A Core Representation Theorem for Scheme-Invariant Collinear Factorization in QCD

This paper establishes a categorical "Core Representation Theorem" that formalizes the scheme-redundancy in QCD collinear factorization by modeling coefficients and correlators as modules over an interface algebra, thereby identifying the universal scheme-invariant physical observable as the relative tensor product of these components.

The Gist

Imagine you are trying to describe the flavor of a complex dish, like a rich stew, to a friend. You want to explain exactly how the taste is created.

In the world of particle physics (specifically Quantum Chromodynamics, or QCD), scientists do something similar. They try to explain how particles (like protons) behave when smashed together at high speeds. They break the problem down into two parts:

1. The Short-Distance Part: The high-energy "explosion" that happens when particles collide. This is easy to calculate with math.
2. The Long-Distance Part: The messy, internal structure of the proton itself (the "stew" ingredients). This is hard to calculate and depends on the specific conditions.

The paper you shared is about a fundamental problem with how scientists currently write down these descriptions. Here is the breakdown in simple terms:

The Problem: The "Recipe" is Arbitrary

Right now, when physicists write the formula for a particle collision, they have to make an arbitrary choice about where to draw the line between the "explosion" and the "stew."

Think of it like this: You are baking a cake. You decide that the "sugar" belongs to the recipe, and the "flour" belongs to the pantry. But then, your friend says, "No, let's put half the sugar in the pantry and half in the recipe."

• The Result: The final cake tastes exactly the same.
• The Confusion: The list of ingredients (the "recipe") looks completely different depending on who wrote it, even though the cake (the physical reality) is identical.

In physics, this is called Scheme Dependence. The "Short-Distance" numbers and the "Long-Distance" numbers change depending on how you draw the line, but the final prediction for the experiment never changes. This redundancy is annoying because it makes it hard to compare different theories or to use machine learning to find the simplest description of nature.

The Solution: The "Core" Recipe

The author, Dustin Keller, uses advanced mathematics (specifically a branch called Category Theory) to solve this. He proposes a way to strip away the arbitrary choices and find the Core of the description.

Here is the analogy:
Imagine you have two lists of ingredients:

• List A (The Chef): "I used 2 cups of flour and 1 cup of sugar."
• List B (The Baker): "I used 1.5 cups of flour and 1.5 cups of sugar."

If you know that the Chef and the Baker are just using different measuring cups (a "scheme change"), you can realize that they are actually describing the same total amount of dough.

Keller's paper builds a mathematical machine that takes all these different lists (schemes) and smashes them together to find the one true, invariant core.

• It doesn't matter if you call it "flour" or "sugar" or "half-and-half."
• The machine ignores the labels and only keeps the total dough that actually matters for the final cake.

The "Interface Algebra": The Rulebook for Swapping

To do this, the author invents a concept called an Interface Algebra. Think of this as a Rulebook for Swapping.

• If you take a piece of "flour" from the Chef's list and move it to the Baker's list, the Rulebook tells you exactly how to adjust the "sugar" to keep the total dough the same.
• This Rulebook ensures that no matter how you shuffle the ingredients between the two lists, the final result remains consistent.

The paper proves that if you follow these rules and "cancel out" all the shuffling, you are left with a Terminal Object (a fancy math term for the "final destination"). This destination is the Core Representation. It is the most compact, simplest, and most honest description of the physics possible. You cannot make it any simpler without losing real information.

Why This Matters (The "So What?")

1. Clarity for AI and Machine Learning: If you want a computer to learn the laws of physics from data, you don't want it to learn the arbitrary "flour vs. sugar" labels. You want it to learn the "Core Dough." This paper gives a blueprint for how to feed data to a computer so it only learns the real physics, not the mathematical noise.
2. Universal Language: It allows scientists using different methods (different "schemes") to talk to each other. They can translate their messy, specific recipes into this universal "Core Language" and know they are talking about the same thing.
3. Efficiency: It proves that you don't need to carry around extra baggage. The "Core" contains everything that is physically real and nothing that is just an artifact of how you chose to write the math.

Summary

Think of this paper as a universal translator for particle physics.

• Before: Scientists speak different dialects (schemes) where the ingredients look different, even though the dish is the same.
• After: This paper provides a machine that translates every dialect into a single, perfect, "Core" description. It strips away the confusion, leaving only the pure, unchangeable truth of how the universe works.

The author isn't trying to discover new particles; he is trying to clean up the dictionary we use to describe them, ensuring that when we say "proton," we all mean the exact same thing, regardless of how we calculated it.

Technical Summary

1. Problem Statement

In perturbative Quantum Chromodynamics (QCD), factorization theorems (such as those for Deep Inelastic Scattering, DIS) express physical observables as a composite of short-distance coefficients ($C$) and long-distance non-perturbative correlators (e.g., Parton Distribution Functions, PDFs, denoted as $f$).

A persistent structural issue is presentation non-uniqueness:

• The individual components $C$ and $f$ are not physical observables; they depend on arbitrary choices of factorization schemes, renormalization scales, and operator bases.
• These choices induce finite redefinitions (mixing) via invertible kernels $Z$. Specifically, $f \to Z \ast f$ and $C \to C \ast Z^{-1}$.
• While the physical observable (the recomposed product $C \ast f$) remains invariant, the intermediate constituents contain "redundant" information that varies with the scheme.
• The Challenge: There is no rigorous, universal mathematical framework to define the "scheme-invariant core" of these factorized descriptions. Current treatments often rely on ad-hoc definitions or specific representations (e.g., Mellin moments vs. $x$-space), lacking a general categorical characterization of the invariant content.

2. Methodology

The author employs category theory to formalize the structural redundancy of QCD factorization. The methodology proceeds as follows:

• Categorical Modeling:

  • The paper models the "recomposition calculus" (how $C$ and $f$ combine to form an observable) as a symmetric monoidal category $(\mathcal{V}, \otimes, \mathbf{1})$. In QCD, $\otimes$ typically represents Mellin convolution (in $x$-space) or pointwise multiplication (in moment space).
  • Admissible Redundancies: The set of finite scheme transformations (kernels $Z$) is formalized as an interface algebra object $\mathcal{A}$ within $\mathcal{V}$.
  • Module Structures:
    • Short-distance coefficients $C$ are modeled as a right $\mathcal{A}$-module.
    • Long-distance correlators $f$ are modeled as a left $\mathcal{A}$-module.
    • The transformation rules ($f \to Z \ast f$, $C \to C \ast Z^{-1}$) correspond to the module actions.

• Balanced Morphisms:

  • A physical evaluation map $\Phi: C \otimes f \to \mathcal{O}_{R,\alpha}$ (where $\mathcal{O}_{R,\alpha}$ is the observable at a specific accuracy) is defined as $\mathcal{A}$-balanced if it respects the scheme redundancy:
    $$\Phi((C \cdot a) \otimes f) = \Phi(C \otimes (a \cdot f))$$
    for all $a \in \mathcal{A}$. This encodes the physical requirement that inserting a counterterm on one side is equivalent to inserting it on the other.

• The Core Construction:

  • The paper constructs the relative tensor product (or balanced tensor product) $C \otimes_{\mathcal{A}} f$.
  • Mathematically, this is the coequalizer of the two action maps:
    $$(C \otimes \mathcal{A}) \otimes f \rightrightarrows C \otimes f \twoheadrightarrow C \otimes_{\mathcal{A}} f$$
  • This construction effectively "quotients out" the internal scheme redundancy while preserving all scheme-invariant information.

3. Key Contributions

A. The Core Representation Theorem (Theorem 5.1)

This is the central result of the paper. It states that the functor of $\mathcal{A}$-balanced pairings is representable by the relative tensor product $C \otimes_{\mathcal{A}} f$.

• Universal Property: Any scheme-invariant evaluation $\Phi: C \otimes f \to \mathcal{O}$ factors uniquely through the core $C \otimes_{\mathcal{A}} f$.
• Terminality: The core is the terminal object among all quotients of the naive composite $C \otimes f$ that preserve scheme-invariant semantics.
• Parsimony: The core is "minimal" in the sense that it contains no redundancy beyond what is forced by the scheme equivalence. Any further quotient would lose physical information.

B. Canonical Induction of the Interface Algebra (Section 6)

The paper demonstrates how standard physics inputs (symmetry constraints, locality/OPE, and truncation accuracy $\alpha$) canonically induce the algebraic structures:

• Operator Sector: The truncated space of local operators $\text{Op}_\alpha$ is decomposed into irreducible representations of the symmetry group $G$.
• Interface Algebra ($\mathcal{A}_\alpha$): Defined as the commutant algebra $\text{End}_G(\text{Op}_\alpha)$. This algebra encodes all admissible finite renormalization/mixing maps that respect the symmetries and accuracy constraints.
• Minimal Closure Principle: The paper proves that for any set of "primitive" long-distance operators, the minimal sector stable under renormalization is the $\mathcal{A}_\alpha$-closure $\langle G_0 \rangle_{\mathcal{A}_\alpha}$.

C. Instantiation in DIS (Section 7)

The framework is applied to inclusive DIS:

• $x$-space: The algebra $\mathcal{A}$ is realized as matrix-valued distribution kernels under Mellin convolution.
• Moment Space: The algebra reduces to matrix algebras acting on spin-dependent operator blocks (e.g., the $2\times2$ mixing between quark singlets and gluons).
• Toy Computation: A concrete example shows how the relative tensor product collapses a pair of vectors $(C, f)$ and a mixing matrix $Z$ into a single scalar invariant, explicitly demonstrating the removal of presentation dependence.

D. Compatibility with Representations (Proposition 5.4)

The theorem proves that the core construction commutes with strong monoidal functors. This means the scheme-invariant core is independent of the representation used (e.g., $x$-space vs. Mellin moments vs. Laplace space), provided the transformation preserves the monoidal structure.

4. Results

1. Formalization of Redundancy: The paper successfully categorizes factorization scheme dependence as a module action over an interface algebra, moving the concept from a heuristic "choice" to a rigorous algebraic structure.
2. Existence of a Universal Carrier: It proves the existence of a unique, minimal object ($C \otimes_{\mathcal{A}} f$) that carries exactly the scheme-invariant physical content of a factorized observable.
3. Resolution of "Parsimony": It provides a mathematical definition of "parsimony" in physics: a representation is parsimonious if it is isomorphic to the relative tensor product, containing no internal redundancy.
4. Systematic Refinement: The framework allows for systematic refinement of accuracy (e.g., including higher twists) via filtrations, producing a tower of cores that refine the leading-order result without altering the universal property at each level.

5. Significance

• Foundational Rigor: The paper provides a rigorous mathematical foundation for a concept (scheme invariance) that is ubiquitous in QCD but often treated informally. It clarifies that $C$ and $f$ are merely "presentations" of a deeper, invariant reality.
• Bridge to Machine Learning and Symbolic Regression: The paper explicitly connects to modern data-driven physics. By defining the "core" as an abstract object independent of specific functional forms, it allows long-distance correlators (which might be represented by neural networks or complex fits) to be manipulated in a symbolic calculus that guarantees scheme invariance by construction. This is crucial for "learned" long-distance objects.
• Modular Workflow: It proposes a standardized workflow for factorization:
  1. Define the symmetry and accuracy ($G, \alpha$).
  2. Construct the interface algebra $\mathcal{A}_\alpha$.
  3. Treat data as modules.
  4. Compute the core $C \otimes_{\mathcal{A}} f$.
     This separates physical content from presentation artifacts, facilitating comparisons between different schemes or effective field theories (like SCET).
• Limitations and Scope: The authors clarify that the theorem does not prove factorization from first principles, nor does it resolve intrinsic non-perturbative ambiguities (like renormalons) that lie outside the chosen accuracy truncation. It assumes a factorized description exists and optimizes its structural representation.

In summary, this paper transforms the understanding of QCD factorization from a collection of specific formulas into a universal categorical structure, identifying the "scheme-invariant core" as the relative tensor product, thereby providing a robust tool for organizing, comparing, and extending perturbative QCD calculations.