Tensor decomposition of $e^+e^-\to\pi^+\pi^-\gamma$ to higher orders in the dimensional regulator

This paper presents the first beyond-next-to-leading-order study of the $e^+e^-\to\pi^+\pi^-\gamma$ process by developing a complete four-dimensional tensor decomposition, evaluating one-loop polarized amplitudes analytically to higher orders in the dimensional regulator, and establishing an efficient numerical framework for five-point Feynman integrals to enable future NNLO predictions for radiative return processes.

The Gist

Imagine you are trying to measure the weight of a feather, but you are doing it while standing on a shaking, vibrating trampoline. That is essentially what physicists are trying to do when they study the collision of electrons and positrons (the antimatter twins of electrons) to create pions (heavy cousins of electrons) and a photon (a particle of light).

This paper, written by a team from the University of Liverpool, is about building a super-precise ruler to measure that feather, even while the trampoline is shaking.

Here is the breakdown of what they did, using everyday analogies:

1. The Goal: Why do we care?

Physicists are currently trying to solve a massive mystery: the "Muon Anomaly." It's like a detective trying to solve a case where the suspect (the Standard Model of physics) doesn't quite match the crime scene evidence (experimental data).

To solve this, they need to know the "hadronic vacuum polarization" contribution. Think of this as the "background noise" of the universe. To measure it, they look at a specific event: an electron and a positron smashing together to create two pions and a flash of light (a photon). This is called a "Radiative Return" process. It's like a car crash where the cars bounce off each other, but one of them throws a flare into the air. By studying that flare, they can figure out exactly how the cars hit.

2. The Problem: The "Next-to-Next-to-Leading Order" (NNLO) Challenge

Scientists have already calculated this crash with "Next-to-Leading Order" (NLO) precision. That's like measuring the crash with a standard tape measure. It's good, but not good enough to solve the mystery.

They need NNLO precision. This is like trying to measure the crash with a laser micrometer that can detect the vibration of a single atom.

• The Catch: To get this level of precision, the math gets incredibly messy. You have to account for "loops" in the calculation (virtual particles popping in and out of existence).
• The Specific Hurdle: The math requires expanding the answer into a series of tiny numbers (called the "dimensional regulator" or $\epsilon$). Most people stop at the first few numbers. This team had to calculate way further down the line (up to the third or fourth decimal place of these tiny numbers) to make the final result accurate.

3. The Solution: A New Way to Organize the Chaos

The authors faced two main problems:

1. The Math is Too Complicated: The equations describing the collision involve complex shapes (tensors) that are hard to untangle.
2. The Computer Crashes: When you try to calculate these complex shapes on a computer, the numbers often get unstable, like trying to balance a house of cards in a windstorm.

Their Innovation: The "Tensor Decomposition"
Imagine you have a giant, tangled ball of yarn (the scattering amplitude).

• Old Way: You try to pull the whole ball apart at once. It gets knotted, and you break the string.
• This Team's Way: They invented a new way to cut the yarn. They broke the giant ball down into 8 specific, simple strands (called "tensor structures").
  • They made sure these strands were "clean." They removed any "spurious" knots (mathematical artifacts that don't exist in reality but mess up the computer) that usually cause the calculation to crash.
  • They used a special language called "spinor-helicity" (think of it as a secret code) to write these strands down in the most compact, efficient way possible.

4. The Engine: The "Differential Equation" Car

Once they had the strands organized, they still needed to calculate the value of the Feynman integrals (the math that describes the probability of the crash).

• The Analogy: Imagine you need to drive from London to Edinburgh. You could try to drive straight there, but the road is full of potholes (mathematical singularities) that will break your car.
• Their Strategy: Instead of driving straight, they built a car that follows a specific set of instructions (a system of differential equations). They mapped out a "safe path" that stays strictly within the "physical region" (the part of the road where the laws of physics actually apply).
• The Result: They built a custom computer program (in C++) that drives this car. It can calculate the answer for any point in the collision zone in about 230 milliseconds (less than a quarter of a second). This is fast enough to be used in "Monte Carlo generators"—the massive computer simulations that physicists use to predict what will happen in real experiments.

5. The Verification: Did it work?

Before publishing, they had to prove their ruler was accurate.

• They compared their results against other automated tools (like a tool called GoSam).
• They checked millions of random collision scenarios.
• The Verdict: Their numbers matched perfectly, down to 12 significant digits. It's like weighing the feather and getting the exact same result as a gold-standard lab, proving their new method is rock solid.

Summary

In short, this paper is about building a high-speed, ultra-precise calculator for a specific type of particle collision.

• They reorganized the math so it doesn't crash.
• They calculated deeper into the numbers than ever before.
• They built a fast engine to run the numbers.

This work is a crucial stepping stone. It provides the "building blocks" that other scientists will use to finally solve the mystery of the Muon Anomaly, potentially revealing new physics beyond our current understanding of the universe.

Technical Summary

1. Problem Statement

The paper addresses the need for Next-to-Next-to-Leading Order (NNLO) theoretical predictions for the radiative return process $e^+e^- \to \pi^+\pi^-\gamma$.

• Context: Precision measurements of this process at low energies are critical for determining the hadronic vacuum polarization contribution to the muon anomalous magnetic moment ($(g-2)_\mu$). Current tensions between data-driven theory and Lattice QCD require unprecedented theoretical accuracy.
• Challenge: While Next-to-Leading Order (NLO) predictions exist, achieving NNLO accuracy requires:
  1. Evaluating genuine two-loop corrections.
  2. Controlling the structure of one-loop amplitudes expanded to higher orders in the dimensional regulator ($\epsilon$) (specifically up to $\mathcal{O}(\epsilon^2)$).
  3. Handling the complexity of five-point amplitudes involving multiple kinematic scales (electron and pion masses) and ensuring numerical stability for integration into Monte Carlo (MC) event generators.
• Specific Obstacle: Standard tensor decompositions often introduce spurious Gram determinants in the coefficients of form factors, leading to numerical instabilities in the physical region.

2. Methodology

The authors developed a systematic framework combining modern amplitude techniques with robust numerical integration strategies.

A. Tensor Decomposition and Polarized Amplitudes

• 4D Tensor Decomposition: Instead of working in $D$ dimensions, the authors exploited the fact that external states live in 4 dimensions. They constructed a basis of 8 independent tensor structures ($T_1$ to $T_8$) using spinors, gamma matrices, and polarization vectors.
• Spinor-Helicity Formalism: Massive momenta were decomposed into massless ones ($p^\flat_i$) to utilize the spinor-helicity formalism. This allowed for the construction of polarized amplitudes for all 8 non-vanishing helicity configurations.
• Stability Optimization: A key innovation was the organization of polarized amplitudes ($A_h = \Phi_h \tilde{A}_h$) such that spurious Gram determinants in the denominators cancel out explicitly. This ensures numerical stability across the physical production region.

B. Renormalization and Subtraction

• UV Renormalization: The bare form factors were renormalized using on-shell wave-function and mass renormalization constants for leptons, pions, and photons.
• IR Subtraction: Infrared (IR) poles were handled using an IR operator ($I$) acting on tree-level amplitudes. The authors organized the calculation so that UV and IR pole cancellations occur directly at the integral level, avoiding the need to explicitly reconstruct poles during numerical evaluation over finite fields.

C. Feynman Integral Computation

• Integral Families: The one-loop diagrams (81 contributing diagrams after Furry's theorem cancellations) were reduced to two independent integral families: Box and Pentagon (5-point).
• Master Integrals (MIs): Using Integration-by-Parts (IBP) identities (via LiteRed) and finite field reconstruction (via FiniteFlow), the integrals were reduced to a set of 38 Master Integrals.
• Differential Equations: A canonical system of differential equations was constructed for these MIs. The system involves a kinematic alphabet of 119 letters (39 rational, 80 algebraic).
• Numerical Integration: The differential equations were solved numerically using a custom C++ implementation. To navigate the physical region without crossing branch cuts, the authors employed a complex deformation of the integration path (homotopic to a straight line) and adjusted Riemann sheets dynamically.

3. Key Contributions

1. First Study Beyond NLO: This is the first systematic study of $e^+e^- \to \pi^+\pi^-\gamma$ beyond NLO, providing the necessary building blocks for future NNLO calculations.
2. Stable Tensor Basis: The development of a 4D tensor decomposition that eliminates spurious Gram determinants, ensuring the amplitude is numerically stable and suitable for MC generators.
3. High-Order $\epsilon$ Expansion: Analytical evaluation of one-loop polarized amplitudes up to $\mathcal{O}(\epsilon^2)$, which is a prerequisite for NNLO accuracy.
4. Efficient Numerical Framework: A custom C++ routine capable of evaluating the resulting transcendental functions (up to weight 4) with high precision.
   • Performance: Achieves evaluation times of ~230 ms per phase-space point (with ~0.5% of points taking longer due to proximity to singularities).
5. Validation: The results were cross-checked against:
   • AMFlow: For the bare form factors (perfect agreement).
   • GoSam-3.0: For the finite parts of the interference terms (agreement at the level of $10^{-12}$ significant digits).
   • DiffExp: For the numerical integration of the differential equations.

4. Results

• Analytic Structure: The authors provided the complete analytic structure of the one-loop form factors, expressed in terms of transcendental functions up to weight four.
• Functional Complexity: They identified that the pentagon family (5-point) requires transcendental functions up to weight 4, while box families are simpler.
• Numerical Stability: The polarized amplitude approach successfully removed numerical instabilities associated with $\text{tr}_5^2$ (Gram determinants) that typically plague form-factor-based approaches.
• Implementation Readiness: The framework is designed for direct implementation in Monte Carlo event generators, bridging the gap between high-precision analytic calculations and phenomenological applications.

5. Significance

• Muon $g-2$ Precision: By enabling NNLO predictions for radiative return processes, this work directly contributes to reducing theoretical uncertainties in the determination of the hadronic vacuum polarization, which is crucial for resolving the tension in the muon anomalous magnetic moment.
• Methodological Advancement: It demonstrates that low-energy processes with multiple mass scales can be treated with the same modern amplitude techniques (tensor decomposition, differential equations, symbol maps) used in high-energy physics, provided that numerical stability is carefully engineered.
• Future Roadmap: This work lays the foundation for computing the full two-loop corrections to the radiative return process and implementing soft-photon resummation within the Coherent Exclusive Exponentiation (CEE) framework.

In summary, this paper provides the essential analytic and numerical infrastructure required to push the precision of low-energy $e^+e^-$ annihilation calculations to the NNLO level, addressing both the analytic complexity of five-point amplitudes and the computational demands of Monte Carlo integration.