Correlation Length: 1D vs. Mean-Field

Introduction

In our last lesson, we established conclusively that the 1D Ising model lacks a finite-temperature phase transition. We saw this manifest in two ways: through a direct analysis of the exact solution, which showed a finite correlation length $ξ$ for any temperature $T > 0$ , and through the powerful domain-wall argument, which demonstrated that entropy always wins over energy in one dimension.

This lesson builds directly on that foundation. Our objective is to contrast the behavior of the correlation length in the 1D model with the mean-field prediction for higher dimensions. You have seen that order is fragile in 1D; now we will explore how mean-field theory (MFT), an approximation that becomes more accurate in higher dimensions, captures the emergence of robust, long-range order. This contrast is not merely a comparison of two formulas; it goes to the heart of why dimensionality is a decisive factor in critical phenomena and sets the stage for understanding why more sophisticated theories are needed.

1. Mean-Field Theory and the "Error Correction" of Higher Dimensions

Recall the "game of telephone" analogy from the previous lesson: in one dimension, a single error (a flipped spin) propagates unimpeded, destroying information (correlation) down the line. In higher dimensions, a spin has multiple neighbors. As Professor Susskind described, this provides a "support system" or an "error-correcting" influence that resists thermal fluctuations and helps maintain order.

Mean-Field Theory is a powerful approximation that formalizes this idea. Its central assumption is to replace the complex, fluctuating interactions of a spin with its neighbors by a single, non-fluctuating, average interaction with a "mean field." This effectively assumes that the number of neighbors is large enough that their individual fluctuations average out. Given your experience with simulating many-body systems like bubble rafts and sand dunes, you can appreciate this as a classic coarse-graining step, where microscopic details are replaced by an effective, average behavior.

Statistical Mechanics Lecture 9

Let's revisit Professor Susskind's intuitive explanation for why higher dimensions support a phase transition, which serves as the physical motivation for the mean-field approach.

Please watch the segment from this discussion on how a higher number of neighbors provides an "error correction" mechanism. This leads directly into the rationale for the mean-field approximation, which is valid in the limit of many neighbors (high dimensionality).

By making this approximation, the interacting many-body problem is reduced to a solvable one-body problem of a single spin in an effective magnetic field. As we'll see, this simplified model predicts a phase transition at a finite temperature $T_{c}$ for dimensions $d > 1$ .

2. The Mean-Field Correlation Length

A key signature of a phase transition is the divergence of the correlation length, $ξ$ . To see how this plays out in MFT, we need to calculate the spin-spin correlation function, $C (r) = ⟨ σ_{i} σ_{i + r} ⟩ - ⟨ σ_{i} ⟩ ⟨ σ_{i + r} ⟩$ , within this framework.

The full derivation involves considering the system's response to a small, spatially varying magnetic field, which is a standard technique in linear response theory. While we won't reproduce every algebraic step, the lecture notes below provide a concise and rigorous derivation that your mathematical background should make quite accessible.

[PDF] Chapter 5 Ising model and phase transitions

These notes derive the mean-field correlation function and extract its behavior near the critical point.

In the document, find the subsection "Correlation function" (page 70). Read from the introductory paragraph to understand the setup. Follow the derivation which uses a Fourier transform to solve the linearized self-consistency equation. The crucial result is Equation (5.33), which gives the correlation function in Fourier space. Pay close attention to the approximation made for large distances r, which leads to Equation (5.34). This Ornstein-Zernike form shows an exponential decay modulated by a power law. The key is to identify the term that defines the decay length.

From the reading, we see that the correlation function at large distances behaves as $C (r) \sim e^{- r / ξ_{M F T}}$ . The derivation shows that the correlation length $ξ_{M F T}$ has a very specific temperature dependence near the critical point $T_{c}$ .

The analysis reveals that the MFT correlation length diverges as:

ξ_{M F T} (t) \propto ∣ t ∣^{- ν_{M F T}} with ν_{M F T} = \frac{1}{2}

Here, $t = (T - T_{c}) / T_{c}$ is the reduced temperature, and $ν$ is the critical exponent for the correlation length. This power-law divergence at a finite critical temperature is the hallmark of a continuous phase transition, a behavior fundamentally absent in the 1D model.

3. A Tale of Two Models: The Contrast

We can now place the behavior of the correlation length in the 1D exact solution and the mean-field approximation side-by-side. The comparison starkly reveals the impact of dimensionality.

Feature	1D Ising Model (Exact)	Mean-Field Theory (Approximate for d > 1)
Critical Temp. $T_{c}$	$T_{c} = 0$	$T_{c} = J z / k_{B} > 0$
Correlation Length $ξ (T)$	$ξ_{1 D} = - \frac{1}{l n ( t a n h ( β J ))}$	$\xi_{MFT} \propto
Behavior at T > 0	$ξ_{1 D}$ is always finite.	$ξ_{M F T}$ diverges as $T \to T_{c}$ .
Decay of Correlations	Always exponential, $e^{- r / ξ_{1 D}}$ .	Exponential away from $T_{c}$ , becomes a power-law $r^{- (d - 2)}$ at $T = T_{c}$ .
Physical Implication	Order is always short-range. No phase transition.	Long-range order emerges at $T_{c}$ . A phase transition occurs.

The exact solution in 1D shows us a world where thermal fluctuations are always strong enough to prevent macroscopic ordering. The correlation length, while growing as temperature drops, only reaches infinity at the unattainable point of absolute zero.

Mean-field theory, by averaging over fluctuations, paints a different picture. It describes a world where, below a certain temperature $T_{c}$ , the "error-correcting" effect of having multiple neighbors is strong enough to establish order across the entire system. The divergence of $ξ$ at $T_{c}$ is the mathematical signature of this collective phenomenon: at the critical point, fluctuations are correlated across all length scales.

4. Reconciling the Views: The Ginzburg Criterion

We are left with a critical question: the 1D model is exact, and MFT is an approximation. We know MFT fails for $d = 1$ , but is it correct for $d = 2$ , $d = 3$ , or higher?

The failure of MFT stems from its core assumption: neglecting fluctuations, i.e., approximating $σ_{i} σ_{j}$ by terms involving only the average magnetization $⟨ σ ⟩$ . This assumption breaks down catastrophically near the critical point, where the divergence of $ξ$ tells us that fluctuations become long-ranged and highly correlated.

The Ginzburg criterion provides a formal test for the validity of mean-field theory. It compares the magnitude of fluctuations within a correlated volume (a region of size $ξ^{d}$ ) to the mean value of the order parameter itself. MFT is self-consistent only when the fluctuations are small compared to the mean.

Phase Transitions & The Ising Model - Ryan Peters

This article gives an excellent conceptual summary of the Ginzburg criterion and the idea of the upper critical dimension.

Please read the section titled "The Upper Critical Dimension". Focus on the core idea: MFT breaks down when fluctuations within a correlated volume become comparable to the order parameter itself. The derivation shows this leads to a condition on the spatial dimension d.

The Ginzburg criterion for the Ising model with short-range interactions yields a remarkable result:

d > 4

This defines an upper critical dimension, $d_{c} = 4$ .

For dimensions $d > 4$ , fluctuations are sufficiently suppressed that they become irrelevant near the critical point. Mean-field theory becomes exact, and its predicted critical exponents (like $ν = 1/2$ ) are correct.
For dimensions $d \leq 4$ , fluctuations dominate the physics near $T_{c}$ . Mean-field theory still correctly predicts that a phase transition occurs (for $d > 1$ ), but its quantitative predictions for critical exponents are wrong.

For instance, Onsager's exact solution for the 2D Ising model ( $d = 2$ ) finds $ν_{2 D} = 1$ , which differs from the mean-field value of $ν_{M F T} = 1/2$ . This discrepancy is a direct consequence of the fluctuations that MFT neglects. This is precisely the physics that the Renormalization Group was designed to handle.

Conclusion

In this lesson, we have juxtaposed the exact behavior of the 1D Ising model with the predictions of mean-field theory to understand the pivotal role of dimensionality in phase transitions.

Key Takeaways:

1D Model: The correlation length $ξ_{1 D}$ is finite for all non-zero temperatures, diverging only at $T = 0$ . This confirms the absence of a phase transition and the persistence of short-range order.
Mean-Field Theory: In contrast, $ξ_{M F T}$ diverges with a power-law at a finite critical temperature $T_{c}$ , with a critical exponent $ν = 1/2$ . This divergence signals the onset of long-range order and a continuous phase transition.
Role of Dimensionality: The difference is rooted in dimensionality. Higher dimensions provide a cooperative environment that stabilizes order against thermal fluctuations, a feature captured qualitatively by MFT.
Limits of MFT: MFT's neglect of fluctuations renders it quantitatively inaccurate for dimensions $d \leq 4$ , as revealed by the Ginzburg criterion. The breakdown of MFT near the critical point highlights the need for a more powerful theoretical framework.

This exploration has set the stage for a more general study of critical phenomena. Having seen how different models yield different exponents, our next step is to abstract these ideas. In the next module, we will formally classify phase transitions, define the full set of critical exponents, and introduce the scaling hypothesis, which posits universal relationships between them.

Recommended resources