Low-energy $e^+\,e^-\toγ\,γ$ at NNLO in QED

This paper presents a fully differential next-to-next-to-leading order QED calculation of the $e^+e^- \to \gamma\gamma$ process, implemented in the McMule framework to enable precise luminosity measurements for electron-positron colliders with center-of-mass energies up to a few GeV.

The Gist

Imagine you are trying to measure the exact size of a room, but the room is filled with invisible, bouncing balloons (photons) that keep popping in and out of existence. To get a perfect measurement, you need to account for every single bounce, every pop, and every ripple they cause.

This paper is about scientists doing exactly that, but instead of a room, they are studying the "room" inside a particle collider where electrons and positrons (matter and anti-matter) crash into each other and turn into two flashes of light (photons).

Here is the story of their work, broken down into simple concepts:

1. The Goal: Measuring the "Luminosity"

In the world of particle physics, "luminosity" is like the brightness of a lightbulb, but for a collider. It tells you how many collisions are happening. If you want to know if you've discovered a new particle, you first need to know exactly how many times you've looked.
The process of an electron and positron turning into two photons ($e^+ e^- \to \gamma \gamma$) is the "standard candle" of this field. It's a very clean, predictable event that scientists use as a ruler to measure the brightness of their experiments.

2. The Problem: The "Perfect" Calculation is Hard

For a long time, scientists had a ruler that was accurate to the "Next-to-Leading Order" (NLO). Think of this as a ruler accurate to the nearest millimeter. It's good, but for the ultra-precise experiments happening today (like the KLOE and Belle II experiments mentioned in the paper), they need a ruler accurate to the nearest micrometer.

They needed to go one step further: NNLO (Next-to-Next-to-Leading Order). This means calculating not just the main event, but also the tiny, complex ripples and secondary effects that happen after the main crash.

3. The Solution: The "McMule" Super-Computer

The team, led by researchers from Switzerland, Italy, and the UK, used a powerful software framework called McMule. You can think of McMule as a highly sophisticated simulation engine.

• The Old Way: Previous calculations were like using a blurry camera. They could guess the general shape of the event but missed the fine details.
• The New Way: This paper presents a "4K Ultra-HD" calculation. They didn't just guess; they calculated every possible way the particles could interact, including:
  • Pure Light: Photons interacting with other photons.
  • Ghost Particles: Virtual particles that pop in and out of existence (called "loops") that briefly change the energy of the collision.
  • Heavy Hitters: Even though they are looking at low-energy collisions, they checked if heavy particles (like muons or hadrons) could sneak in and mess up the math. They found that for these low energies, the heavy particles are too lazy to show up, so they only needed to worry about the lightest ones (electrons).

4. The Analogy: The Traffic Jam

Imagine a highway where cars (electrons) are driving toward each other.

• Leading Order (LO): Two cars crash and bounce off. Simple.
• NLO: The cars crash, but a few other cars (photons) get scared and swerve nearby. You have to account for that.
• NNLO (This Paper): The cars crash, swerving cars cause other cars to brake, which causes a ripple effect of honking and dust clouds. The scientists calculated the honking, the dust, and the exact path of every swerving car.

They compared their new, ultra-precise calculation with the old "blurry" method (which used a "Parton Shower" approach—basically a statistical guess for the swerving cars).

• The Result: The two methods agreed almost perfectly (within 0.1%). This is huge! It means the old "blurry" ruler was actually very good, but now we have a "laser" ruler to be absolutely sure.

5. Why Does This Matter?

• For Dark Matter Hunters: If you are looking for a "Dark Photon" (a ghostly particle that might explain dark matter), you need to know the background noise perfectly. If your ruler is off by 1%, you might think you found a ghost when it was just a measurement error. This paper tightens the ruler.
• For Future Experiments: As experiments get more powerful, they need better math. This paper provides the "gold standard" math for low-energy collisions.
• The "McMule" Library: The scientists didn't just write a paper; they put their code into a public library. Any other scientist can now download this "laser ruler" and use it for their own experiments.

Summary

In short, these scientists built the most precise mathematical map ever created for a specific type of particle crash. They proved that their new map matches the old maps in the places they overlap, giving everyone confidence that they are ready to hunt for new physics with extreme precision. They turned a "good enough" measurement into a "perfect" one.

Technical Summary

1. Problem Statement

The annihilation of an electron-positron pair into two photons ($e^+ e^- \to \gamma \gamma$) at low center-of-mass energies (up to a few GeV) is a fundamental Quantum Electrodynamics (QED) process. It serves two critical roles:

• Luminosity Measurement: It is a standard candle for determining the luminosity of electron-positron colliders (e.g., KLOE, Belle II).
• Background Process: It acts as a significant background in searches for dark photons.

While Next-to-Leading Order (NLO) QED corrections have been known for decades, precise experimental comparisons require fully differential calculations. The current state-of-the-art tool, BabaYaga, provides NLO results combined with parton-shower (PS) resummation (NLO+PS). However, this approach lacks complete Next-to-Next-to-Leading Order (NNLO) terms, specifically non-logarithmic $\mathcal{O}(\alpha^2)$ corrections, vacuum polarization (VP) effects, and light-by-light (LbL) scattering contributions. Conversely, fixed-order NNLO calculations often miss the resummation of collinear logarithms found in parton showers. There is a need for a complete, fixed-order NNLO calculation to cross-check theoretical uncertainties and improve precision for luminosity measurements.

2. Methodology

The authors performed a fully differential calculation using the McMule framework, which specializes in multi-loop QED computations for $2 \to 2$ processes.

• Framework & Tools:

  • The calculation utilizes the OpenLoops library for real-virtual corrections.
  • Infrared (IR) singularities are handled using the FKS subtraction scheme (Frixione-Kunszt-Signer).
  • Renormalization is performed in the on-shell scheme.

• Decomposition of Corrections:
  The NNLO cross-section ($\sigma^{(2)}$) is decomposed into:

  1. Purely Photonic Corrections ($\sigma^{(2,\gamma)}$): Includes double-virtual, real-virtual, and double-real contributions.
     • Mass Effects: For the double-virtual amplitude, the authors use "massified" amplitudes derived from massless fermion results, neglecting terms polynomially suppressed by the electron mass ($m_e$). This is justified as $\sqrt{s} \gg m_e$.
     • Real-Virtual: Evaluated with full mass dependence.
  2. Fermion-Loop Corrections:
     • Vacuum Polarization (VP): Includes electron loops ($\sigma^{(2,VPe)}$), muon/tau loops ($\sigma^{(2,VP\mu\tau)}$), and hadronic loops ($\sigma^{(2,VPh)}$). The hadronic contribution is calculated using a dispersive approach via the alphaQED23 package.
     • Light-by-Light (LbL): Includes double-box diagrams ($\sigma^{(2,LbL)}$) and real corrections with three photons ($\sigma^{(2,rLbL)}$).
  3. Approximations:
     • Only electron loops are considered for LbL contributions; contributions from heavier leptons ($\mu, \tau$) and hadrons in LbL loops are neglected due to strong suppression at low energies.
     • Massless electrons are used for LbL loop calculations.

3. Key Contributions

• First Fully Differential NNLO QED Calculation: This work presents the first complete, fully differential NNLO calculation for $e^+ e^- \to \gamma \gamma$ in QED, including all photonic corrections and dominant fermion-loop effects.
• Integration into McMule: The process is implemented in the McMule user library, completing the set of NNLO calculations for the most important $2 \to 2$ processes in the framework.
• Complementarity to NLO+PS: The authors explicitly contrast their fixed-order NNLO results with the NLO+PS approach (BabaYaga). This allows for a mutual cross-check:
  • McMule provides exact $\mathcal{O}(\alpha^2)$ terms missing in NLO+PS.
  • NLO+PS provides resummation of leading logarithms beyond NNLO missing in the fixed-order calculation.
• Massive vs. Massless Treatment: The paper addresses the technical challenge of external fermion masses in QED (unlike QCD where quarks are often treated as massless), demonstrating that massified amplitudes are sufficient for the energy regime of interest.

4. Results

The results are validated against analytic NLO results and compared with the NLO+PS predictions from reference [12] (BabaYaga).

• Total Cross Section:

  • Calculated for $\sqrt{s} \in \{1, 3, 10\}$ GeV with specific kinematic cuts (photon energy $E_\gamma > 0.3\sqrt{s}$, angular range $45^\circ < \theta < 135^\circ$, acollinearity $\xi < 10^\circ$).
  • Agreement: The NLO results agree with previous literature within $0.01\%$.
  • NNLO Impact: The total NNLO correction is between 6% and 8% relative to the Born level.
  • Comparison with NLO+PS: The difference between the exact NNLO photonic corrections and the parton-shower induced corrections (beyond NLO) is small, roughly 0.05%. This confirms the theoretical uncertainty estimate of 0.1% given in previous studies.
  • Dominant Corrections: The electron vacuum polarization ($\sigma^{(2,VPe)}$) is the largest non-photonic correction (negative, $\sim 20\%$ of the photonic NNLO term). Muon/tau and hadronic VP contributions are negligible (factor of 100 smaller). Light-by-light contributions are also negligible.

• Differential Distributions:

  • KLOE-like Scenario ($\sqrt{s}=1$ GeV): NNLO corrections are a few permille, with enhancements up to 2% at the distribution edges. Non-photonic corrections are negligible compared to photonic ones.
  • Belle II-like Scenario ($\sqrt{s}=10.58$ GeV): NNLO corrections remain at the permille level. Non-photonic corrections (specifically electron VP) grow to $\sim 0.1\%$, becoming relevant for high-precision goals.

5. Significance

• Precision Luminosity: The calculation provides the necessary theoretical precision for luminosity measurements at low-energy $e^+ e^-$ colliders, reducing theoretical uncertainties to the permille level.
• Validation of NLO+PS: The comparison confirms that NLO+PS approaches (like BabaYaga) are highly reliable, with missing higher-order effects impacting results only at the permille level.
• Framework Completeness: This work solidifies McMule as a comprehensive tool for precision QED phenomenology, covering the most critical $2 \to 2$ scattering processes at NNLO.
• Future Prospects: The authors note that the techniques used here can be extended to $2 \to 3$ processes and Compton scattering (via crossing symmetry), paving the way for even more complex precision calculations.

In summary, this paper delivers a high-precision, fully differential NNLO QED prediction for diphoton production, validating existing phenomenological tools while providing the exact fixed-order terms required for the next generation of low-energy collider experiments.