections do not fade away quickly; instead, they follow a specific, universal pattern that depends only on the dimension of the space, not on the tiny details of the grid.

For decades, scientists have been able to predict these patterns perfectly for a simple square grid, using advanced theories that treat the system like a fluid with special mathematical properties. However, nature is rarely so simple. Real materials and landscapes often have more complex shapes, like the honeycomb pattern of a beehive or the intricate triangles of a kagome lattice, which looks like a woven mesh of stars. For these complex shapes, the mathematical tools used for the square grid often break down, leaving researchers with only guesses about how the sand should behave. They knew the general rules should be the same, but they lacked the numbers to prove it. This uncertainty created a gap between the elegant theory and the messy reality of different geometric structures.

A team of researchers has now bridged this gap by running massive computer simulations on three different types of grids: the square, the honeycomb, and the kagome. Instead of letting the sandpile evolve grain by grain, which is slow and creates messy, repetitive data, they used a clever trick from graph theory. They generated random, branching trees that connect every point on the grid to a central sink, much like a river network flowing to the sea. They then translated these trees directly into sandpile configurations. This method allowed them to create millions of independent, perfectly random snapshots of the sandpile, free from the statistical noise that usually plagues such studies. With this clean data, they could measure exactly how the height of the sand at one point influenced the height at another point across vast distances.

The results confirmed the deep universality of the phenomenon. On all three very different grids, the connection between distant points faded away at exactly the same rate. The researchers found that the influence of one spot on another drops off according to a precise power law, meaning that if you double the distance, the connection weakens by a specific, predictable factor. This held true whether the grid was a simple square, a honeycomb, or the complex kagome pattern. The data showed that the underlying physics does not care about the specific shape of the grid; the critical behavior is the same everywhere. This is a strong validation of the theoretical framework that describes these systems, proving that the rules of self-organized criticality are robust enough to survive changes in geometry.

Beyond confirming the general rule, the team also measured the specific numbers that define how the sand behaves on each grid. For the square grid, where the answers were already known from complex math, their new simulations matched the old predictions almost perfectly, serving as a rigorous check on their methods. For the honeycomb grid, they found that the known predictions for the simplest case were correct, but they also provided the first detailed measurements for more complex interactions, filling in a missing piece of the puzzle. The most significant breakthrough came with the kagome lattice. Until this study, almost nothing was known analytically about the sandpile on this specific shape. The researchers not only measured the probabilities of finding different sand heights but also derived a new, exact mathematical formula for the most basic probability on this lattice, a result that had never been written down before. They then used their simulations to verify this new formula, finding a match so close that the difference was less than one part in a million.

The study also looked at how different heights of sand interact with each other. They found that a low, stable pile of sand tends to suppress the formation of other low piles nearby, while encouraging the formation of the highest, most stable piles. This pattern held true across all three grids, suggesting a universal mechanism where the system organizes itself to avoid instability. The researchers were able to extract the specific coefficients that describe these interactions for the honeycomb and kagome lattices, numbers that were previously unknown. These values act as fingerprints for each lattice, showing how the local geometry tweaks the universal behavior. While the general scaling law remained identical, the specific strength of the connections varied slightly depending on the grid, a nuance that only high-precision simulation could reveal.

This work demonstrates that the laws governing self-organized criticality are far more universal than previously demonstrated. By combining a sophisticated sampling technique with the power of modern computing, the researchers were able to peer into the statistical heart of these complex systems. They showed that even when the geometry becomes intricate and the mathematical tools fail, the system still obeys the same fundamental rules. The findings provide a solid foundation for understanding how criticality emerges in natural systems with complex structures, from the branching of lightning to the flow of traffic. The team has not only confirmed what was suspected for the simplest cases but has also charted new territory for complex geometries, providing the first systematic map of how sandpiles behave on these unique lattices. Their work leaves no doubt that the critical state is a robust feature of nature, persisting regardless of the shape of the stage on which it plays out.

Technical Summary: Universal Correlations in the Abelian Sandpile Model

Problem and Context
The Abelian Sandpile Model (ASM) is a paradigmatic example of self-organized criticality (SOC), where open systems evolve spontaneously to a critical stationary state without fine-tuning external parameters. The scaling limit of the two-dimensional ASM is predicted to be described by a logarithmic conformal field theory (LCFT) with central charge $c = -2$ . While exact analytical results for bulk correlation functions exist for the square lattice, derived via spanning-tree representations and graph-theoretic methods, these techniques rely heavily on specific lattice symmetries and known Green functions. For more complex geometries, such as non-Bravais lattices (e.g., honeycomb and kagome), analytical derivations of bulk observables are significantly more difficult, and scaling-field coefficients remain largely unknown. Furthermore, standard numerical approaches based on Markov-chain dynamics suffer from strong autocorrelations, requiring massive sample sizes ( $\sim 10^9$ – $10^{10}$ ) to achieve statistical precision, particularly for two-point correlation functions in the bulk.

Methodology
To address the challenges of sampling and analytical intractability, the authors employ a hybrid numerical-analytical approach:

Sampling Strategy: The study bypasses the correlated Markov-chain dynamics of the sandpile entirely. Instead, it utilizes Wilson's algorithm to generate large-scale uniform spanning trees (USTs) in parallel. These trees are mapped to recurrent sandpile configurations via the Majumdar–Dhar burning bijection. This method produces independent samples, eliminating sample autocorrelations and allowing for fast convergence. The authors report that $10^6$ independent samples on an $8000 \times 8000$ lattice suffice for well-converged results, a significant reduction compared to previous studies.
Numerical Implementation: Correlation functions are computed using a Fast Fourier Transform (FFT)-based convolution theorem to efficiently evaluate two-point correlations over the bulk of the lattice. The analysis focuses on one-point height probabilities ( $P_a$ ) and connected two-point correlation functions ( $C_{a,b}(r)$ ) for heights $a, b \in \{1, 2, 3, 4\}$ .
Analytical Derivation: For the kagome lattice, where analytical results were previously scarce, the authors derive a closed-form expression for the bulk height-1 probability ( $P_1$ ) using the lattice Green function and a defect-matrix approach. This involves calculating specific Brillouin-zone integrals to determine relative Green function values required for the defect matrix determinant.
Lattices Studied: The analysis covers three planar lattices with varying structural complexity: the square lattice (Bravais, $z=4$ ), the honeycomb lattice (non-Bravais, two sublattices, $z=3$ ), and the kagome lattice (non-Bravais, three sublattices, $z=4$ ).

Key Contributions and Results

Universal Scaling Exponent: Across all three lattices, the numerical data confirms the universal power-law decay of the height-1 correlation function, $C_{1,1}(r) \sim r^{-4}$ , corresponding to a scaling exponent $2\Delta = 4$ . This validates the universality of the LCFT prediction independent of lattice structure.
Square Lattice (Benchmark): The numerical results for height probabilities and correlation amplitudes agree with exact analytical predictions to within $\sim 10^{-6}$ for probabilities and $\sim 12\%$ for correlation amplitudes. The study successfully reconstructs the $C_{4,1}$ correlation via sum rules, confirming the predicted positive slope and matching the exact theoretical value to within 0.5%.
Honeycomb Lattice: The study reproduces exact height probabilities ( $P_1=1/12, P_2=7/24, P_3=5/8$ ) with high precision. While exact analytical amplitudes for mixed correlations are not available, the numerical data confirms the expected functional forms: $C_{a,1}(r)$ exhibits linear dependence on $\log r$ , and $C_{2,2}(r)$ exhibits quadratic dependence on $\log r$ , consistent with the LCFT framework.
Kagome Lattice: This work provides the first systematic numerical study of bulk correlations on the kagome lattice.
- Analytical Contribution: The authors derive the exact bulk height-1 probability: $P_1^{\text{kag}} = \frac{85}{1296} + \frac{7\sqrt{3}}{432\pi} - \frac{1}{24\pi^2} \approx 0.070298$ .
- Numerical Findings: The numerical $P_1$ matches the derived analytical value to within $10^{-6}$ . The three sublattices (A, B, C) are found to be statistically indistinguishable in their height distributions and correlation functions. The correlation functions follow the predicted LCFT asymptotic forms (constant, linear, and quadratic in $\log r$ for different height pairs).
LCFT Amplitudes: The study extracts lattice-dependent expansion parameters ( $\alpha_a, \beta_a$ ) for the honeycomb and kagome lattices. While the exact values are unknown for these lattices, the extracted coefficients are of comparable magnitude to the exact square-lattice values, and consistency checks (e.g., $A_{22} \approx A_{21}^2/A_{11}$ ) support the validity of the LCFT decomposition.

Significance and Claims
The paper claims that its primary significance lies in providing a robust numerical framework to probe universality and lattice-dependent coefficients in regimes where analytical methods are difficult to apply. By eliminating autocorrelations through Wilson's algorithm, the authors achieve high statistical precision with feasible computational resources, enabling the study of complex non-Bravais lattices like the kagome.

Specifically, the authors claim:

Validation of Universality: The $r^{-4}$ decay of $C_{1,1}(r)$ is confirmed across square, honeycomb, and kagome lattices, reinforcing the universality of the $c=-2$ LCFT description.
First Systematic Study of Kagome: The work offers the first comprehensive numerical characterization of bulk correlations on the kagome lattice, filling a gap in the literature where few bulk observables were known analytically.
Analytical Advancement: The derivation of the closed-form expression for the kagome height-1 probability via the lattice Green function represents a novel analytical result for this geometry.
Methodological Efficiency: The combination of Wilson's algorithm and FFT-based correlation evaluation is demonstrated to be a superior alternative to Markov-chain dynamics for studying bulk correlations in the ASM, yielding cleaner data and faster convergence.

The authors remain modest regarding the precision of higher-order amplitude extractions for the kagome and honeycomb lattices, noting that larger system sizes or more samples would be required to disentangle asymptotic amplitudes from sub-leading corrections with higher precision. They do not propose new experimental applications but suggest future theoretical directions, such as extending the analysis to triangular lattices, studying higher-point correlations, and deriving further analytical bulk amplitudes.

More like this