🔬 physics

The geometry-first formulation of gauge theory is not equivalent to the symmetry-first one

This paper argues that the geometry-first and symmetry-first formulations of gauge theory are fundamentally inequivalent because the former admits fewer theories, fails to uniquely determine generating structures from principal bundles, and lacks essential surjectivity and fullness in its natural functor to the latter framework.

Original authors: Henrique Gomes

Published 2026-07-29

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Henrique Gomes

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

Imagine you are trying to describe a massive, invisible machine that holds the universe together. In the world of high-energy physics, this machine is called a "gauge theory," and it explains how particles like electrons and quarks talk to each other through forces like electromagnetism and the strong nuclear force. For decades, physicists have used a specific way of thinking about this machine, which we can call the "Symmetry-First" approach. In this view, you start by declaring, "There is a rulebook (a group) that tells particles how to transform," and then you build the machine around that rulebook. It's like saying, "We have a dance troupe with a specific choreography, so let's build a stage for them."

However, a new way of thinking has emerged, called the "Geometry-First" approach. Instead of starting with the abstract rulebook, this method starts with the physical "dancers" themselves—the fundamental particles and the shapes of the spaces they live in. It asks, "If we look at the geometry of these particles, what rules naturally pop out?" It's like looking at the dancers' movements and deducing the choreography from their steps, rather than assuming the choreography exists first. The big question physicists have been asking is: Are these two ways of thinking just different languages describing the exact same machine, or are they actually describing two different machines?

This paper, written by Henrique Gomes, argues that the answer is a definitive "no." The two approaches are not equivalent. While they might look similar in some simple cases, they actually describe different possibilities for how the universe works. The author shows that the Geometry-First approach is more restrictive; it cannot describe every theory that the Symmetry-First approach can. Specifically, the paper proves that if you try to build a theory using only the "dancers" and their shapes (Geometry-First), you run into walls that don't exist in the "rulebook" approach. You might find yourself unable to explain certain types of electric charges, or you might lose the ability to distinguish between a machine built from three separate parts versus one built from a single, complex part.

The Two Ways to Build a Machine

To understand why this matters, let's use an analogy. Imagine you are a toy designer.

The Symmetry-First Approach (The Rulebook Method)
In this method, you start with a manual. You say, "I want a toy that can rotate in 360 degrees, flip upside down, and change color." You write these rules down in a big book called the "Gauge Group." Then, you build the toy (the particles) to fit these rules. If you want a toy that can spin forever without stopping, you just write that into the book. If you want a toy that can spin in weird, irrational amounts, you just add that to the book. The rules come first; the toy is built to obey them. This is the standard way physicists have described the universe for a long time.

The Geometry-First Approach (The Shape Method)
In this method, you don't start with a manual. You start with a pile of raw materials: a few specific shapes (fundamental vector bundles) and some tools (like inner products or volume forms). You say, "I have a sphere and a cube. If I try to rotate them without breaking their shape, what rules naturally appear?" You look at the sphere and realize, "Ah, you can only rotate this in specific ways to keep it looking the same." The rules (the Gauge Group) are discovered as a result of the shapes, not chosen beforehand.

The Big Problem: The Rules Don't Always Match

Gomes argues that these two methods are not interchangeable. If you try to translate a Symmetry-First toy into a Geometry-First toy, you often hit a dead end. Here are the three main ways they fail to match:

1. The "Irrational Charge" Wall
Imagine you have a toy that can have a "charge" (like electric charge). In the Symmetry-First world, you can easily design a toy with a charge of 1 and another with a charge of $\sqrt{2}$ (an irrational number). They just follow the rules you wrote down.
But in the Geometry-First world, charges are built by stacking shapes on top of each other (like stacking Lego bricks). If you start with one brick, you can only make stacks of 1, 2, 3, or 4 bricks. You can never make a stack of $\sqrt{2}$ bricks because you can't cut a brick in half and keep the shape intact in this specific mathematical way.
The paper proves that if you try to build a theory with "irrational" charges using only shapes, you simply can't do it. The Geometry-First approach forces charges to be whole-number multiples of a base unit (a lattice). The Symmetry-First approach allows for any number. So, if the universe actually contains particles with irrational charge ratios, the Geometry-First method would fail to describe it, while the Symmetry-First method would breeze right through.

2. The "Hidden Identity" Trap
Sometimes, the Symmetry-First method allows you to build a machine where the rules are a bit "sloppy." For example, you might have a machine with three gears (SU(3), SU(2), and U(1)), but the way the parts connect means that a tiny, invisible twist in all three gears at once does nothing to the toy. The toy doesn't notice.
In the Symmetry-First world, you can just say, "Okay, let's call the machine 'G'."
But in the Geometry-First world, you are forced to build the machine from three separate, independent shapes. You can't easily "glue" them together to hide that tiny twist. The paper shows that for the Standard Model (our best theory of particles), the "real" machine that acts on the particles is actually a simplified version of the three-gear machine. The Geometry-First method, which builds from three independent shapes, cannot naturally produce this simplified version without adding extra, arbitrary "background" rules that weren't there to begin with. It's like trying to build a car from three separate engines and realizing you can't make them work as one smooth unit without adding a secret manual.

3. The "Mirror" Problem
Finally, there is a problem with how the machines can be transformed. In the Symmetry-First world, you can imagine a transformation that flips the machine inside out (an "outer automorphism"). It's like taking a glove and turning it inside out; it's still a glove, but the rules have changed in a way that the Symmetry-First manual allows.
However, in the Geometry-First world, the rules are tied strictly to the shape of the material. If you try to turn the shape inside out, you might break the "volume" or the "orientation" of the shape. The paper shows that there are transformations allowed in the Symmetry-First world that simply have no corresponding move in the Geometry-First world. It's like having a dance move that is legal in the choreography book but physically impossible for the dancers because their joints don't bend that way.

Why This Matters

The author isn't saying the Geometry-First approach is "wrong." It's actually very powerful because it explains why the rules exist based on the shapes of particles. It removes the need to just "guess" the rules. However, the paper concludes that this approach is not just a different way of writing the same thing. It is a different theory with a different "menu" of what is possible.

If the universe turns out to have particles with irrational charges, or if the specific way our particles interact requires a "sloppy" rulebook that the Geometry-First method can't build, then the Geometry-First approach would be incomplete. The paper proves that the two approaches are mathematically distinct. You cannot simply swap one for the other without losing information or changing the set of theories you are allowed to write down.

In short, the "Rulebook" and the "Shape" methods are not twins. They are cousins who look alike but have different family secrets. The Geometry-First method is stricter, more explanatory, but it cannot describe every scenario the Symmetry-First method can. This means that if we want to understand the deep structure of the universe, we have to be very careful about which "lens" we are looking through, because they don't show us the exact same picture.

Problem Statement
This paper investigates the theoretical equivalence between two distinct formulations of gauge theory: the standard "symmetry-first" approach (Principal Fibre Bundle Point of View, or PFB-POV) and an alternative "geometry-first" approach (Vector Bundle Point of View, or VB-POV). In the PFB-POV, the theory begins with a principal $G$ -bundle and a connection, with matter fields defined as sections of associated vector bundles. In the VB-POV, the theory begins with a finite collection of fundamental vector bundles equipped with specific fibre structures (e.g., inner products, volume forms), from which the gauge group arises as the automorphism group of these structures. While a companion paper (Gomes, 2026a) demonstrated that the geometry-first formulation can reproduce many explanatory successes of the symmetry-first approach (such as the Higgs mechanism and charge quantization), it left open the question of whether the two formulations are merely notational variants or if they define distinct theoretical structures. This paper argues they are not equivalent.

Methodology
The author employs a multi-faceted approach combining differential geometry, representation theory, and category theory to compare the two formulations:

Theory-Space Comparison: The paper analyzes the set of theories admissible in each formulation. It examines whether every PFB-POV theory can be generated by a finite set of fundamental bundles with classical linear-algebraic structures (the "menu").
Reconstruction Analysis: The paper tests whether the "provenance" of a gauge theory (i.e., the specific fundamental bundles and their tensorial genealogy) can be uniquely recovered from the resulting matter bundles and connections alone.
Categorical Analysis: Formalizing the comparison, the author defines a functor $F$ from the category of geometry-first generating objects ( $\mathbf{VB}_{gen}$ ) to the category of principal bundles with matter ( $\mathbf{PFB}_{mat}$ ). The paper evaluates whether $F$ constitutes an equivalence of categories by testing for three conditions: faithfulness, essential surjectivity, and fullness.

Key Contributions and Results

1. Strict Inclusion of Theory-Spaces (Failure of Essential Surjectivity)
The geometry-first formulation admits a strictly smaller set of theories than the symmetry-first formulation.

Non-Compact Groups and Charge Quantization: The VB-POV requires gauge groups to be automorphism groups of structured fibres generated by finite tensorial data. Such groups are real-algebraic. Consequently, the VB-POV cannot present theories with the additive group $\mathbb{R}$ as the structure group coupled to charged matter. In the PFB-POV, one can postulate matter with irrational charge ratios (using $\mathbb{R}$ ), but the VB-POV's tensorial construction enforces a lattice of charges (integer multiples of a generator). Proposition 2 proves that no finite tensorial menu can present the additive $\mathbb{R}$ theory with non-trivial charged matter.
Diagonal Quotients (Standard Model): Even for compact groups, the VB-POV faces obstructions. The Standard Model gauge group $G = SU(3) \times SU(2) \times U(1)$ acts on matter with a kernel $\mathbb{Z}_6$ . The faithful quotient $G' = G/\mathbb{Z}_6$ is a valid PFB theory. However, $G'$ is not isomorphic to a product of automorphism groups of independent fundamental fibres on the classical menu. While $G'$ can be generated by a single fundamental bundle with a fixed background splitting, the matter sector alone cannot distinguish between the product origin ( $G$ ) and the quotient origin ( $G'$ ). Thus, $G'$ lacks a preimage in the standard geometry-first setup.

2. Non-Recoverability of Geometric Provenance
The paper demonstrates that the matter sector does not uniquely determine the geometry-first starting point.

Factor Identification: If a gauge group is a product $G = G_1 \times G_2$ , and matter bundles are constructed such that $G_1$ acts trivially on some sectors and $G_2$ on others, the total matter bundle does not reveal the product structure.
Tensorial Genealogy: Even with full knowledge of the connection and the tensorial construction of matter bundles, one cannot distinguish between a theory generated by independent fundamental bundles (yielding a product group) and a theory generated by a single bundle with a fixed background splitting (yielding a diagonal quotient). The matter sector fixes the action but not the global group structure or the specific geometric origin.

3. Categorical Non-Equivalence (Failure of Fullness)
The functor $F: \mathbf{VB}_{gen} \to \mathbf{PFB}_{mat}$ is faithful but neither essentially surjective nor full.

Failure of Fullness: The paper constructs a counterexample using a real oriented fibre $\mathbb{R}^{2m}$ with an inner product and volume form, recovering the group $SO(2m)$. The PFB category admits morphisms induced by outer automorphisms of $SO(2m)$, specifically conjugation by an improper orthogonal map (a reflection $R$ with $\det R = -1$ ). However, in the VB category, a morphism must be induced by a structure-preserving map of the fundamental fibre. Since $R$ reverses the volume form, it is not an automorphism of the structured fibre and thus does not exist in the source category. Consequently, the functor is not full; there are symmetries in the PFB description that have no counterpart in the geometry-first description.

Significance and Claims
The paper concludes that the geometry-first and symmetry-first formulations are not equivalent. They differ in three fundamental respects:

Theory-Space: The geometry-first approach is more restrictive, excluding certain non-compact theories (like additive $\mathbb{R}$ with arbitrary charges) and specific global forms of compact groups (like the diagonal quotient of the Standard Model) unless the "menu" of allowed structures is expanded.
Recoverable Structure: The symmetry-first formulation (principal bundle + matter) does not encode the specific geometric provenance (fundamental bundles and their independence) required by the geometry-first formulation.
Explanatory Order: The geometry-first formulation derives the gauge group from fibre structures, whereas the symmetry-first formulation posits the group independently. The categorical analysis shows that the "interaction structure" (which factors of the group act on which sectors) is not recoverable from the symmetry-first objects alone.

The author asserts that the geometry-first formulation is not a mere notational variant but a distinct theoretical framework with stricter primitive objects and a different explanatory order. For compact groups, the difference is often one of "geometric provenance and economy" rather than mere presentability. The paper suggests that if physicists intend to restrict gauge theories to those with specific tensorial genealogies, these assumptions should be explicitly stated rather than treated as automatic features of the principal bundle formalism.

The Two Ways to Build a Machine

The Big Problem: The Rules Don't Always Match

Why This Matters

More like this