Marked GUE-corners process in doubly periodic dimer models

This paper establishes that the fluctuations of periodically weighted Aztec diamond dimer models near their turning points, as the system size tends to infinity, converge to a marked GUE-corners process where independent Bernoulli marks encode the model's periodicity, a result proven via a double-contour integral representation of the inverse Kasteleyn matrix on a higher-genus Riemann surface.

The Gist

The Big Picture: A Tiling Puzzle with a Secret Pattern

Imagine you have a giant, diamond-shaped floor made of tiles. You want to cover this floor completely with dominoes (two-tile blocks) so that no gaps are left and no dominoes overlap. This is called a dimer model.

In the simplest version of this puzzle, every domino is identical, and the floor is perfectly uniform. Mathematicians have known for a long time that if you look at the edges of this tiled floor, the way the dominoes wiggle and fluctuate follows a very famous, predictable pattern called the GUE-corners process. Think of this as the "standard rhythm" of the puzzle.

But this paper asks a "What if?" question:
What if the floor isn't uniform? What if the dominoes have a secret, repeating pattern? For example, maybe every second row has a slightly different weight, or the tiles change color in a cycle.

The authors of this paper studied a specific type of patterned floor (a "doubly periodic" Aztec diamond). They wanted to know: Does the secret pattern disappear when the floor gets huge, or does it leave a permanent scar on the edge?

The Discovery: The "Marked" Rhythm

Their answer is surprising and beautiful.

1. The Scale: As the floor gets infinitely large, the "wiggles" of the dominoes at the very edge (the turning point) still look like the standard GUE-corners rhythm.
2. The Twist: However, the pattern doesn't disappear. Instead, it survives as a hidden tag or a mark on every single domino.

Imagine the standard rhythm is a song played on a piano. The authors found that in their patterned version, every note in that song is now stamped with a tiny, invisible label (either "Red" or "Blue").

• If you ignore the labels and just listen to the music, it sounds exactly like the standard song.
• But if you look closely at the labels, you see a specific, repeating sequence that matches the original floor pattern.

They call this the "Marked GUE-corners process." It's the standard rhythm, but with a memory of its origins attached to every single particle.

The Analogy: The Crowd at a Concert

To visualize this, imagine a massive crowd of people (the dominoes) at a concert.

• The Uniform Case: If everyone is wearing the same plain white shirt, and they start swaying to the music, the movement of the crowd at the very edge of the venue follows a specific, smooth wave pattern (the GUE-corners process).
• The Patterned Case (This Paper): Now, imagine the crowd is arranged in a specific pattern: every other person is wearing a red shirt, and the rest are wearing blue.
• The Result: As the concert gets huge, the swaying motion of the crowd at the edge still looks like that same smooth wave. BUT, if you zoom in, you realize the wave is made of two distinct groups. The "Red" people and the "Blue" people sway together, but they are distinct. The mathematical "wave" now carries a tag: "I am Red" or "I am Blue."

The paper proves that this "tag" (the mark) is mathematically precise. It tells you exactly how likely a particle is to be "Red" or "Blue" based on where it is in the wave.

How They Did It: The Mathematical Telescope

How do you prove something about an infinitely large, patterned floor? You can't count the dominoes.

The authors used a powerful mathematical tool called a Double-Contour Integral.

• The Analogy: Imagine trying to hear a whisper in a hurricane. You can't just listen with your ears. You need a special microphone that filters out the wind and amplifies the whisper.
• In math terms, they built a "microscope" using complex geometry (specifically, a shape called a Riemann surface, which is like a multi-layered map). They used this to zoom in on the "turning point" of the floor (the corner where the frozen, rigid part of the pattern meets the liquid, wiggly part).

They found that even though they were zooming in on a microscopic scale, the "periodicity" (the repeating pattern of the weights) didn't wash away. Instead, it transformed into those "marks" on the particles.

Why Does This Matter?

1. Universality: In physics, we often look for "universal laws"—things that happen regardless of the specific details. This paper shows that even when you add complex, repeating details (periodicity), the system still finds a universal rhythm (GUE-corners).
2. New Mechanism: It reveals a new way nature preserves information. Usually, when things get huge, small details get lost. Here, the small details (the pattern) survive by becoming "tags" on the big picture.
3. Thinning: The paper also shows that if you only look at the "Red" dominoes, you get a "thinned" version of the standard rhythm (a rhythm where some notes are randomly missing). This connects their work to other areas of probability theory.

Summary

The authors took a complex, patterned tiling puzzle and proved that as it grows to infinity, the chaos at the edge settles into a famous, predictable rhythm. However, unlike the simple version, this rhythm carries a permanent memory of the original pattern in the form of invisible "marks" on every particle.

It's like finding out that even in a chaotic, infinite ocean, every wave still remembers exactly which ripple it came from.

Technical Summary

1. Problem Statement and Context

The paper investigates the asymptotic behavior of doubly periodic Aztec diamond dimer models near their turning points.

• Background: The Aztec diamond is a classic model in statistical mechanics and combinatorics. In the uniformly weighted case, the local fluctuations at the "turning points" (where the arctic curve touches the boundary) are known to converge to the GUE-corners process (a determinantal point process related to the eigenvalues of principal sub-matrices of a Gaussian Unitary Ensemble matrix).
• The Gap: While the uniform case is well-understood, the behavior of models with doubly periodic edge weights (specifically, weights that vary periodically in both horizontal and vertical directions) near turning points was not fully characterized.
• The Question: Does the microscopic periodicity of the edge weights survive the macroscopic scaling limit at the critical turning point, or does it wash out to the classical GUE-corners process?

2. Methodology

The authors employ a rigorous analytical approach combining algebraic combinatorics, complex analysis, and asymptotic analysis on Riemann surfaces.

A. Model Setup

• Graph: The Aztec diamond of size $2\ell N$ is considered.
• Weights: Edge weights are periodic with period $\ell$ in the horizontal direction and period $2$ in the vertical direction. The weights are parameterized by $\alpha_i$ and $\beta_i$.
• Particle System: The dimer cover is mapped to an interlacing particle system. Black vertices adjacent to South or West dimers are identified as particles. Due to the vertical period-2, these particles are assigned a "color" (mark) $j \in \{0, 1\}$ based on the parity of their $y$-coordinate.

B. Correlation Kernel Representation

• Inverse Kasteleyn Matrix: The authors utilize a double-contour integral representation of the inverse Kasteleyn matrix (and thus the correlation kernel $K_{Int}$) derived in previous work [3].
• Riemann Surface: The integral is defined on a higher-genus Riemann surface $\mathcal{R}$ (genus $g = \ell - 1$). The surface is constructed by gluing two copies of the complex plane along cuts determined by the discriminant of the characteristic polynomial.
• Action Function: The asymptotic analysis relies on an action function $G(z, w)$ defined on $\mathcal{R}$. The turning point corresponds to a specific point $q_\infty$ on the surface where the differential $dG$ has a simple zero.

C. Asymptotic Analysis (Steepest Descent)

The core of the proof involves analyzing the double-contour integral as $N \to \infty$ using the method of steepest descent:

1. Contour Deformation: The integration contours are deformed to pass through the saddle point $q_\infty$.
2. Local Expansion: Near $q_\infty$, the authors perform a local expansion of the integrand. They establish that the action function behaves quadratically ($G(z) \approx G(q_\infty) + \frac{\sigma^2}{2}(z-1)^2$).
3. Gauge Transformation: A specific gauge function $g$ is introduced to normalize the kernel.
4. Separation of Terms: The kernel is split into a dominant term ($J_d$, the double integral) and a sub-dominant term ($J_s$, a single integral). The authors prove that $J_s$ vanishes or becomes negligible in the limit for the relevant scaling regime.
5. Limiting Kernel: The dominant term converges to a double integral representation of the GUE-corners kernel, but multiplied by a factor dependent on the particle's mark.

3. Key Contributions and Results

A. The Marked GUE-Corners Process

The primary result is the identification of a new limiting object: the Marked GUE-Corners Process.

• Definition: This process is constructed by taking a standard GUE-corners process and independently assigning a binary mark $j \in \{0, 1\}$ to each particle $(t, \mu)$ with probability $\theta(t)$.
• The Marking Function: The probability $\theta(t)$ is explicitly determined by the edge weights of the model:
  $$\theta(t) = \frac{\alpha_{\ell+1-t}\beta_{\ell-t}^{-1}}{1 + \alpha_{\ell+1-t}\beta_{\ell-t}^{-1}}$$
  This function reflects the horizontal periodicity ($\ell$) and the specific position $t$ relative to the turning point.

B. Main Theorems

• Theorem 1.1 (Kernel Convergence): The correlation kernel of the colored interlacing particle system, under the scaling $x \sim 2\ell N$ and $y \sim 2N\tau + 2\sigma\sqrt{N}\mu$, converges to:
  $$\nu(t_2, j_2) \sigma^{-1} K_{GUE}(t_1, \sigma^{-1}\mu_1; t_2, \sigma^{-1}\mu_2)$$
  where $\nu(t, j) = \theta(t)\delta_{1,j} + (1-\theta(t))\delta_{0,j}$. The factor $\nu$ cannot be removed by a gauge transformation, proving the limit is distinct from the classical GUE-corners process.
• Theorem 1.2 (Weak Convergence): The rescaled particle system converges in distribution to the Marked GUE-corners process.
• Corollary 1.3 (Thinning): If one restricts the process to only one color (e.g., only $j=1$), the limit is a thinned GUE-corners process, where particles are deleted with probability $1-\theta(t)$.

C. Parameters $\tau$ and $\sigma^2$

The authors provide explicit formulas for the scaling parameters:

• $\tau$ (Location of the turning point): Determined by the edge weights and the geometry of the spectral curve.
• $\sigma^2$ (Variance of fluctuations): Derived from the second derivative of the action function at the turning point $q_\infty$.

4. Significance and Implications

1. Persistence of Microscopic Structure: The paper demonstrates that periodicity does not vanish in the critical scaling limit at turning points. Unlike bulk limits where periodicity often averages out, the turning point retains a "trace" of the microscopic lattice structure in the form of marks (colors) on the particles.
2. Universality Class: It establishes a new universality class for dimer models. While the classical GUE-corners process is universal for uniform weights, the Marked GUE-corners process is the universal limit for this class of periodic models.
3. Connection to Random Matrices: The result bridges the gap between discrete periodic dimer models and continuous random matrix theory, showing how discrete symmetries manifest as specific probabilistic structures (marking/thinning) in the continuous limit.
4. Methodological Advance: The use of double-contour integrals on higher-genus Riemann surfaces provides a robust framework for analyzing doubly periodic models, which are significantly more complex than the uniform or singly periodic cases.

5. Conclusion

Berggren and Bradinoff prove that for doubly periodic Aztec diamond dimer models, the local fluctuations at the turning point are governed by a Marked GUE-corners process. This limit "remembers" the vertical periodicity through independent Bernoulli marks assigned to particles, with the marking probability determined by the horizontal periodicity and the specific location in the scaling limit. This work enriches the understanding of critical phenomena in statistical mechanics by revealing how microscopic periodicity can survive and shape macroscopic critical limits.