Isoperimetric inequality for nonlocal bi-axial discrete perimeter

Imagine you are a city planner tasked with building a new neighborhood. You have a fixed number of bricks (let's say 100 bricks) to build a single, solid block of houses. Your goal is to arrange these bricks into a shape that is as "compact" as possible.

In the old, classic world of geometry, "compact" simply meant having the shortest possible fence around the outside. If you used 100 bricks, the best shape was a perfect square (10x10), because it had the shortest fence compared to a long, skinny rectangle (1x100). This is the famous Isoperimetric Problem: for a fixed area, the shape with the smallest perimeter is a circle (or a square in a grid world).

But this paper introduces a twist. It asks: What if the "fence" isn't just the outside edge?

The New Rule: The "Telepathic" Neighborhood

In this new world, the bricks don't just care about their immediate neighbors. They have a "long-range connection."

Imagine every brick in your neighborhood can "feel" every other brick in the entire city, even if they are far apart.

If a brick is on the edge of your block, it feels a strong pull from the empty space outside.
But here's the kicker: If there is a gap inside your block (a hole), or if your block is shaped like a "C" or a "U," the bricks on the inside of the curve also feel a pull from the outside world, even though they are surrounded by other bricks.

The "cost" of your shape isn't just the length of the outer fence. It's the sum of all these long-distance "tugs" between the inside of your block and the outside world. The further apart two points are, the weaker the tug, but it never completely disappears.

The authors of this paper asked: Given this new "telepathic" rule, what is the best shape for our 100 bricks?

The Discovery: Squares are Still Kings, but with a Catch

The researchers found that the answer is surprisingly specific, but with a weird quirk:

No Holes Allowed: If you make a shape with a hole in the middle (like a donut), it's a terrible idea. The "telepathic" pull from the inside of the hole makes the energy cost huge. The best shapes must be solid, with no gaps.
No "C" Shapes: If you make a "C" shape, the bricks on the inside of the curve are still "exposed" to the outside world through the gap. This costs extra energy. The best shapes must be "convex" (bulging out, not in).
The Winner is a Square (or close to it): Just like in the old world, a square is the most efficient shape.
The "Protuberance" Rule: What if you have 101 bricks? You can't make a perfect square. You have to add one extra brick.
- In the old world, it didn't matter where you put that extra brick; a square with a bump on the top is the same as a square with a bump on the side.
- In this new world, it matters a lot! The paper proves that if you have a rectangle that is slightly longer than it is wide, you must attach your extra brick to the shorter side.
- Analogy: Imagine a long, skinny loaf of bread. If you add a crouton to the long side, it sticks out far and feels a lot of "wind" from the outside. If you add it to the short side, it's tucked in closer to the main body. The math shows that tucking it in on the short side saves energy.

Why Does This Matter? (The Metastability Connection)

You might wonder, "Who cares about arranging bricks?"

The authors explain that this isn't just a puzzle; it's a key to understanding how magnetic materials behave.

Think of a magnet. Inside, tiny atomic magnets (spins) want to point in the same direction. Sometimes, due to temperature or external forces, they get stuck in a "metastable" state. They are happy, but not perfectly happy. They are like a ball sitting in a small dip on a hillside. They want to roll down to the bottom of the valley (the stable state), but they are stuck.

To escape, they have to form a "nucleus" or a "droplet" of the new state.

In simple magnets (short-range interactions), this droplet is just a square.
In complex, "long-range" magnets (where atoms feel each other from far away), the shape of this droplet is exactly what this paper calculates.

The Big Picture:
This paper solves the "shape puzzle" for these complex magnets. By knowing exactly what shape the "escape droplet" takes (a square with a specific bump on the short side), scientists can finally predict how long it takes for the magnet to switch states and how much energy is needed to make it happen.

Summary in a Nutshell

The Problem: Find the most efficient shape for a cluster of items when every item feels a weak pull from everything outside the cluster, not just its immediate neighbors.
The Solution: The best shapes are solid squares or rectangles. If you have extra items, you must attach them to the shortest side of the rectangle.
The Analogy: It's like building a fort. In a normal world, you just want the shortest wall. In this new world, you also want to minimize the "exposure" of your inner soldiers to the enemy outside, even if they are deep inside the fort.
The Impact: This helps physicists understand how complex magnetic materials switch states, which is crucial for developing better data storage and understanding phase transitions in nature.

The authors didn't just guess; they built a rigorous mathematical "algorithm" (a step-by-step recipe) to prove that any other shape (like a donut, a "C", or a rectangle with a bump on the long side) is strictly worse than the optimal square-with-a-bump-on-the-short-side.

Here is a detailed technical summary of the paper "Isoperimetric inequality for nonlocal bi-axial discrete perimeter" by Vanesa Jacquier, Cristian Spitoni, and Wioletta M. Ruszel.

1. Problem Statement

The paper addresses a nonlocal discrete isoperimetric problem, a generalization of the classical isoperimetric problem (finding the shape with minimal perimeter for a fixed area).

Context: While the classical problem in $\mathbb{R}^n$ is solved by the sphere, and its discrete analog on $\mathbb{Z}^2$ is solved by squares/quasi-squares, this work introduces a nonlocal term.
The Object: The study focuses on polyominoes (finite unions of unit squares on the integer lattice $\mathbb{Z}^2$ ) with a fixed area $n$ .
The Functional: The authors define a nonlocal bi-axial discrete perimeter, denoted $Per_\lambda(P)$ $P e r_{λ} (P)$ , for a polyomino $P$ $P$ and a parameter $\lambda > 1$ $λ > 1$ . Unlike the classical perimeter, which only counts boundary edges, this functional sums the interactions between all sites inside $P$ $P$ and all sites outside $P$ $P$ ( $P^c$ $P^{c}$ ).
- The interaction decays as an inverse power of the distance, specifically weighted by horizontal and vertical distances raised to the power $\lambda$ .
- Mathematically:
  $Per_\lambda(P) := \sum_{x \in P \cap \mathbb{Z}^2} \sum_{y \in P^c \cap \mathbb{Z}^2} \frac{1}{d_\lambda(x,y)}$
  where $d_\lambda$ is defined via horizontal ( $d_H$ ) and vertical ( $d_V$ ) distances:
  $\frac{1}{d_\lambda(x,y)} = \frac{\mathbb{1}_{\{x_1=y_1, x_2 \neq y_2\}}}{|x_2-y_2|^\lambda} + \frac{\mathbb{1}_{\{x_2=y_2, x_1 \neq y_1\}}}{|x_1-y_1|^\lambda}$
Goal: Characterize the set of minimizers of $Per_\lambda(P)$ among all polyominoes with area $n$ , specifically for sufficiently large $\lambda$ .

2. Methodology

The proof of the main theorem relies on a combination of geometric transformations, combinatorial arguments, and analytic estimates involving the Hurwitz-zeta function.

A. Geometric Reduction (The "Main Algorithm")

The authors first establish that minimizers must belong to a specific class of "well-behaved" polyominoes. They prove that:

Connectivity: Disconnected polyominoes always have a larger nonlocal perimeter than their connected counterparts.
Convexity: Concave polyominoes (those with "holes" or disconnected strips in rows/columns) can be transformed into convex ones with strictly lower or equal perimeter by shifting unit squares to fill gaps.
Cross-Convexity: Even among convex polyominoes, "cross-convex" shapes (rectangles with internal rectangular holes or specific non-rectangular convexities) are excluded via an algorithmic transformation that reduces the perimeter.

This process reduces the search space from all polyominoes ( $M_n$ ) to a specific set of extremal polyominoes ( $M_n^{ext}$ ), which includes squares, quasi-squares, and rectangles (with or without protuberances).

B. Analytic Comparison

Once the search space is restricted to $M_n^{ext}$ , the authors perform a rigorous comparison of the nonlocal perimeter values for different shapes:

Squares vs. Rectangles: They prove that for a fixed area $n=l^2$ , the square $Q_l$ has a strictly smaller nonlocal perimeter than any rectangle $R_{a,b}$ (where $a \neq b$ ) provided $\lambda > \lambda_c$ .
Protuberances: They analyze shapes with "protuberances" (strips of unit squares attached to the side). They show that if a square (or quasi-square) and a rectangle with a protuberance have the same area, the square/quasi-square is the minimizer unless their classical perimeters are equal, in which case both are candidates.
Critical Threshold: The authors identify a critical value $\lambda_c \approx 1.78788$ (used as $1.8 $in proofs). For$ \lambda > \lambda_c$, the nonlocal term introduces an anisotropy: protuberances must be attached to the shorter side of a rectangle to minimize energy, unlike the isotropic classical case.

C. Mathematical Tools

Hurwitz-Zeta Function: The perimeter calculations are expressed using the Hurwitz-zeta function $\zeta(\lambda, i) = \sum_{r=0}^\infty (r+i)^{-\lambda}$ .
Monotonicity and Induction: The proofs utilize monotonicity arguments to show that specific functions of the dimensions (e.g., $F_1(a,l)$ ) are strictly positive for $\lambda > 1.8$ .
Algorithmic Proofs: The paper introduces a "Main Algorithm" and an "Algorithm for Cross-Convex P" to iteratively transform any non-optimal polyomino into an optimal one while strictly decreasing the perimeter.

3. Key Contributions and Results

Main Theorem (Theorem 1)

For any fixed area $n \in \mathbb{N}$ and $\lambda > \lambda_c \approx 1.8$ , the set of minimizers for the nonlocal perimeter $Per_\lambda(P)$ consists of:

Squares ( $Q_l$ ) if $n = l^2$ .
Quasi-squares ( $R_{l, l+1}$ ) if $n = l(l+1)$ .
Squares or Quasi-squares with a protuberance attached to the shortest side if $n = l^2 + k$ or $n = l(l+1) + k$ .
Rectangles with protuberances only if they share the same classical perimeter as the square/quasi-square minimizer.

Key Insight: The nonlocal term breaks the symmetry found in the classical case. In the classical isoperimetric problem, a rectangle with a protuberance on the long side is equivalent to one on the short side. In the nonlocal case, attaching the protuberance to the shorter side minimizes the long-range interaction energy.

Characterization of Minimizers

The paper provides a complete classification of the minimizers based on the area $n$ :

If $n = l^2$ , the unique minimizer is the square $Q_l$ .
If $n = l(l+1)$ , the unique minimizer is the quasi-square $R_{l, l+1}$ .
If $n = l^2 + k$ ($0 < k < l $), the minimizers are either the square with a protuberance$ Q_l^k$ or a rectangle with a protuberance, depending on whether their classical perimeters match. If they differ, the square/quasi-square wins.

4. Significance and Applications

Connection to Metastability

The primary motivation for this work is the rigorous study of metastability in the two-dimensional long-range bi-axial Ising model.

Hamiltonian Connection: The energy of a configuration in the long-range Ising model (with external field $h$ ) can be approximated by the nonlocal perimeter plus a bulk term.
$\Delta H_\lambda(\sigma) \approx 2 Per_\lambda(P) - 2h|P|$
Critical Droplets: In the path-wise approach to metastability, the system transitions from a metastable state (all spins down) to a stable state (all spins up) by forming a "critical droplet" of up spins. The shape of this droplet corresponds to the minimizer of the energy functional at a specific magnetization (area).
Impact: By solving the isoperimetric problem, the authors identify the critical droplet shapes for the long-range Ising model. They show that these droplets are not simple squares but can be quasi-squares with protuberances on the shorter side, a feature absent in short-range models.

Novelty

This is the first solution to a nonlocal discrete isoperimetric problem.
It bridges the gap between discrete geometry and statistical mechanics, providing the necessary variational ingredients to analyze the nucleation process in long-range interacting systems.
It demonstrates that long-range interactions induce anisotropy in optimal shapes, even on a square lattice.

5. Conclusion

The paper successfully characterizes the minimizers of a nonlocal bi-axial discrete perimeter. The results reveal that for sufficiently strong long-range interactions ( $\lambda > 1.8$ ), the optimal shapes are squares, quasi-squares, or these shapes with protuberances attached to the shorter side. This characterization is a crucial step toward rigorously solving the metastability problem for the 2D long-range bi-axial Ising model, allowing for the precise determination of critical configurations and transition times in such systems. Future work will focus on constructing the optimal nucleation paths and proving the recurrence properties required for a complete metastability analysis.