From Landau to Renormalization: Understanding Critical Phenomena

Introduction

In our last lesson, we derived the Ginzburg criterion, which provided a stark verdict on the limits of mean-field theory. We established that for systems in dimensions $d < 4$ , thermal fluctuations inevitably dominate in a region close to the critical temperature $T_{c}$ , causing the mean-field approximation to fail catastrophically. The divergence of the correlation length, $ξ \to \infty$ , signals that the neat separation between microscopic and macroscopic scales, upon which Landau theory is built, breaks down. At a critical point, the system is correlated across all length scales simultaneously.

This final lesson of our course addresses the profound question that arises from this failure: what comes next? You will learn how the very nature of this breakdown does not represent a dead end, but rather provides the crucial motivation for the Renormalization Group (RG), the modern and powerful framework for understanding critical phenomena. We will explore how the failure of standard perturbation theory to "correct" Landau theory forces us to adopt a new perspective—one that embraces the concept of scale invariance rather than fighting it. This lesson will serve as a conceptual bridge from the phenomenological theories you have mastered to the foundational ideas that underpin much of contemporary condensed matter and statistical physics.

The Breakdown of Perturbation Theory

The Ginzburg criterion tells us that treating the order parameter as a uniform field—the core of mean-field theory—is untenable near $T_{c}$ . A natural next step for a theorist would be to try to improve the theory by systematically accounting for fluctuations using perturbation theory. The Landau-Ginzburg functional, with its quadratic (Gaussian) and quartic ( $ϕ^{4}$ ) terms, provides a perfect setting for this. We can treat the Gaussian part as an exactly solvable "free" theory and the $ϕ^{4}$ interaction term as a small perturbation.

The central question is whether this perturbative approach can fix the incorrect mean-field critical exponents. The answer, as you will see, is a resounding no. The attempt to "correct" the theory leads to its own, more profound, failure.

The following video lecture from MIT OpenCourseWare provides an exceptional walk-through of this calculation. It calculates the first-order correction to the susceptibility due to the $u ϕ^{4}$ term and analyzes its behavior near the critical point.

9. Perturbative Renormalization Group Part 1

This video, "Perturbative Renormalization Group Part 1" from the course on Statistical Physics in Biology, rigorously demonstrates why a standard perturbative expansion fails near a critical point.

First, watch the setup and calculation of the first-order correction to the two-point correlation function (susceptibility) from this segment. The key is to see how the correction term, which is proportional to the interaction strength u, involves an integral over all fluctuation modes. Next, focus on the crucial analysis of this correction from this point onwards. The lecturer shows that while the correction shifts the critical temperature, its behavior as t = (T-T_c)/T_c \to 0 is highly problematic. Pay close attention to the dimensional analysis and the final conclusion about the divergent nature of the perturbation series for d<4.

As the video demonstrates, the attempt to calculate corrections to a quantity like the susceptibility, $χ$ , reveals a fatal flaw. The correction term is not simply proportional to the "small" coupling constant $u$ . Instead, the expansion parameter is effectively a dimensionless quantity that behaves like $u ξ^{4 - d}$ . Since the correlation length $ξ$ diverges as $t \to 0$ , this expansion parameter blows up for any $d < 4$ , no matter how small we make $u$ .

This is a divergent perturbation series. It tells us that the fluctuations are so strong that they cannot be treated as a small correction to a mean-field background. The interaction term fundamentally alters the nature of the critical state. From a field theory perspective, this issue arises from what are known as infrared (IR) divergences. These are divergences in loop integrals (like the one you saw calculated) that come from long-wavelength, small-momentum modes. At a critical point, where the mass term of the field goes to zero, these modes are exactly the ones that dominate, invalidating the entire perturbative framework.

As the following text succinctly puts it, this is the core issue.

[PDF] Critical Phenomena, Phase Transitions and Statistical Field Theory

This section from a set of lecture notes by R. R. Horgan at DAMTP Cambridge provides the formal field-theoretic diagnosis of Landau theory's failure.

In Section 4, "The Partition Function and Field Theory", find the part that analyzes the loop correction to the mass term. Read the paragraph starting with the analysis of the integral for dimensions D>4 versus D<4. This explicitly shows how the integral becomes sensitive to low-momentum modes (an "Infra-Red divergent" behavior) for D<4, making it inconsistent with the basic assumptions of Landau's expansion.

The Renormalization Group Idea

The failure of perturbation theory implies we need a non-perturbative method, or at least a much more sophisticated one. This is where Kenneth Wilson's profound insight comes in: instead of trying to suppress or ignore the fluctuations at all scales, the Renormalization Group (RG) provides a mathematical apparatus to systematically account for them.

The core idea is to analyze how the description of the system changes as we change our observation scale. This is done through an iterative procedure that "integrates out" degrees of freedom, moving from short scales to long scales. An intuitive way to picture this is through Kadanoff's block-spin transformation:

Group: Group the microscopic spins on a lattice into blocks.
Average: Define a new "block spin" for each block, for example, by taking the majority rule of the microscopic spins within it.
Rescale: Find the new effective Hamiltonian that describes the interactions between these block spins.

This process effectively reduces the number of degrees of freedom and looks at the system at a coarser level of detail. The RG formalizes this into a momentum-space procedure, which generally consists of two steps:

Coarse-Graining: In the functional integral for the partition function, integrate out the high-momentum (short-wavelength) field modes. This leaves an effective theory for the remaining low-momentum modes.
Rescaling: Rescale momenta, lengths, and the field itself so that the new effective Hamiltonian can be compared to the original one. For instance, momenta are rescaled to fill the original momentum shell, and lengths are rescaled to restore the original lattice spacing.

The combination of these two steps forms the RG transformation, $R$ . Applying this transformation repeatedly generates a trajectory, or RG flow, in the space of all possible Hamiltonians (parameterized by their coupling constants like $t, u, K, \dots$ ).

Landau Ginzburg theory of Phase Transitions

Fixed Points, Universality, and Scaling

The RG flow provides a powerful geometric picture for understanding critical phenomena.

Fixed Points: A critical point is characterized by scale invariance—the system looks the same at all length scales. In the RG framework, this means that the Hamiltonian describing the system must be a fixed point ( $H^{*}$ ) of the RG transformation. Applying the transformation leaves the Hamiltonian unchanged: $R (H^{*}) = H^{*}$ . The system ceases to "flow" under coarse-graining.
Universality: Consider different physical systems with different microscopic Hamiltonians. They may start at different points in the space of couplings. However, as the RG transformation is applied, their trajectories might flow towards the same fixed point. This explains the phenomenon of universality: systems with vastly different microscopic details exhibit identical critical exponents because their long-distance behavior is governed by the properties of the same RG fixed point. The flow washes away the "irrelevant" details of the microscopic interactions.
Critical Exponents and Scaling: The behavior of the RG flow near a fixed point determines the critical exponents. Couplings are classified as relevant, irrelevant, or marginal based on how they behave under the RG flow away from the fixed point. To reach a critical point, one must tune the relevant parameters (like temperature) to specific values. The way the system flows away from the fixed point as a relevant parameter is detuned from its critical value dictates the scaling laws and the values of the critical exponents.

The following text provides a more formal introduction to these central concepts of the RG.

[PDF] Critical Phenomena, Phase Transitions and Statistical Field Theory

This reading introduces the general formalism of the RG transformation, the concept of a flow in coupling space, and the ultimate emergence of scaling laws at a fixed point.

Please read Section 4.5, "The Renormalization Group in General". Start by understanding the two key steps: thinning and rescaling. Follow the logic that leads to the definition of the RG transformation R(\lambda) on the couplings, which generates a flow described by the beta function: \frac{dg_i}{db} = \beta_i(g). Then, focus on the definition of a fixed point and how it leads to scaling laws for Green's functions. This is where concepts like the anomalous dimension \eta_\phi arise naturally, correcting the naive "engineering dimensions" of fields and explaining why mean-field exponents are wrong.

Conclusion

This lesson, and with it our course, concludes at the threshold of one of the most powerful ideas in modern physics. We have traced the path from the fundamental postulates of statistical mechanics to the sophisticated phenomenology of Landau theory, and finally, to the very reasons for its limitations.

Key Takeaways:

The breakdown of Landau theory near a critical point for $d < 4$ is not just a failure of a simple approximation, but a signal of new, profound physics: fluctuations on all length scales are inextricably coupled.
This coupling of scales manifests as a divergent perturbation series when one tries to "correct" mean-field theory by perturbatively including interaction terms like $u ϕ^{4}$ . The expansion parameter is not small near $T_{c}$ .
The Renormalization Group (RG) was conceived to solve this problem by systematically analyzing how the effective description of a system changes under a change of scale (coarse-graining and rescaling).
This leads to a picture of RG flow in the space of Hamiltonians. Fixed points of this flow correspond to scale-invariant critical points.
The RG framework naturally explains universality (different systems flow to the same fixed point) and provides the machinery to calculate critical exponents from the properties of the flow near a fixed point.

You began this course seeking a foundational "reset" to bridge the gap toward advanced theoretical research. You have now arrived at that bridge. The failure of Landau theory is not an end, but the beginning of a deeper understanding, one that is built upon the beautiful and powerful concepts of the Renormalization Group. You are now fully equipped to tackle a dedicated study of RG methods, the essential tool for any theorist working in condensed matter, high-energy physics, or statistical mechanics.

Recommended resources

Introduction

The Breakdown of Perturbation Theory

The Renormalization Group Idea

Fixed Points, Universality, and Scaling

Conclusion