Towards a Fully Automated Differential $\text{NNLO}_\text{EW}$ Generator for Lepton Colliders

This paper presents a solution for achieving fully automated, process-independent $\text{NNLO}_\text{EW}$ precision for future lepton colliders by utilizing the Yennie-Frautschi-Suura theorem to combine local infrared subtraction with all-order resummation of soft and soft-collinear logarithms.

The Gist

Imagine you are trying to predict exactly how a crowd of people will move through a busy train station. If you only look at the main flow of people, your prediction is okay. But if you want to predict the exact path of every single person, including the tiny shoves, the accidental bumps, and the way people slow down to check their phones, you need a much more sophisticated model.

This paper is about building that super-sophisticated model for the world's most powerful particle colliders, specifically the future ones that smash electrons and positrons together.

Here is the breakdown of what the authors, Alan Price and Frank Krauss, have achieved, using simple analogies:

The Problem: The "Static Noise" of the Universe

When scientists smash particles together at these colliders, they hope to see new, rare events. But the universe is messy. As soon as particles interact, they emit a swarm of "soft" photons (particles of light). Think of these photons like static noise on a radio or dust motes dancing in a sunbeam.

• The Old Way: Previous computer programs (generators) could handle the big, loud interactions well. But when it came to the tiny, constant "static" (the soft photons), they struggled. They had to use a "slicing" method: cutting the data into neat chunks to avoid mathematical errors. This was like trying to measure a messy room by only counting the furniture and ignoring the dust. It worked, but it wasn't precise enough for the next generation of experiments.
• The Goal: The new experiments will be so precise that the "dust" (the soft photons) matters. If the theory doesn't account for every single photon, the predictions will be wrong, and scientists might miss a discovery.

The Solution: The "YFS" Magic Trick

The authors present a new way to handle this mess, based on a mathematical theorem called Yennie-Frautschi-Suura (YFS).

Think of the YFS theorem as a magic noise-canceling headset for particle physics.

• Instead of trying to calculate every single photon interaction one by one (which creates infinite mathematical errors), the YFS method reorganizes the math.
• It takes all the "infinite noise" (the divergences) and subtracts it out before doing the hard calculations.
• It then "resums" (adds up) the important effects of all those photons into a smooth, manageable formula.

The authors have taken this method, which was previously only used for very specific, simple scenarios, and turned it into a fully automated machine. They built it into a software package called SHERPA.

What They Actually Did (The "How")

The paper details how they automated this process to reach a level of precision called NNLOEW (Next-to-Next-to-Leading Order in Electroweak corrections).

1. The "Subtraction" Engine: They created a system that automatically identifies the "infinite" parts of the math and subtracts them locally. Imagine trying to balance a scale. If you have a heavy weight on one side (the real physics) and a heavy weight on the other (the mathematical error), they cancel out perfectly, leaving you with the true, finite answer. They proved this works for complex scenarios with many particles.
2. Handling the "Double Trouble": They successfully automated the calculation for when two photons are emitted at once (Double Real) or when a photon is emitted while a loop of virtual particles is involved (Real-Virtual). This is like handling a traffic jam where two cars swerve at the exact same time; the math gets incredibly complicated, but their code handles it automatically.
3. The Missing Piece (The "Two-Loop" Bottleneck): The only part they couldn't fully automate yet is the "Double-Virtual" correction (where two loops of virtual particles interact). This is because there isn't a public tool yet that can automatically calculate these specific two-loop diagrams. However, they built the framework so that as soon as such a tool exists, their system can plug it in immediately. For now, they tested this part on simple processes where the answers are already known from other papers.

The Results: A Clearer Picture

They tested their new "YFSNLOEW" and "YFSNNLOEW" tools against standard methods and found:

• Better Precision: The new method reduces the uncertainty in predictions from about 2.5% down to 0.1% for certain processes. That's like going from guessing the weight of a person within a few pounds to guessing within a few ounces.
• Stability: The math is much more stable. Old methods sometimes produced "negative weights" (mathematical nonsense that has to be discarded), which slows down simulations. The new method produces fewer of these, making the computer run faster and more efficiently.
• Versatility: They showed it works for various scenarios, from creating pairs of muons (heavy electrons) to creating pairs of pions (particles made of quarks). They even compared their predictions for pion production against real data from the BESIII experiment, and the match was excellent.

The Bottom Line

This paper doesn't claim to have discovered a new particle or solved a medical mystery. Instead, it provides the ultimate ruler and calculator for future particle physics experiments.

By automating the handling of "soft photons" and pushing the precision to the NNLOEW level, they have ensured that when the next generation of lepton colliders (like the FCC-ee or ILC) comes online, the theoretical predictions will be sharp enough to match the incredible precision of the machines. They have essentially upgraded the software that tells scientists what to expect, so that when the real data arrives, any deviation will be a genuine sign of new physics, not just a glitch in the math.

Technical Summary

1. Problem Statement

Future lepton collider experiments (e.g., FCC-ee, ILC, CEPC) aim for unprecedented precision in measuring electroweak (EW) observables and probing the Higgs boson. To match this experimental precision, theoretical predictions must reach Next-to-Next-to-Leading Order in the Electroweak coupling (NNLOEW).

Current challenges include:

• Systematic Automation: While Next-to-Leading Order (NLO) EW calculations have been automated for various processes, extending this to NNLOEW is difficult due to the complexity of two-loop amplitudes and the lack of process-independent subtraction schemes.
• Infrared (IR) Divergences: Higher-order calculations involve soft and collinear photon emissions which generate IR divergences. These must be cancelled between real emission and virtual loop contributions.
• Resummation vs. Fixed-Order: Fixed-order predictions alone are insufficient because large logarithms (e.g., $\alpha \log(Q^2/m^2)$) can spoil accuracy. These must be resummed to all orders. Existing solutions often rely on process-specific codes (like KKMC) or amplitude-level subtraction (Coherent Exclusive Exponentiation, CEEX), limiting their generality.
• Matching: There is a need for a framework that systematically includes NNLOEW corrections while matching them to all-order resummation of soft/collinear logarithms in a fully automated, process-independent Monte Carlo (MC) generator.

2. Methodology

The authors propose a solution based on the Yennie-Frautschi-Suura (YFS) theorem, implemented within the SHERPA event generator.

Core Theoretical Framework

• YFS Theorem: The perturbative expansion is reordered to exponentiate IR divergences arising from soft photon emissions. The differential cross-section is expressed as:
  $$d\sigma^{(\infty)} = e^{Y(\Omega)} \sum_{n_\gamma=0}^\infty \frac{1}{n_\gamma!} \left( \prod d\Phi_\gamma \tilde{S}(k) \right) \left( \tilde{\beta}_0 + \sum \tilde{\beta}_1 + \dots \right)$$
  Here, $Y(\Omega)$ is the YFS form factor (resumming IR divergences), and $\tilde{\beta}_n$ are the finite residuals (higher-order corrections) that remain after IR subtraction.
• Local Subtraction: Unlike traditional schemes (e.g., Catani-Seymour) that rely on the Kinoshita-Lee-Nauenberg (KLN) theorem to cancel poles between separate integrals, the YFS scheme constructs local subtraction terms such that each individual residue ($\tilde{\beta}_n$) is IR finite.
• Automation Strategy:
  • NLOEW: Automated for arbitrary processes using standard one-loop tools (OPENLOOPS, RECOLA).
  • NNLOEW:
    • Real-Real (RR): Fully automated using tree-level matrix elements.
    • Real-Virtual (RV): Automated using one-loop amplitudes.
    • Double-Virtual (VV): Currently limited to processes where two-loop amplitudes are available in the literature or via specific packages (e.g., GRIFFIN for $Z$-pole physics), as a fully automated two-loop tool for arbitrary EW processes is not yet public.

Technical Implementation Details

• Momentum Mappings: The authors developed specific phase-space mappings to handle the transition between the global event phase-space (with $n$ photons) and the local phase-spaces required for evaluating subtraction terms (where photon momenta vanish). This ensures the subtraction terms are evaluated at the correct kinematic points to cancel singularities.
• Subtraction Terms:
  • Virtual Terms: Constructed by subtracting the YFS form factor contribution from the one-loop amplitude.
  • Real Terms: Constructed by subtracting the eikonal factor times the lower-order finite residual from the real emission amplitude.
  • Recurrence: The scheme is recursive; the $\tilde{\beta}_1$ (NLO finite) term is used to construct the subtraction for the $\tilde{\beta}_2$ (NNLO) terms.
• Regularization: The implementation supports both Dimensional Regularization (DR) and massive regularization (fictitious photon mass), allowing for cross-checks.

3. Key Contributions

1. First Fully Automated NNLOEW Matching: This work presents the first implementation of a fully automated matching of YFS resummation to NNLOEW corrections for arbitrary processes in a general-purpose MC generator (SHERPA).
2. Process-Independent Framework: Unlike previous CEEX-based approaches which were process-specific, this method constructs finite remainders at the level of squared amplitudes, making it applicable to any process supported by SHERPA.
3. Recursive Subtraction Scheme: The authors demonstrated how to construct subtraction terms for Double-Real and Real-Virtual corrections using the finite residuals from lower orders, ensuring IR finiteness without introducing ad-hoc terms.
4. Integration with GRIFFIN: The authors interfaced the GRIFFIN package (providing two-loop QED corrections for $e^+e^- \to f\bar{f}$) with SHERPA to achieve complete NNLOEW accuracy for $Z$-pole processes, including missing IR-finite contributions previously absent in other generators like KKMC.
5. Validation: The method was validated against fixed-order Catani-Seymour (CS) NLOEW calculations and existing literature, showing excellent agreement.

4. Results

The paper presents several validation and phenomenological results:

• IR Pole Cancellation:
  • For NLOEW and NNLOEW (Real-Virtual), the cancellation of IR poles (both single and double poles in DR) was verified to 13–18 decimal digits, confirming the correctness of the subtraction terms.
  • The subtraction terms were shown to be numerically stable down to photon energies of $10^{-9}$ GeV.
• Cross-Section Comparisons ($e^+e^- \to \mu^+\mu^-\tau^+\tau^-$):
  • At 250 GeV, the YFSNLOEW predictions agreed excellently with the standard CS NLOEW approach.
  • NLOEW corrections induced a $\sim 14\%$ reduction in the fiducial cross-section compared to Leading Order (LO).
• NNLOEW Precision ($e^+e^- \to \mu^+\mu^-$):
  • Uncertainty Reduction: The perturbative uncertainty for the total cross-section around the $Z$-pole dropped from $\approx 2.5\%$ (at YFSNLOEW) to $\approx 0.1\%$ (at YFSNNLOEW).
  • Resonance Structure: The differential distributions near the $Z$-pole and at the $WW$ threshold showed that YFSNNLOEW provides a stable prediction, whereas YFSNLOEW showed larger uncertainties away from the pole due to uncorrected hard photon emissions.
• Application to Other Processes:
  • The framework was successfully applied to pion production ($e^+e^- \to \pi^+\pi^-$), demonstrating its utility for scalar production and low-energy experiments (e.g., BESIII data comparison).
  • It is applicable to fixed-target experiments (MUonE, MOLLER) and low-energy scans.

5. Significance

• Future Collider Readiness: This work provides the necessary theoretical infrastructure to analyze data from future lepton colliders (FCC-ee, ILC) with the required sub-percent precision.
• Automation Milestone: By moving from amplitude-level (CEEX) to squared-amplitude level automation, the authors have removed the bottleneck of process-specific coding, allowing for the rapid inclusion of NNLOEW corrections for complex multi-particle final states.
• Negative Weight Reduction: The YFS approach significantly reduces the number of negative weights in event generation compared to traditional subtraction methods, improving computational efficiency for high-statistics simulations.
• Path Forward: While currently limited by the lack of a fully automated two-loop amplitude generator for arbitrary processes, the authors have established the complete framework. Once such tools become available, SHERPA can be immediately upgraded to full YFSNNLOEW accuracy for any process.

In conclusion, this paper represents a major step forward in precision physics for lepton colliders, delivering a robust, automated, and process-independent generator capable of handling the complex interplay of fixed-order NNLOEW corrections and all-order QED resummation.