NNLOJET: a parton-level event generator for jet cross sections at NNLO QCD accuracy

This paper introduces the open-source NNLOJET parton-level event generator, which implements the antenna subtraction method to compute jet cross sections and related observables at NNLO QCD accuracy across electron-positron, lepton-hadron, and hadron-hadron collisions.

The Gist

The Big Picture: Building a Better Crystal Ball

Imagine you are a physicist trying to predict what will happen when two tiny particles smash into each other at nearly the speed of light. You have a set of rules (the Standard Model) that tells you how these particles should behave. But nature is messy. When particles collide, they don't just bounce off; they spray out a chaotic shower of new particles, like confetti exploding from a cannon.

To make a truly accurate prediction, you can't just guess the main explosion. You have to calculate the tiny, almost invisible ripples and secondary sprays that happen because of the explosion. In the world of particle physics, these calculations are called NNLO (Next-to-Next-to-Leading Order). They are incredibly complex, like trying to predict the exact path of every single grain of sand in a hurricane.

This paper introduces NNLOJET, a new, open-source computer program designed to do these super-precise calculations for "jets" (sprays of particles). It's like giving scientists a high-definition crystal ball that can see details the old models missed.

The Core Problem: The "Infinite" Mess

When physicists try to calculate these collisions, they run into a mathematical nightmare called "infrared singularities."

• The Analogy: Imagine trying to count the number of people in a room, but every time someone whispers, the room gets infinitely louder. If you try to add up the noise, the number goes to infinity.
• The Reality: In particle physics, when particles move very slowly or very close together, the math blows up to infinity. Since the real world doesn't have infinite energy, these infinities are just errors in the calculation method.

To fix this, the authors use a method called Antenna Subtraction.

• The Analogy: Think of it like noise-canceling headphones. The headphones listen to the annoying background noise (the mathematical infinities) and generate a "negative" sound wave to cancel it out perfectly.
• How it works: The program calculates the messy, infinite parts separately and then subtracts them out, leaving only the clean, finite, real-world answer.

What is NNLOJET?

NNLOJET is the "engine" that runs these calculations. Before this paper, many of these calculations were like secret recipes locked in a single chef's kitchen. If you wanted to use them, you had to beg the chef or build your own kitchen from scratch.

NNLOJET changes the game by being open-source.

• The Analogy: It's like releasing the full recipe book and the kitchen tools to the public. Anyone with a computer can download it, cook the meal, and check the ingredients.
• The "Jet" Part: The program specializes in "jets." In particle physics, when a quark or gluon (the building blocks of matter) is shot out, it doesn't stay alone. It instantly turns into a spray of particles. We call this spray a "jet." NNLOJET predicts exactly how big, how fast, and in what direction these sprays will go.

How Do You Use It? (The Runcard)

You don't need to be a coding wizard to use NNLOJET. The paper explains that you control the program using a Runcard.

• The Analogy: Think of the Runcard as a flight plan or a recipe card. You don't need to know how the airplane engine works; you just need to tell the pilot where to go.
• What you write: You tell the program:
  • Where: Are we smashing protons (like at the Large Hadron Collider) or electrons?
  • What: Are we looking for a Z-boson, a Higgs boson, or just a spray of jets?
  • Rules: How big should the jets be? How fast?
  • Output: What charts do you want to see at the end?

The paper provides a detailed manual (Sections 5 and 6) on how to write this "flight plan," covering everything from the type of collision to the specific math settings.

The Workflow: The Assembly Line

Calculating these collisions is so heavy that one computer can't do it alone. It would take years.

• The Analogy: Imagine you need to paint a massive mural. One person would take a lifetime. Instead, you hire a crew of 100 painters.
• The Tool: The paper describes a workflow script (nnlojet-run) that acts as the foreman.
  1. Warm-up: The foreman sends out a few painters to test the wall and figure out the best way to paint (this is the "VEGAS grid" adaptation).
  2. Production: Once the plan is set, the foreman sends out hundreds of painters to do the actual work simultaneously.
  3. Finalize: When everyone is done, the foreman collects all the pieces, checks for mistakes (outliers), and glues them together into one perfect picture.

What Can It Calculate?

The paper lists the specific "recipes" currently available in the cookbook (Version 1.0.0):

• Electron-Positron Collisions: Like the old LEP collider.
• Proton-Proton Collisions: Like the current Large Hadron Collider (LHC).
• Specific Events: It can predict the creation of:
  • Z-bosons (heavy cousins of the photon).
  • W-bosons (particles that carry the weak force).
  • Higgs bosons (the particle that gives mass to others).
  • Photons (light) and jets.
  • Combinations of all the above (e.g., a Higgs boson plus a jet).

Why Does This Matter?

The paper concludes that by making this code open and user-friendly, it allows scientists to compare their theoretical predictions directly with real experimental data.

• The Analogy: If you are building a model car, you need to know if it will actually drive. NNLOJET provides the most accurate "wind tunnel" test possible. If the real data matches the NNLOJET prediction, our understanding of the universe is correct. If they don't match, it might mean we found something new!

In short: This paper releases a powerful, free, and user-friendly tool that helps physicists calculate the messy details of particle collisions with extreme precision, ensuring our map of the universe is as accurate as possible.

Technical Summary

Technical Summary of NNLOJET: A Parton-Level Event Generator for Jet Cross Sections at NNLO QCD Accuracy

Problem Statement
Precision tests of the Standard Model (SM) rely heavily on comparing experimental measurements of hadronic final states (jets) with theoretical predictions. These predictions require the calculation of Next-to-Next-to-Leading Order (NNLO) Quantum Chromodynamics (QCD) corrections to handle the large strong coupling constant ($\alpha_s$) and the complex infrared (IR) singularity structure arising from soft and collinear radiation. While Next-to-Leading Order (NLO) calculations have been largely automated, NNLO calculations remain challenging due to the need to cancel IR divergences between double-virtual (VV), mixed real-virtual (RV), and double-real (RR) contributions. Existing NNLO codes are often process-specific, not broadly available as open-source, or require significant manual tuning for each observable. There is a need for a flexible, open-source parton-level event generator capable of computing jet cross sections at NNLO accuracy across various collider environments (electron–positron, lepton–hadron, and hadron–hadron).

Methodology: Antenna Subtraction
The NNLOJET code implements the antenna subtraction method to handle IR singularities. This method derives subtraction terms based on the universal factorization of QCD real-radiation amplitudes in their IR limits.

• Core Mechanism: The method identifies unresolved real-radiation configurations with pairs of hard radiator partons. The radiation between them is described by "antenna functions."
• Subtraction Terms: Unintegrated subtraction terms are constructed from products of antenna functions and reduced matrix elements of lower partonic multiplicity. These terms are integrated over the antenna phase space to produce integrated subtraction terms ($d\hat{\sigma}_T$ and $d\hat{\sigma}_U$) that explicitly cancel the $\epsilon$-poles from virtual corrections.
• Implementation: The code utilizes a phase-space mapping to factorize the kinematics of the antenna functions from the reduced matrix elements. The implementation covers final–final, initial–final, and initial–initial radiator configurations.
• Computational Framework: NNLOJET is a parton-level event generator written primarily in modern Fortran, with C++ and Python modules. It uses the adaptive Monte Carlo routine VEGAS for phase-space integration. The code includes libraries for evaluating special functions (harmonic polylogarithms) via HPLOG, TDHPL, and CHAPLIN, and interfaces with LHAPDF for parton distribution functions (PDFs).

Key Contributions and Features
The paper presents the open-source release (v1.0.0) of the NNLOJET code, distributed under the GNU General Public License v3.0. Key technical contributions include:

1. Broad Process Coverage: The code supports NNLO calculations for jet production in:

   • Electron–positron collisions: $e^+e^- \to 2j$ and $3j$.
   • Lepton–hadron collisions: Neutral-current (NC) and charged-current (CC) DIS processes ($ep \to \ell + 2j$, $ep \to \nu + 2j$, etc.).
   • Hadron–hadron collisions: Inclusive jet production ($pp \to j$), dijet production ($pp \to 2j$), vector-boson plus jet ($V+1j$), photon plus jet ($\gamma+1j$), di-photon ($\gamma\gamma$), and Higgs-boson plus jet ($H+1j$) production.
   • Derived Orders: The framework implicitly provides NLO corrections for processes with one additional jet and LO for two additional jets (e.g., calculating $Z+2j$ at NLO via the $ZJ$ process with a 2-jet requirement).

2. Flexible User Interface (Runcard): The code is controlled via a plain-ASCII "runcard" file, allowing users to define:

   • Process and Collider: Selection of collider type (pp, ep, $e^+e^-$), beam energies, and jet algorithms (anti-$k_T$, Cambridge-Aachen, $k_T$, Durham, Jade).
   • Observables and Selections: Definition of fiducial cuts, object-level selectors (jet $p_T$, rapidity), and event-level selectors.
   • Histograms: Flexible binning (linear, logarithmic, custom edges) and multi-differential distributions via histogram selectors.
   • Scales: Simultaneous evaluation of multiple renormalization and factorization scale combinations for uncertainty estimation.

3. Automated Workflow: The distribution includes a Python-based workflow script (nnlojet-run) to manage job submission, resource allocation, and result merging. This workflow handles:

   • Warmup and Production Phases: Adaptive grid generation followed by production runs.
   • Resource Management: Dynamic adaptation of computing resources to reach a target relative accuracy.
   • Result Merging: A robust algorithm (involving trimming, k-scan, and weighted combination) to merge partial results from independent jobs while mitigating the impact of outliers.

4. Validation: The implementation includes rigorous validation checks:

   • Point tests: Verifying the ratio of matrix elements to subtraction terms approaches unity near singular limits.
   • Technical cut checks: Ensuring independence of results from the technical infrared cutoff.
   • Pole tests: Explicit cancellation of $\epsilon$-poles in RV and VV contributions.

Results and Validation
The paper details the implementation of various processes, referencing specific publications for validation details. As a concrete example, the authors present a calculation for $Z + 1j$ production at the LHC ($\sqrt{s} = 7$ TeV):

• Setup: Anti-$k_T$ jet algorithm ($R=0.4$), NNPDF3.1 NNLO PDFs, and specific fiducial cuts on leptons and jets.
• Outcome: The total NNLO fiducial cross section is calculated to be 68.5(2) pb.
• Differential Distributions: The paper displays the transverse momentum distribution of the leading jet, showing the convergence from LO to NLO to NNLO, with uncertainty bands representing a 7-point scale variation.

Significance and Claims
The authors claim that NNLOJET provides a universally applicable, open-source framework for NNLO QCD calculations of jet observables, addressing the gap left by previous codes that were often proprietary or limited to specific process classes.

• Accessibility: By distributing the code under GPLv3.0 and providing a user-friendly runcard syntax and workflow script, the authors aim to make high-precision NNLO predictions accessible to a broader community of phenomenologists and experimentalists.
• Flexibility: The ability to define arbitrary fiducial cuts and observables without recompiling the code allows for direct comparison with experimental data.
• Extensibility: The modular design allows for the addition of new processes and features in future releases.

The paper concludes that NNLOJET is a mature tool for precision physics studies, stress-testing the Standard Model, and refining the determination of SM parameters and parton distribution functions. The code is available on HEPForge, and the authors invite user feedback and bug reports.