Coarse-Graining and Scale Invariance

Welcome to Module 8. In our previous lesson, we confronted the limits of mean-field theory using the Ginzburg criterion. We discovered that for systems in our three-dimensional world, thermal fluctuations near a critical point are not just a small correction but the dominant physical reality. This breakdown is not a failure, but a signpost pointing toward a more profound theoretical structure. It tells us that to understand critical phenomena, we cannot ignore fluctuations; we must develop a framework that systematically accounts for physics at all length scales.

This lesson introduces that framework: the Renormalization Group (RG). This is not just a calculation technique but a fundamental shift in perspective on how physical laws at different scales relate to one another. Our goal is to understand the two conceptual pillars of the RG: coarse-graining and scale invariance. We will explore how systematically ignoring information at small scales can, paradoxically, reveal universal truths about the large-scale behavior of a system.

From Microscopic Complexity to Macroscopic Simplicity

At the heart of the RG is a question that pervades much of physics: how do the immensely complex interactions of countless microscopic constituents (like atoms or molecules) give rise to the relatively simple, predictable laws we observe at macroscopic scales? The Renormalization Group provides a formal answer to this question by modeling the process of "zooming out."

The following video provides an excellent high-level introduction to what renormalization is and the key questions it helps us answer.

Renormalization: Why Bigger is Simpler

This video from the "Broken Symmetries" channel, titled "Renormalization: Why Bigger is Simpler," introduces the core ideas and motivation behind the RG.

Please watch the first two sections. The first segment introduces the two fundamental questions RG addresses: why macroscopic properties are simple despite microscopic complexity, and why phase transitions are sharp. The second segment outlines the general procedure of starting with a microscopic model and iteratively averaging out small-scale behavior. This process is the essence of coarse-graining.

As the video explains, the central methodological step is coarse-graining. This is the process of creating an effective theory at a larger length scale by averaging over, or "integrating out," the degrees of freedom at smaller scales. Let's see how this works in practice.

Coarse-Graining in Real and Momentum Space

There are two primary ways to conceptualize coarse-graining, which are ultimately equivalent. One works directly with the system in real space (e.g., on a lattice), while the other, more aligned with your QFT background, works in momentum space.

Real-Space Coarse-Graining: The Block-Spin Transformation

The pioneering idea, developed by Leo Kadanoff, was to apply coarse-graining to a lattice model like the Ising model. The procedure involves grouping spins into blocks and replacing each block with a single, new "block spin" that represents the average behavior of the original spins within it.

Kadanoff's real-space renormalization procedure for the 2D Ising model. Original spins on a lattice with spacing $a$ (left) are grouped into blocks. A "decimation" step replaces each block with a single new spin, creating a new lattice with a larger spacing, here $3a$ (middle). A "rescaling" step then shrinks the new lattice back to the original spacing $a'$ for comparison (right). This two-step process—coarse-grain then rescale—is the core RG transformation.

This procedure generates a new, effective Hamiltonian for the block spins. The crucial insight is that this new Hamiltonian should have the same form as the original one, but with different, "renormalized" coupling constants.

The following video provides a concrete example of this process using a specific recipe for determining the new spin values.

Renormalisation group

This lecture by Prof. Jos Thijssen demonstrates a specific coarse-graining procedure for the 2D Ising model.

Watch the segment from this description of the procedure. The key idea is the "majority rule" recipe for defining the new block spin T_k from the four original spins S_i in a plaquette. This makes the abstract idea of coarse-graining into a specific, executable algorithm. The lecturer also introduces the probability distribution W(T_k | \{S_i\}) that formalizes this rule.

Momentum-Space Coarse-Graining: Integrating Out Fast Modes

Kenneth Wilson reformulated and generalized Kadanoff's idea in the language of quantum field theory. Since you are studying QFT, this perspective should feel quite natural. In this context, short-distance physics corresponds to high-momentum (or high-energy) modes, often called "UV" or "fast" modes. Long-distance physics corresponds to low-momentum ("IR" or "slow") modes.

Coarse-graining in momentum space means integrating out the high-momentum degrees of freedom to obtain an effective theory, or effective action, for the low-momentum modes.

Renormalization

This section from Markus Luty's lecture notes provides a clear and formal definition of the Wilsonian RG procedure in the language of path integrals.

Please read the section titled "The Wilson Effective Action". This section formalizes the idea of "integrating out" modes. You'll see how the field \phi is split into low-momentum (\phi') and high-momentum (\phi'') parts, and the path integral is performed over the high-momentum part to define a new effective action S'[\phi']. This S' is the Wilsonian effective action for the low-energy physics.

The core idea is the same in both pictures: we have a procedure, the RG transformation, that takes us from a description of a system at one length scale to an effective description at a larger one.

H_{micro} Coarse-Graining H_{eff}

Scale Invariance: The Signature of a Critical Point

Now, let's consider what makes a critical point so special from this new perspective. As we established in the last lesson, the correlation length $ξ$ diverges at $T_{c}$ . This means there are fluctuations correlated over all possible length scales. The system has no characteristic scale; it is scale-invariant. If you were to "zoom in" or "zoom out," the statistical properties of the system would look the same.

This visual metaphor of "islands within islands," or fractality, is a powerful way to picture scale invariance.

Renormalisation group

This part of Prof. Thijssen's lecture provides an intuitive "cartoon" of what the system looks like at a critical point.

Please watch from the explanation of fluctuations. The key takeaway is the image of red islands within blue islands, within which are smaller red islands, and so on. This self-similar structure at all scales is the physical manifestation of scale invariance.

How does this property relate to our RG transformation? If a system is truly scale-invariant, then applying the coarse-graining and rescaling procedure should leave its effective Hamiltonian unchanged. The transformation maps the system onto itself. Such a point in the space of all possible Hamiltonians is called a fixed point of the Renormalization Group.

H^{*} RG H^{*}

This is the defining property of a critical point in the RG framework. This scale invariance is the deep reason for the power-law behavior of physical quantities near $T_{c}$ . A system with no intrinsic length scale $L$ cannot have observables that depend on, say, $e^{- x / L}$ , because that would introduce a scale. The only functions that do not introduce a scale are power laws, $x^{p}$ .

The Flow of Renormalization

So, critical points are fixed points of the RG transformation. What happens away from a critical point, where the correlation length $ξ$ is finite? In this case, the system is not scale-invariant. Applying the RG transformation will change the effective Hamiltonian. This generates a "flow" in the space of Hamiltonians.

Let's consider the two simple cases:

High Temperature ( $T ≫ T_{c}$ ): The system is disordered with short-range fluctuations. When we average over blocks, these random fluctuations tend to cancel out. The resulting coarse-grained system looks even more disordered, as if it were at an even higher temperature. The RG flow takes the system towards the infinite-temperature fixed point (complete disorder).
Low Temperature ( $T ≪ T_{c}$ ): The system is highly ordered, with a few small, isolated islands of flipped spins. When we average over blocks, these rare fluctuations get smoothed over. The coarse-grained system looks even more perfectly ordered, as if it were at an even lower temperature. The RG flow takes the system towards the zero-temperature fixed point (perfect order).

The critical point is the unstable fixed point that separates these two behaviors. A system starting exactly at $T_{c}$ stays at the critical fixed point under RG, while a system infinitesimally away from $T_{c}$ will flow away toward one of the trivial, stable fixed points (T=0 or T= $\infty$ ). It is the behavior of the RG flow near the critical fixed point that determines the universal critical exponents.

3. The Renormalisation Group

This section from David Tong's lecture notes connects the concepts of RG flow, fixed points, and scale invariance to explain the origin of universality.

Please read the subsection "Universality Explained". Focus on how fixed points are defined as being invariant under the RG transformation and must have a correlation length that is either zero (trivial fixed points) or infinite (critical fixed points). This text beautifully summarizes the conceptual picture we've been building.

Conclusion

In this lesson, we have laid the conceptual groundwork for the Renormalization Group, moving from the problem (breakdown of mean-field theory) to the solution (a theory of scales).

Key Takeaways:

Coarse-graining is a procedure for systematically removing short-distance degrees of freedom to build an effective theory for long-distance physics. It can be visualized in real space (Kadanoff's block spins) or formalized in momentum space (Wilson's integration of fast modes).
Scale invariance is the defining characteristic of a critical point. It means the system exhibits self-similar fluctuations on all length scales, due to a diverging correlation length.
The Renormalization Group (RG) is a transformation that implements coarse-graining and rescaling. A system's effective Hamiltonian "flows" under repeated applications of this transformation.
Fixed points of the RG flow correspond to scale-invariant theories. Critical points are associated with unstable fixed points, which control the universal behavior in their vicinity.

We have established the core concepts. The next logical step is to see them in action. In our next lesson, we will perform a concrete real-space RG calculation for a simple system. This will allow us to explicitly calculate how coupling constants "flow," identify the fixed points, and extract critical properties, turning these powerful abstract ideas into a practical calculational tool.

Recommended resources