Griffiths inequalities for the $O(N)$-spin model

This paper proves Griffiths inequalities for the $O(N)$-spin model with inhomogeneous couplings and external magnetic fields for any $N \geq 2$ by utilizing a random path representation and an identity analogous to the switching lemma for random currents.

The Gist

Imagine you are in a giant, crowded dance hall. The dancers are spins (tiny magnets), and they are trying to decide which way to face. In this specific dance hall, the dancers can face any direction in a multi-dimensional space (not just North/South or East/West, but any angle). This is the O(N)-spin model.

The paper by Benjamin Lees is about proving a very specific rule about how these dancers influence each other.

The Big Question: Do Dancers "Stick Together"?

In physics, we often ask: If I force one dancer to face North, does it make it more likely that their neighbor also faces North?

For a long time, we knew this was true for simple dancers (the Ising model, where they can only face North or South). But for complex dancers who can face any direction (like the O(N) model), proving this "stickiness" (called Griffiths inequalities) was a massive mathematical headache, especially when the dance floor is messy (different strengths of connection between different dancers) and there's a wind blowing (an external magnetic field).

Lees finally cracked the code for any number of dimensions ($N \ge 2$).

The Secret Weapon: The "Random Path" Metaphor

To solve this, Lees didn't look at the dancers directly. Instead, he imagined the dance floor as a network of invisible strings or paths.

1. The Strings (Random Paths): Imagine that every time two dancers interact, a string is laid down between them.

   • Some strings are loops (they start and end at the same dancer).
   • Some strings are walks (they start at one dancer and end at another).
   • Crucially, these strings come in N different colors.

2. The Rules of the Dance:

   • Color 1 is special. It represents the direction we are interested in (like "North").
   • Colors 2 through N are the other directions.
   • The rule is: You can have a walk (an open string) of any color, but you can only have "loose ends" (unpaired strings) for Color 1. All other colors must form perfect loops.

The Magic Trick: The "Switching Lemma"

The core of the paper is a clever mathematical trick called the Switching Lemma. Think of it like a game of musical chairs with these colored strings.

• The Setup: Imagine you have two separate dance groups. Group A has a loose string ending at dancer $X$. Group B has a loose string ending at dancer $Y$.
• The Switch: Lees shows that you can mathematically "swap" parts of these strings. You can take the path from Group A and the path from Group B, cut them, and re-attach them in a new way.
• The Result: After the switch, you still have valid configurations, but now the "loose ends" might have moved.

Why is this important?
This switching trick proves that the probability of finding a loose string at $X$ and a loose string at $Y$ is always higher than the probability of finding them separately.

In plain English: If the dancers are connected by these invisible strings, they are more likely to agree with each other than to disagree. This confirms that the system is "ferromagnetic" (tending to align).

The "Ghost" Trick for Wind (External Fields)

The paper also handles the case where there is an "external wind" (a magnetic field) blowing on the dancers, trying to push them in a specific direction.

Lees uses a clever metaphor here: Ghost Dancers.

• He imagines invisible "ghost" dancers standing outside the main hall.
• If the wind is blowing North, he connects every real dancer to a "North Ghost" with a string.
• This turns the problem of "wind" into a problem of "boundary conditions" (dancers connecting to the edge of the room).
• This allows him to use the same string-switching magic trick even when the wind is blowing.

The Takeaway

Before this paper, we had to guess or use complicated, case-by-case math to prove that these complex magnetic systems behave "nicely" (that they align).

Lees provided a universal instruction manual (the Random Path Model and the Switching Lemma) that works for:

• Any number of dimensions ($N$).
• Any messy arrangement of connections (inhomogeneous couplings).
• Any wind blowing on the system (external fields).

In summary: The paper proves that in a complex magnetic dance, the dancers are naturally inclined to move in sync. It does this by translating the problem from "spinning magnets" to "tangled colored strings," and then showing that you can untangle and retangle those strings in a way that mathematically guarantees they will stick together.

Technical Summary

1. Problem Statement

The paper addresses a long-standing open problem in statistical mechanics: the proof of Griffiths inequalities for the O(N)-spin model (also known as the classical Heisenberg model for $N=3$ and XY model for $N=2$) for arbitrary $N \geq 2$.

• Context: Griffiths inequalities (specifically the first and second inequalities) are fundamental correlation inequalities originally proven for the Ising model ($N=1$). They state that spin correlations are non-negative and that correlations increase with coupling strength. These inequalities are crucial for proving the existence of infinite-volume limits, monotonicity of spontaneous magnetization, and comparing critical temperatures across different models.
• Gap: While Griffiths inequalities were established for the Ising model and extended to the XY model ($N=2$) and Heisenberg model ($N=3$) under specific conditions (e.g., homogeneous couplings), a general proof for the O(N)-spin model with $N > 4$ and inhomogeneous parameters (coupling constants varying by edge, spin direction, and external field) was missing.
• Goal: The author aims to prove the Griffiths inequality for the correlation of the first spin component ($S^1$) for any $N \geq 2$ on an arbitrary finite simple graph, allowing for:
  • Inhomogeneous ferromagnetic coupling constants ($J^i_e \geq 0$) that can vary by edge and spin direction.
  • Inhomogeneous external magnetic fields.
  • General boundary conditions (free and "+").

2. Methodology

The proof relies on a probabilistic representation of the spin model using random paths (walks and loops), rather than traditional analytical methods.

A. The Random Path Model (RPM)

The author utilizes a representation of the O(N)-spin model introduced in previous work (Lees-Taggi, 2020), which generalizes the random current representation of the Ising model.

• Configuration Space: The model is mapped to a system of colored links on a graph. Links are assigned one of $N$ colors (corresponding to spin components).
• Structure: A configuration consists of:
  1. Links: Multiple edges between vertices.
  2. Colorings: Each link is assigned a color $i \in \{1, \dots, N\}$.
  3. Pairings: At each vertex, links of the same color are paired up to form continuous paths (loops or walks).
• Key Distinction: Unlike the Ising case ($N=1$) where pairings cancel out with vertex weights, for $N > 1$, the pairings do not cancel. This introduces combinatorial complexity that must be handled carefully.
• Connection to Spins: The correlation functions of the spin model (e.g., $\langle \prod S^1_x \rangle$) are shown to be equivalent to the partition functions of specific subsets of these random path configurations (specifically, configurations where only color-1 links have unpaired endpoints).

B. The Switching Lemma Analogue

The core of the proof is a new combinatorial identity, Lemma 3.1, which serves as an analogue to the famous Switching Lemma used in the random current representation of the Ising model.

• The Switching Lemma (Ising): Allows the "switching" of sources (vertices with odd current) between two configurations, proving that the sum of products of correlations equals the product of sums.
• The Challenge for $N > 1$: Because pairings are not cancelled by the vertex weight function $U^{(N)}$, one cannot simply switch sources. The author must account for the specific set of pairings that allow paths to connect sources without violating the parity constraints of the other colors.
• The Solution: The author constructs a bijection (a one-to-one mapping) between two sets of configurations:
  1. Pairs of configurations $(m, \bar{m})$ where $m$ has sources at set $A$ and $\bar{m}$ has sources at set $B$.
  2. Pairs where the combined configuration has sources at $A \cup B$ and the second configuration has no sources, but with a restriction on the pairings (specifically, no path connects a vertex in $A$ to a vertex in $B$).
• Geometric Interpretation: The proof involves a geometric manipulation of the "paths" and "vertices" in the configuration space.
  • Case $N$ even: The proof uses a specific ordering of pairs at each vertex to define a bijection that moves "walks" ending in $B$ from one configuration to the other.
  • Case $N$ odd: The proof is more complex due to half-integer factors in the Gamma functions of the vertex weights. It requires a more elaborate labeling scheme for vertices to maintain the bijection.

C. Extension to External Fields

The methodology is extended to include external magnetic fields by introducing "ghost vertices" ($g_1, \dots, g_N$).

• An external field $h^i_x$ on vertex $x$ is modeled as an edge connecting $x$ to a ghost vertex $g_i$.
• This allows the external field to be treated mathematically as a boundary condition, preserving the structure of the random path model.

3. Key Contributions

1. Generalization of Griffiths Inequalities: The paper provides the first proof of Griffiths inequalities for the O(N)-spin model for any $N \geq 2$ (including $N > 4$) with inhomogeneous couplings and external fields.
2. New Combinatorial Identity (Lemma 3.1): The author derives a switching lemma analogue for $N > 1$. This identity handles the non-trivial pairing constraints of the O(N) model, which was the primary obstacle in previous attempts.
3. Refined Random Path Representation: The paper presents a slightly modified version of the random path model (using "pre-colored" links) that simplifies the notation and proof structure compared to previous formulations, while remaining equivalent to the original model.
4. Handling Boundary Conditions: The proof rigorously handles both free and "+" boundary conditions, as well as the extension to external fields via ghost vertices.

4. Main Results

The central result is Theorem 1.1, which states:
For a finite simple graph $G$, $N \geq 2$, and non-negative coupling constants $J^i_e$:
$$\left\langle \prod_{x \in A} S^1_x \prod_{y \in B} S^1_y \right\rangle_{\eta} \geq \left\langle \prod_{x \in A} S^1_x \right\rangle_{\eta} \left\langle \prod_{y \in B} S^1_y \right\rangle_{\eta}$$
where $\eta \in \{\text{free}, +\}$ denotes the boundary condition.

Corollary 1.2 (Monotonicity):
The correlation functions are monotonically increasing with respect to the coupling constants:
$$\frac{\partial}{\partial J^1_e} \left\langle \prod_{x \in A} S^1_x \right\rangle_{\eta} \geq 0$$

These results hold for arbitrary graphs, inhomogeneous couplings (varying by edge and spin direction), and inhomogeneous external fields.

5. Significance

• Theoretical Foundation: This work resolves a significant gap in the theory of spin systems, extending the powerful toolkit of correlation inequalities to a much broader class of models.
• Applications: The established inequalities are essential for:
  • Proving the existence of the infinite volume limit for correlation functions in the O(N) model.
  • Establishing the monotonicity of spontaneous magnetization (critical for understanding phase transitions).
  • Comparing critical temperatures between models with different spin dimensions or spatial structures.
• Methodological Impact: The derived switching lemma analogue (Lemma 3.1) is expected to have further applications in the study of O(N) models, potentially aiding in proofs of sharpness of phase transitions or continuity of magnetization, similar to how the Ising switching lemma was used in recent years.
• Universality: By handling inhomogeneous parameters, the results apply to disordered systems and models on complex networks, not just regular lattices.

In summary, Benjamin Lees successfully bridges the gap between the well-understood Ising model and higher-dimensional O(N) models by adapting the random path representation and developing a novel combinatorial switching lemma that overcomes the complexities introduced by higher spin dimensions.