Next-to-next-to-next-to-leading order QCD corrections to photon-pair production

This paper presents novel next-to-next-to-next-to-leading order ($\text{N}^3$LO) predictions for photon-pair production in high-energy hadron collisions, successfully demonstrating perturbative convergence and addressing significant computational challenges for Large Hadron Collider phenomenology.

The Gist

Imagine you are trying to predict exactly how two specific billiard balls will bounce off each other on a massive, chaotic pool table. But this isn't a normal pool table; it's the Large Hadron Collider (LHC), where protons smash together at near light speed, creating a storm of particles.

The paper you're looking at is a major breakthrough in predicting what happens when two protons collide and produce two isolated photons (particles of light). This is a big deal because these "photon pairs" are the background noise scientists have to filter out to find the Higgs boson (the "God particle"). If you can't predict the noise perfectly, you might miss the signal.

Here is the story of this paper, broken down into simple concepts:

1. The Problem: The "Messy Room" of Physics

In the world of quantum physics, calculations are done in layers of accuracy, like peeling an onion.

• LO (Leading Order): The basic sketch.
• NLO (Next-to-Leading): Adding more detail.
• NNLO: Getting very precise.

For a long time, scientists could only get to the NNLO layer for this specific photon collision. But here's the catch: every time they added a new layer of detail, the answer kept jumping around wildly, and the "error bars" (the margin of uncertainty) got bigger instead of smaller. It was like trying to measure a room with a ruler that kept changing length. They knew the math was right, but the calculation was too messy to trust.

2. The Solution: The "Slicing" Technique

To fix this, the authors used a clever trick called $q_T$ slicing.

Imagine you are trying to calculate the total volume of a very strange, jagged rock. It's hard to measure the whole thing at once. So, you slice it into two parts:

1. The Smooth Core: The part where the rock is nice and round. You can calculate this using a neat mathematical formula (like a recipe).
2. The Jagged Edges: The messy, spiky parts. You calculate these by simulating them on a computer, piece by piece.

The authors used this "slicing" method to separate the easy math from the hard math. They then glued the two results back together.

3. The Challenge: The "Mathematical Storm"

Even with the slicing trick, the calculation was a nightmare.

• The Cancellation Problem: Imagine you have two giant piles of sand. One pile is positive (+1,000,000 grains) and the other is negative (-1,000,000 grains). When you mix them, the result should be zero. But if your scale isn't precise enough, you might get a result of "5 grains" or "-3 grains" just because of tiny measurement errors.
• In this physics problem, the "positive" and "negative" numbers are so huge and cancel each other out so perfectly that standard computer math (like the kind your phone uses) wasn't good enough. The "noise" of the calculation was drowning out the "signal."

The Fix: The team upgraded their computers to use super-precision math (quadruple and octuple precision). Think of this as switching from a standard ruler to a laser-measuring device that can detect the width of a single atom. This allowed them to see the true result after the massive cancellations.

4. The Breakthrough: Finally, Convergence!

After years of work and using massive supercomputers (like the ones at Cambridge and Aachen), they finally reached the N3LO level (Next-to-Next-to-Next-to-Leading Order).

• The Result: For the first time, the layers of the onion stopped jumping around. The predictions stabilized. The "error bars" shrank from a huge 8% down to a tiny 3%.
• The Analogy: It's like finally tuning a radio. For years, the station was just static and noise. Now, with this new calculation, the music is crystal clear.

5. Why Does This Matter?

• Trust: It proves that our theory of how the universe works (Quantum Chromodynamics) is solid. The math finally agrees with itself.
• Discovery: Since the prediction for the "background noise" (photon pairs) is now so precise, scientists at the LHC can be much more confident when they see something new. If the data doesn't match this new, ultra-precise prediction, it's a much stronger signal that they've found a new particle or a new law of physics.
• The Future: This paper is a "proof of concept." It shows that we can now do these incredibly difficult calculations for more complex collisions, paving the way for even deeper discoveries in the future.

In a nutshell: The authors built a super-precise mathematical microscope, used a clever slicing trick to handle the messiness, and finally got a clear, stable picture of how two photons are born from a proton collision. It's a massive step forward in understanding the fundamental rules of our universe.

Technical Summary

1. Problem Statement

The production of two isolated photons ($pp \to \gamma\gamma$) at the Large Hadron Collider (LHC) is a critical process for testing the Standard Model and serves as a primary background for Higgs boson discovery ($H \to \gamma\gamma$). However, perturbative Quantum Chromodynamics (QCD) calculations for this process have historically struggled with perturbative convergence.

• The Issue: Previous Next-to-Leading Order (NLO) and Next-to-Next-to-Leading Order (NNLO) calculations showed that increasing the order of approximation did not stabilize the cross-section prediction. Instead, the central values shifted significantly outside the theoretical uncertainty bands of the previous order, and the scale uncertainty remained large (approx. 8% at NNLO).
• The Challenge: Extending calculations to Next-to-Next-to-Next-to-Leading Order (N3LO) for processes with actual $2 \to 2$ kinematics (as opposed to inclusive $2 \to 1$ processes like Higgs or Drell-Yan) is computationally formidable. It requires handling massive cancellations between real and virtual corrections, managing infrared divergences, and computing complex multi-loop amplitudes with high numerical stability.

2. Methodology

The authors employed the $q_T$-slicing method to decompose the N3LO cross-section calculation. The total cross-section is split into two regions based on the transverse momentum ($q_T$) of the photon pair relative to a cutoff parameter $q_T^{\text{cut}}$:

$$\sigma_{N3LO} = \int_0^{q_T^{\text{cut}}} dq_T \frac{d\sigma}{dq_T} + \int_{q_T^{\text{cut}}}^{\infty} dq_T \frac{d\sigma}{dq_T}$$

A. The Low-$q_T$ Region ($q_T < q_T^{\text{cut}}$)

• Factorization: This region is calculated using the factorization theorem, where the cross-section is expressed in terms of Transverse Momentum Dependent (TMD) parton distribution functions (PDFs), beam functions, and soft functions.
• Implementation: The authors performed the $b_T$ (impact parameter) and $q_T$ integrals analytically using matching kernels known up to N3LO. The result was implemented in the STRIPPER framework.
• Virtual Corrections: Three-loop virtual amplitudes for $gg \to \gamma\gamma$ were taken from existing literature.

B. The High-$q_T$ Region ($q_T > q_T^{\text{cut}}$)

• Fixed-Order Calculation: This region corresponds to the NNLO calculation of $pp \to \gamma\gamma + \text{jet}$.
• Technical Improvements: To handle the extreme numerical cancellations required for a very small $q_T^{\text{cut}}$, the authors implemented several critical upgrades to the STRIPPER code:
  • Hybrid Precision: Use of double, quadruple, and octuple floating-point precision (via the qd library) to manage cancellations.
  • Selector Function Modification: Adjusted angular hierarchies to avoid extreme collinear limits that cause numerical instability.
  • Amplitude Stability: Replaced unstable libraries (OpenLoops) with analytic expressions for six-point one-loop amplitudes.

C. Amplitude Reconstruction (Six-Point One-Loop)

A major theoretical contribution was the derivation of analytic expressions for the scattering amplitudes of $0 \to \gamma\gamma gg q\bar{q}$ and $0 \to \gamma\gamma Q\bar{Q}q\bar{q}$ at one-loop order.

• Method: They utilized rational reconstruction over finite fields.
  • Generated Feynman diagrams and performed color/Dirac algebra using FORM.
  • Reduced loop integrals to master integrals using Kira with finite-field arithmetic (FireFly).
  • Constructed compact rational ansätze using spinor products $\langle ij \rangle$ and $[ij]$.
  • Determined denominators and numerators using $p$-adic probes and partial fractioning.
• Result: These stable, compact analytic expressions were compiled into C++ code, allowing for robust evaluation in octuple precision.

3. Key Contributions

1. First N3LO Calculation for $2 \to 2$ Kinematics: This is the first N3LO QCD calculation for a process with actual $2 \to 2$ final state kinematics (diphoton production), marking a milestone beyond the inclusive $2 \to 1$ calculations (Higgs, Drell-Yan) previously achieved.
2. Demonstration of Perturbative Convergence: The study provides the first evidence that the diphoton cross-section converges at N3LO, resolving the long-standing issue of non-convergence observed at NNLO.
3. Analytic Amplitude Derivation: The derivation and public release of compact analytic expressions for six-point one-loop amplitudes, which are essential for stable high-precision calculations.
4. Algorithmic Improvements: Significant enhancements to the STRIPPER package, including hybrid precision strategies and modified subtraction selectors, enabling the handling of the severe numerical cancellations inherent in $q_T$ slicing at N3LO.

4. Results

• Cross-Section Prediction:
  The fiducial cross-section at N3LO (for $\sqrt{s}=13$ TeV) is calculated as:
  $$\sigma_{pp \to \gamma\gamma}^{N3LO} = 31.2^{+0.5}_{-0.7} \, (\text{scale}) \pm 0.6 \, (\text{stat}) \, \text{pb}$$

  • Convergence: The N3LO result is consistent with the NNLO value but shows a significant reduction in scale uncertainty.
  • Uncertainty Reduction: The total theoretical uncertainty is reduced to approximately 3%, a major improvement over the ~8% uncertainty at NNLO.
  • Comparison with Data: The result agrees well with the ATLAS measurement ($31.4 \pm 2.4$ pb), shifting the theoretical prediction slightly closer to the experimental central value.

• Differential Distributions:
  The authors provided fully differential distributions (e.g., invariant mass $m_{\gamma\gamma}$) through N3LO. These distributions show stable behavior across different phase-space regions and confirm perturbative convergence.

• Computational Cost:
  The calculation required massive computational resources (millions of CPU hours) due to the need for high-precision arithmetic and the cancellation of large positive and negative contributions in the NNLO $pp \to \gamma\gamma + \text{jet}$ sector.

5. Significance

• Theoretical Validation: This work validates the convergence of the perturbative QCD series for a complex, realistic LHC process, confirming that higher-order corrections are under control.
• Precision Physics: By reducing the theoretical uncertainty to the level of experimental systematic errors (3% vs. ~8%), this calculation enables more stringent tests of the Standard Model and better constraints on New Physics in the diphoton channel.
• Methodological Blueprint: The techniques developed (specifically the analytic reconstruction of amplitudes and the hybrid precision handling in $q_T$ slicing) provide a roadmap for future N3LO calculations for other processes involving colored particles and resolved jets, which were previously considered intractable.
• Future Outlook: While the current result is limited by statistical uncertainty from Monte Carlo integration, the framework established here paves the way for future improvements in local subtraction schemes and automation of higher-order predictions.