⚛️ phenomenology

Information Geometry of DGLAP Evolution: From Geodesic Tests to Emergent Small-Circle Geometry

This paper employs information geometry to demonstrate that leading-order DGLAP evolution of valence quark distributions follows a highly organized, near-small-circle trajectory on the Fisher-Rao manifold, characterized by negligible geodesic curvature and effectively captured by a two-dimensional affine subspace.

Original authors: Wei Kou, Xurong Chen

Published 2026-10-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Wei Kou, Xurong Chen

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ 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

Inside every proton, the building blocks of matter known as quarks and gluons are not static; they are in a constant state of flux. As physicists probe deeper into the proton with higher energy, they see more of these particles, and their arrangement changes. This change is governed by a set of rules called the DGLAP equations, which act like a map for how the internal structure of a proton evolves as the scale of observation shifts. For decades, scientists have tracked these changes by looking at how the probability of finding a particle at a certain speed changes. However, a new approach asks a different kind of question: instead of just tracking the numbers, what does the shape of this evolution look like if we treat the entire distribution of particles as a single point moving through a vast, abstract space?

This space is not made of physical distance, but of statistical difference. Imagine every possible way the particles could be arranged as a location on a map. As the proton's energy increases, the arrangement of its particles shifts, and this shift traces a path across that map. The researchers in this study wanted to know if nature follows the most direct, straightest possible line through this space, or if it takes a curved, winding route. In the language of geometry, a straight line on a curved surface is called a geodesic. If the proton's evolution followed a geodesic, it would mean the change is happening in the most efficient, unforced way possible. If it does not, the path reveals a hidden structure or a specific force guiding the change.

The team, led by Wei Kou and Xurong Chen, took a specific type of proton data, focusing on the "valence" particles which give the proton its identity, and mapped their evolution from a low energy level up to a very high one. They translated the changing probabilities of finding these particles into a geometric curve on a sphere, a shape that naturally fits the rules of probability. By calculating the speed and the turning force of this curve, they could test whether it was a straight line or a bend. Their calculations showed that the path is not a straight line. The curve has a measurable bend, meaning the evolution is not following the simplest possible route through the statistical landscape.

However, the story does not end with a simple curve. The researchers found that while the path is not straight, it is remarkably orderly. When they analyzed the curve in detail, they discovered that it stays almost perfectly within a flat, two-dimensional sheet that cuts through the sphere. This is a significant finding because the space of all possible particle arrangements is incredibly complex and high-dimensional. The fact that the proton's evolution is confined to such a simple, flat slice suggests a deep underlying simplicity in how these particles change. The curve traced out within this flat slice is not a random wobble; it is a precise arc of a small circle.

The study revealed that this small-circle shape is not just a visual coincidence but a mathematical reality. The researchers compared the actual turning force of the proton's evolution with the turning force predicted by this small-circle geometry. The two matched with astonishing precision, differing by less than two parts in a thousand. This agreement confirms that the complex, chaotic-looking changes in the proton's internal structure are actually following a very specific, organized geometric path. The path is slightly curved, but it is a consistent, predictable curve that stays within a narrow, flat plane.

To understand why this happens, the team looked at which parts of the proton were driving the movement. They found that the regions where the particles are most likely to be found are not necessarily the ones causing the path to move or bend. Instead, the motion is driven by the relative reshaping of the distribution. Some parts of the proton change shape in a way that pushes the path forward, while other parts cause it to turn. These two effects happen in different regions of the proton's internal structure, keeping the overall motion smooth and organized. The study also showed that the speed at which the proton evolves slows down as the energy increases, but the shape of the path remains consistent.

The results paint a picture of a universe where the evolution of matter, even at the most fundamental level, adheres to a strict geometric order. The proton does not wander randomly through the space of possibilities. Instead, it follows a narrow, curved track that is almost perfectly circular. This track is so precise that it can be described by a simple circle cut from a sphere, a shape that emerges naturally from the complex interactions of quarks and gluons. The research provides a new way to see the proton, not just as a bag of changing particles, but as a traveler following a specific, elegant route through the geometry of probability. This discovery does not change the known laws of physics but offers a new, clearer lens through which to view them, revealing that the path of evolution is far more structured and predictable than previously imagined.

Technical Summary: Information Geometry of DGLAP Evolution

Problem Statement
The paper investigates the geometric structure of Leading-Order (LO) non-singlet DGLAP evolution within the space of normalized valence parton distribution functions (PDFs). While PDF evolution is conventionally described via changing distributions and Mellin moments, the authors pose a complementary question: What is the geometric path selected by QCD evolution in the space of normalized probability densities? Specifically, they ask whether the DGLAP-selected trajectory is a geodesic (the shortest path) on the Fisher–Rao manifold and, if not, how its deviation from geodesic motion is organized.

Methodology
The authors formulate the LO non-singlet DGLAP evolution as a trajectory on the Fisher–Rao manifold of normalized valence distributions. The methodology relies on the following steps:

Isometric Embedding: They utilize the square-root embedding $\Phi = 2\sqrt{p}$ , which maps the normalized probability density $p(x, Q^2)$ isometrically into the positive sector of a Hilbert sphere of radius $R_{FR} = 2$ . This transforms the statistical manifold into a geometric sphere where the Fisher–Rao metric corresponds to the standard Euclidean metric on the sphere.
Velocity and Acceleration: The evolution is parameterized by $\tau = \int \frac{\alpha_s}{2\pi} \frac{d\mu^2}{\mu^2}$ , rendering the evolution operator fixed at LO. The velocity ( $\partial_\tau \Phi$ ) and acceleration ( $\partial^2_\tau \Phi$ ) of the trajectory are derived directly from the DGLAP operator acting on the distribution.
Geodesic Curvature Calculation: To test geodesicity, the authors isolate the transverse geodesic acceleration by removing the radial component (fixed by sphere geometry) and the component parallel to the velocity (which only affects parameterization speed). The norm of this residual, normalized by the squared Fisher speed, yields the parameterization-invariant geodesic curvature $k_{FR}$ .
Numerical Implementation: A reference trajectory is constructed using the GRV98 parametrization for the total valence combination ( $u_v + d_v$ ) evolved from $Q^2 = 2$ GeV $^2$ to $2000 $GeV$ ^2$. The analysis includes validation against analytic Mellin-space solutions and finite-chord curvature estimators.
Global Geometry Analysis: An arc-length-weighted Principal Component Analysis (PCA) is performed to determine the effective dimensionality of the trajectory. The authors test whether the trajectory is confined to a two-dimensional affine plane, which would intersect the Fisher sphere in a "small circle." They compare the curvature predicted by this small-circle geometry with the curvature reconstructed from the DGLAP acceleration.

Key Contributions and Results

Non-Geodesic but Regular Motion: The study confirms that the DGLAP-selected trajectory is not a Fisher–Rao geodesic; it possesses a non-zero geodesic curvature ( $k_{FR} \approx 0.8984$ ). However, this curvature is remarkably stable, varying only weakly (at the level of $10^{-3}$ ) across the entire evolution interval, indicating an organized rather than arbitrary bending.
Effective Dimensional Reduction: PCA reveals that the trajectory is effectively confined to a two-dimensional affine sector. The first two principal directions capture more than 99.999% of the trajectory's variance.
Emergent Small-Circle Geometry: The intersection of this dominant affine plane with the Fisher sphere defines a "small circle." The curvature predicted by this geometric construction ( $k_{sc} \approx 0.8970$ ) agrees with the dynamically reconstructed DGLAP curvature to within $1.6 \times 10^{-3}$ .
Path Length Efficiency: While the trajectory bends, it remains extremely close to the metric shortest path between endpoints. The ratio of the geodesic distance to the actual path length ( $\eta_G$ ) is approximately $0.9956$, meaning the physical path exceeds the shortest path by only $\sim 0.44\%$ .
Momentum-Fraction Structure: An $x$ -resolved analysis demonstrates that the Fisher motion is governed by the relative reshaping of the distribution rather than the absolute probability weight. The regions contributing most to the statistical displacement (Fisher speed) differ from those dominating the transverse bending (acceleration). Characteristic momentum fractions (the pivot point $x_p$ and acceleration zeros $x_{\kappa,i}$ ) migrate toward lower $x$ as the scale increases but remain distinct.
Mellin Perspective: The Mellin representation provides a complementary view, suggesting that the effective low-dimensional trajectory arises from the collective evolution of the spectral hierarchy.

Significance
The paper establishes a quantitative information-geometric characterization of the reference LO DGLAP trajectory. Its primary significance lies in revealing that the complex, scale-dependent evolution of parton distributions, when viewed through the lens of information geometry, organizes itself into a highly structured, near-small-circle geometry over a finite scale interval. This finding suggests that the "organizing principle" of QCD evolution in the space of normalized distributions is not random but follows a specific, low-dimensional geometric path that is nearly, but not exactly, a geodesic. The work bridges the gap between the dynamical laws of QCD (DGLAP) and the statistical geometry of probability distributions, offering a new perspective on the effective dimensionality of partonic structure evolution.

Technical Summary: Information Geometry of DGLAP Evolution

More like this