Abstract: We report on a method for matching the next-to-leading order calculation of QCD jet production in ${e}^{+}{e}^{\ensuremath{-}}$ annihilation with a Monte Carlo parton shower event generator (MC) to produce realistic final states. The final result is accurate to next-to-leading order (NLO) for infrared-safe one-scale quantities, such as the Durham 3-jet fraction ${y}_{3}$, and agrees well with parton shower results for multiscale quantities, such as the jet mass distribution in 3-jet events. For our numerical results, the NLO calculation is matched to the event generator Pythia, though the method is more general. We compare one-scale and multiscale quantities from pure NLO, pure MC, and matched NLO-MC calculations.
Introduction
In Quantum Chromodynamics (QCD), there is a historical tension between two primary methods of calculating physics: Fixed-Order Perturbation Theory (NLO) and Parton Showers (Monte Carlo). NLO provides high precision for "inclusive" quantities, such as the total rate of producing three jets, but it often fails at small scales. For example, when calculating the mass of a single jet, NLO math can produce unphysical results like negative probabilities or infinite singularities.
Conversely, Parton Shower programs like Pythia simulate the "showering" of partons into many particles and their eventual "hadronization" into observable particles. While these look like real data, they usually only have Leading Order (LO) accuracy, meaning their total rates can be off by 10-50%. This paper addresses the challenge of combining these methods to obtain NLO precision for the "hard" interaction without double-counting the physics handled by the "soft" parton shower.
Methodology
The authors implement a matching algorithm specifically for $e^+ e^- \to 3$ jets. The core of their approach involves several key components:
- Sudakov Suppression: They use Sudakov form factors to damp the singularities in the NLO calculation, treating the first "splitting" of a parton as part of the NLO calculation.
- Subtraction Terms: To avoid double-counting, they subtract parts of the NLO calculation that are effectively approximated by the parton shower.
- Modified Pythia Interface: The authors modified Pythia to accept a "partially developed shower" (a state with up to 7 partons), allowing the generator to finish the showering and perform hadronization using the Lund string model.
- Soft Gluon Antenna: They include a specific treatment for wide-angle soft gluons to ensure the color structure of the 3-jet event is preserved.
Key Insights: The "Negative Bin" Problem
A primary motivation for this work is the failure of pure NLO for detailed distributions. As illustrated in Figure 1, the NLO calculation for jet mass shows a massive negative value in the first bin. This is a mathematical artifact of infrared cancellations that makes the theory unusable for low-mass jet analysis without matching.
Results and Comparison
The matched NLO-MC method fixes these unphysical artifacts, producing smooth, positive, and realistic distributions. The authors demonstrate that the matched calculation preserves the NLO normalization for the 3-jet fraction ($f_3$) within about 10%, proving that the shower doesn't "wash out" the precision of the underlying theory.
Furthermore, the results highlight the importance of hadronization. The authors show that turning partons into hadrons shifts the jet mass peak by about 2 GeV. This shift underscores why NLO theory alone is insufficient for comparing to actual detector data—the non-perturbative physics of hadronization is numerically significant.
Conclusion and Future Work
The matched NLO+PS+Had calculation successfully matches NLO predictions across various jet resolution scales while eliminating unphysical singularities. The method shows high theoretical stability; varying the renormalization scale ((\mu)) by a factor of 2 resulted in only a ~5% change in the matched result.
While the current implementation is specific to 3-jet production and tied to Pythia, the authors suggest that future NLO-MC hybrids will need to be more sophisticated and less "hand-tuned" to specific processes to be fully utilized in the LHC era.
Summary updated