On logarithmic extensions of local scale-invariance

✨

This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are watching a pot of water boil. If you wait long enough, the bubbles stop changing size and the water reaches a steady, calm state. In physics, we call this equilibrium. Scientists have long known how to predict the behavior of such calm systems using a set of rules called Scale Invariance. Think of this like a fractal: if you zoom in or out, the pattern looks the same. It's a beautiful, symmetrical rule that governs how things behave when they are settled down.

But what happens when you suddenly turn off the heat and the water starts cooling down? Or when you take a magnet and suddenly drop its temperature? The system is far from equilibrium. It's in a state of aging. It's not settled; it's slowly, painfully relaxing. It doesn't look the same if you zoom in or out in the same way, and it doesn't care about "time" in the usual sense (yesterday is different from today).

For a long time, physicists tried to use the old "calm water" rules to predict how these "cooling down" systems behave. They built a new set of rules called Local Scale Invariance (LSI). It worked pretty well for a while, like a map that gets you 90% of the way to your destination. But when they looked at the data with a magnifying glass, they noticed the map was slightly off. The curves didn't quite match the data points. There was a tiny, stubborn error that the old rules couldn't explain.

The New Idea: "Logarithmic" Extensions

This paper, written by Malte Henkel, proposes a fix. He suggests that the old rules need a "logarithmic" upgrade.

To understand this, let's use an analogy. Imagine you are trying to describe the height of a person.

The Old Way (Standard LSI): You say, "This person is 6 feet tall." Simple. One number.
The Logarithmic Way (Logarithmic LSI): You realize that "6 feet" isn't just a single number. It's a "6 feet" plus a tiny bit of "6 feet and a whisper." In math, these two numbers get stuck together in a special package called a Jordan Cell.

In the world of physics, this means that every "scaling operator" (the thing that tells us how the system behaves) isn't just one single entity. It's a doublet. It's like a character in a video game who has a main form and a "ghost" form that is almost identical but slightly different. They are so close that they blur together.

When you try to measure the system, you don't just see the main form; you see the main form plus a tiny correction that involves a logarithm (a mathematical function that grows very slowly, like the sound of a whisper getting quieter).

Why Does This Matter?

The author tested this new "Logarithmic LSI" theory against two very different, messy, real-world scenarios:

Growing Interfaces (The KPZ Equation): Imagine a sandpile growing as you throw sand on it randomly. The surface gets rough and bumpy. The paper looked at how the surface responds to a tiny nudge.
- The Result: The old rules (LSI) were off by about 5%. The new rules (Logarithmic LSI) matched the data with 99.9% accuracy. It was like finally finding the missing piece of a puzzle that everyone thought was just a printing error.
Directed Percolation (The Spreading Fire): Imagine a forest fire spreading through a grid of trees. Will it die out, or will it consume everything? This is a critical moment.
- The Result: Again, the old rules failed to capture the fine details of how the fire spreads over time. The new logarithmic rules perfectly described the shape of the data, even when the fire was just starting to spread.

The "Time Travel" Problem

The most important twist in this story is Time.
In the calm, equilibrium world, time is symmetrical. You can run the movie forward or backward, and the physics looks the same (mostly). This allows for simple rules.
But in aging systems, time is broken. You can't run the movie backward. The system remembers its past.

Because time is broken, the "doublet" nature of the particles (the main form and the ghost form) behaves differently than in the calm world. In the calm world, the ghost form is tightly locked to the main form. In the aging world, they are freer to move independently. This freedom allows for those extra "logarithmic whispers" to appear in the data, which the old rules missed.

The Takeaway

Think of this paper as upgrading the GPS for the universe.

Old GPS (Standard LSI): "Turn left in 5 miles." (Good enough for a highway, but you might miss a small turn).
New GPS (Logarithmic LSI): "Turn left in 5 miles, but also account for the slight curve in the road and the wind resistance." (Perfect for the bumpy, winding roads of non-equilibrium physics).

The author shows that when things are far from equilibrium (like cooling magnets or growing sandpiles), nature is a bit more complex than we thought. It doesn't just scale; it scales with a "logarithmic echo." By listening to that echo, we can finally predict how these chaotic systems behave with incredible precision.

In short: The universe, when it's in a hurry to settle down, doesn't just follow the simple rules. It follows a more complex, "logarithmic" script, and this paper finally wrote down the full script.

Technical Summary: On Logarithmic Extensions of Local Scale-Invariance

Paper: arXiv:1009.4139v2 [hep-th]
Author: Malte Henkel
Date: December 13, 2012

1. Problem Statement

The paper addresses the theoretical description of ageing phenomena in non-equilibrium statistical physics. Ageing is characterized by slow relaxation, the breaking of time-translation invariance, and dynamical scaling. While standard Local Scale-Invariance (LSI) based on the ageing algebra $age(d)$ (a sub-algebra of the Schrödinger algebra) successfully predicts the form of two-time autoresponse functions in many systems, it fails to capture fine details in the scaling functions of certain universality classes (e.g., 1D Kardar-Parisi-Zhang and directed percolation) when compared to high-precision numerical simulations.

Specifically, standard LSI predicts a scaling function $f_R(y)$ (where $y=t/s$ ) that is purely power-law based. However, simulation data often show systematic deviations near $y \to 1$ that cannot be resolved by simply adjusting exponents. The author posits that these deviations arise from logarithmic corrections to scaling, necessitating a generalization of LSI analogous to Logarithmic Conformal Field Theory (LCFT) but adapted for non-equilibrium systems lacking time-translation invariance.

2. Methodology

The author constructs a Logarithmic Extension of Local Scale-Invariance (llsi) by adapting the representation theory of the ageing algebra $age(d)$.

Algebraic Construction:
- The standard ageing algebra $age(d)$ is generated by time dilatations ( $X_0, X_1$ ), special conformal transformations, and spatial translations/rotations.
- In standard LSI, scaling operators are characterized by a single scaling dimension $x$ .
- In the logarithmic extension, the author replaces the scaling dimensions with Jordan cells (Jordan matrices). Specifically, the scaling dimensions $x$ and $\xi$ (associated with the generators $X_0$ and $X_1$ ) are promoted to matrices:
  $x \to \begin{pmatrix} x & x' \\ 0 & x \end{pmatrix}, \quad \xi \to \begin{pmatrix} \xi & \xi' \\ 0 & \xi \end{pmatrix}$
- This implies that scaling operators form doublets $\Psi = (\psi, \phi)^T$ , where the action of the algebra mixes the components.
Derivation of Two-Point Functions:
- The author derives the covariance conditions for two-point functions $F, G, H$ under the logarithmic representation of $age(d)$.
- Unlike logarithmic Schrödinger or conformal invariance (which assume time-translation invariance), the absence of the time-translation generator in $age(d)$ allows for independent logarithmic corrections in the scaling functions and the presence of quadratic logarithmic terms ( $\ln^2 t$ ).
- The resulting two-point functions involve scaling functions that depend on the ratio $y = t_1/t_2$ and include terms proportional to $\ln(t_2)$ and $\ln^2(t_2)$ .
Application to Models:
- The derived theoretical forms are applied to two specific non-equilibrium universality classes:
  1. 1D Kardar-Parisi-Zhang (KPZ) equation: Interface growth.
  2. 1D Critical Directed Percolation (DP): Contact process.
- The author compares the theoretical predictions against high-precision numerical simulation data (specifically autoresponse functions $\chi(t,s)$ or $R(t,s)$ ).

3. Key Contributions

Formulation of Logarithmic Ageing Algebra: The paper defines the logarithmic extension of the ageing algebra $age(d)$, explicitly handling the lack of time-translation invariance. This distinguishes it from previous works on logarithmic Schrödinger or conformal Galilean invariance.
Generalized Scaling Functions: The derivation of the most general form of covariant two-point functions (Eq. 3.16). Crucially, the author shows that without time-translation invariance:
- Logarithmic corrections can appear independently in the scaling functions.
- Quadratic logarithmic terms ( $\ln^2 y$ ) are permitted, unlike in standard LCFT where they are often constrained or absent in specific forms.
- The constraint $F=0$ (found in standard LCFT) is no longer required.
Resolution of Scaling Anomalies: The paper demonstrates that the "fine structure" of scaling functions in ageing systems, which standard LSI fails to capture, is naturally explained by these logarithmic extensions.

4. Results

The theoretical predictions were tested against simulation data for the 1D KPZ and 1D Directed Percolation models.

1D KPZ Equation:
- Standard LSI (with $a \neq a'$ ) fits data to within ~5% accuracy.
- The Logarithmic LSI (llsi) model, incorporating terms like $\ln(1-y^{-1})$ and $\ln^2(1-y^{-1})$ , fits the data with < 0.1% accuracy over the entire range of $y$ .
- The fit parameters ( $A_1, A_2$ ) for the logarithmic terms are non-zero and of comparable magnitude, confirming the necessity of the logarithmic extension.
1D Directed Percolation:
- Similar to KPZ, standard LSI with $a' \neq a$ fits well for large $y$ but deviates as $y \to 1$ .
- The llsi model (Eq. 4.8) provides a superior fit, capturing the systematic deviations near $y=1$ .
- The analysis suggests that the logarithmic partner of the order parameter is physically relevant, even if its exact microscopic identification remains an open question.
Distinction from Other Logarithmic Forms:
- The author explicitly distinguishes llsi from other phenomenological logarithmic scaling forms (e.g., $R(t,s) \sim s^{-1-a} f(t/s, \ln t / \ln s)$ found in some magnetic systems), proving via Appendix B that the latter is incompatible with the algebraic structure of llsi.

5. Significance

Theoretical Advancement: This work extends the powerful framework of conformal and Schrödinger invariance to the non-equilibrium regime with logarithmic corrections, providing a rigorous algebraic basis for phenomena previously described only phenomenologically.
Precision in Non-Equilibrium Physics: It offers a new tool to interpret high-precision simulation data in ageing systems, resolving discrepancies that standard scaling theories could not explain.
Physical Insight: The success of the model suggests that non-equilibrium critical systems may possess "logarithmic partners" to their scaling operators, analogous to LCFT in equilibrium. This hints at potential non-local observables or hidden symmetries in ageing dynamics.
Future Directions: The paper identifies the need to find exactly solvable models of llsi and to physically identify the logarithmic partners in specific systems (e.g., relating them to global persistence exponents).

In summary, Henkel proposes that the breakdown of standard scaling in non-equilibrium ageing is not a failure of scale-invariance but a signal of a logarithmic extension of the underlying symmetry algebra, characterized by Jordan cells in the representation of scaling dimensions. This framework successfully unifies theoretical predictions with high-precision numerical data.