Structure of non-global logarithms with Cambridge/Aachen clustering

This paper determines the four-loop structure of Abelian and non-Abelian non-global logarithms for $e^+e^-$ dijet invariant mass using the Cambridge/Aachen algorithm, demonstrating that it minimizes these logarithmic effects compared to anti-$k_t$ and $k_t$ clustering schemes.

The Gist

Imagine you are at a massive, chaotic concert where thousands of people (particles) are dancing. Your goal is to group these dancers into specific "jets" (clusters) based on how close they are to each other. The way you decide who belongs in which group changes the final picture of the dance floor.

In the world of particle physics, scientists use different "grouping rules" (algorithms) to sort these particles after a collision. This paper investigates three specific rules: anti-$k_t$, $k_t$, and Cambridge/Aachen (C/A).

Here is the breakdown of what the authors discovered, explained simply:

1. The Problem: The "Bad Neighbors" Effect

When particles fly out from a collision, they don't just stay in their neat little groups. Sometimes, a particle from one group gets pulled into another group's territory, or a particle from outside sneaks in.

In physics, this causes a mathematical headache called Non-Global Logarithms (NGLs). Think of these as "noise" or "static" that messes up your measurements.

• The Anti-$k_t$ rule: It's like a strict bouncer who forms perfect, rigid circles. It's easy to calculate, but it lets a lot of this "noise" in.
• The $k_t$ rule: It's a bit more flexible, grouping people based on how energetic they are. It reduces the noise a little, but not enough.
• The Cambridge/Aachen (C/A) rule: This is the most complex one. It groups people only based on how close they are in space, ignoring their energy. It's like sorting a crowd purely by who is standing next to whom, regardless of how loud they are shouting.

2. The Challenge: A Maze Without a Map

The authors wanted to see how much "noise" (NGLs) the C/A rule creates compared to the others.

• The Difficulty: The anti-$k_t$ and $k_t$ rules have a natural order (like a ladder where you climb step-by-step). The C/A rule has no ladder; it's a maze where any two people could be the closest pair at any moment.
• The Analogy: Calculating the noise for anti-$k_t$ is like walking down a straight hallway. Calculating it for C/A is like trying to solve a Rubik's Cube while blindfolded, where every twist changes the rules.

3. The Breakthrough: Cracking the Code

The team spent years (and used powerful computers) to calculate this noise up to four loops. In physics terms, a "loop" is a level of complexity.

• Loop 1 & 2: C/A and $k_t$ were identical.
• Loop 3 & 4: This is where the magic happened. The authors found that the C/A rule introduces a special "correction" that acts like a noise-canceling headphone.

4. The Big Discovery: C/A is the Quietest

When they compared the results:

• Anti-$k_t$: The loudest. Lots of noise.
• $k_t$: Quieter, but still has significant static.
• Cambridge/Aachen (C/A): The quietest.

The authors found that the C/A algorithm naturally reduces the "noise" by more than 50% compared to the anti-$k_t$ rule. It's as if the C/A rule is a master organizer that instinctively knows how to arrange the dancers so that the "bad neighbors" (the ones causing the mathematical noise) don't interfere with each other.

5. Why This Matters

In particle physics, we want to measure things with extreme precision (like the mass of a jet). If the "noise" is too high, our measurements are blurry.

• Because the C/A algorithm minimizes this noise so effectively, the authors conclude that C/A is the best choice for studying these complex particle events, even though it is much harder to calculate.

Summary Analogy

Imagine you are trying to take a clear photo of a fireworks display.

• Anti-$k_t$ is like using a camera with a wide, shaky lens. You get the picture, but it's blurry.
• $k_t$ is a slightly steadier lens. Better, but still a bit fuzzy.
• Cambridge/Aachen is like using a high-tech, image-stabilizing lens. It's much harder to build and program (the math is incredibly complex), but the resulting photo is crystal clear.

The Bottom Line: This paper proves that while the Cambridge/Aachen algorithm is the most difficult to understand and calculate, it is the superior tool for cleaning up the "static" in our view of the subatomic world, giving us the clearest possible picture of nature's building blocks.

Technical Summary

1. Problem Statement

Precision studies of Quantum Chromodynamics (QCD) at colliders increasingly rely on observables sensitive to restricted phase-space regions, such as jet substructure (e.g., jet mass). These are known as non-global observables. Their perturbative distributions are enhanced by:

• Global logarithms: From soft and collinear radiation associated with the hard initiating parton.
• Non-Global Logarithms (NGLs): From correlated emissions in disparate angular regions.
• Clustering Logarithms (CLs): Generated by the specific sequence of the jet algorithm recombining particles.

While NGLs and CLs have been studied extensively for the anti-$k_t$ and $k_t$ jet algorithms, the Cambridge/Aachen (C/A) algorithm presents unique challenges. Unlike $k_t$ (which orders by transverse momentum) or anti-$k_t$ (which has a rigid cone-like structure), C/A clusters purely based on angular separation. This lack of an intrinsic hardness ordering parameter prevents the use of standard evolution equations (like the BMS equation) and complicates analytical resummation. Consequently, progress for C/A has been limited to "brute-force" fixed-order calculations, and the behavior of NGLs and CLs at higher loops (beyond two) for C/A was previously unknown.

2. Methodology

The authors perform fixed-order calculations up to four loops for the normalized dijet invariant mass in $e^+e^-$ annihilation processes.

• Approximations:
  • Soft (Eikonal) Approximation: Retains only leading soft singularities.
  • Strong Energy Ordering: Assumes $Q \gg \omega_1 \gg \omega_2 \gg \dots \gg \omega_n$. This simplifies the combinatorial complexity of the clustering sequence.
  • Single-Logarithmic Accuracy: The calculations target terms of the form $\alpha_s^n L^n$.
• Algorithm Implementation:
  • The C/A algorithm is defined by the distance measure $d_{ij} \propto 1 - \cos\theta_{ij}$.
  • Unlike $k_t$ and anti-$k_t$, where the energy ordering induces a unique clustering hierarchy, C/A requires considering all possible orderings of the distance measures $d_{ij}$ at a given loop order.
  • The authors developed a Python code to automate the generation of all possible gluon configurations (real vs. virtual) and distance orderings. For four loops, this involved analyzing $2^6 = 64$ configurations and 1,957 distinct distance orderings.
• Calculation Strategy:
  • The calculation identifies "mis-cancellations" between real emissions and virtual corrections. These mismatches occur when the clustering algorithm drags a soft gluon in or out of the measured jet, preventing the standard KLN theorem cancellation.
  • The C/A result is decomposed into the known $k_t$ result plus a correction term ($\tilde{\Sigma}^{C/A}$), representing the unique orderings present in C/A but absent in $k_t$.
  • Numerical integration is performed using the Cuba library (Vegas routine) to handle the angular integrations while retaining full color and jet-radius ($R$) dependence.

3. Key Contributions

• First Four-Loop Calculation: This is the first computation of NGLs and CLs for the C/A algorithm up to four loops in the literature.
• Analytical Framework for C/A: The paper establishes a methodology to handle the lack of ordering in C/A by systematically summing over all distance hierarchies, effectively treating C/A as the "superset" of $k_t$ and anti-$k_t$ orderings.
• Exponentiation Ansatz: Based on the fixed-order results, the authors propose a next-to-leading-logarithmic (NLL) resummed form factor where the global Sudakov factor multiplies exponentials encoding the fixed-order CLs and NGLs coefficients.

4. Key Results

A. Three-Loop Results

• CLs and NGLs Coefficients: The authors computed the coefficients $F_3(R)$ (CLs) and $G_{3,a/b}(R)$ (NGLs).
• Reduction of Effects: The unique C/A corrections are generally of the opposite sign to the $k_t$ corrections. This leads to a significant reduction in the magnitude of both CLs and NGLs compared to the $k_t$ algorithm.
  • At $R=0$, CLs are reduced by ~45%.
  • At $R=1$, CLs are reduced by ~95% (becoming negligible).
  • NGLs are reduced by 8–33% depending on the color structure and $R$.
• Edge Effect: The CLs coefficient remains non-vanishing as $R \to 0$, confirming the "edge effect" previously observed in other algorithms.

B. Four-Loop Results

• Complexity: The calculation required handling 1,957 orderings. The results confirm the trend observed at three loops.
• Magnitude Reduction: Across most of the jet radius range ($R \in [0, 1]$), the C/A algorithm reduces the combined CLs and NGLs contribution by 74% to 92% compared to anti-$k_t$, and significantly outperforms $k_t$ (which reduces by 63–78%).
• Finite-$N_c$ Effects: The finite-$N_c$ corrections (terms involving $C_A - 2C_F$) were found to be insignificant up to four loops.

C. All-Order Resummation Comparison

• The authors constructed a resummed jet mass distribution using their fixed-order coefficients in an exponential form factor.
• Comparison: When compared to Monte Carlo (MC) results for anti-$k_t$ and $k_t$ (using the Dasgupta-Salam code), the C/A distribution shows the smallest deviation from the pure Sudakov form factor.
• Peak Shift: The inclusion of NGLs/CLs in anti-$k_t$ shifts the Sudakov peak position and reduces its height by up to 23%. For C/A, this reduction is only ~6–8%, indicating that C/A effectively minimizes the impact of non-global logarithms.

5. Significance and Conclusion

• Optimal Algorithm for Non-Global Observables: The study concludes that the Cambridge/Aachen algorithm is the preferred choice among the three major algorithms (anti-$k_t$, $k_t$, C/A) for analyzing non-global observables like jet mass. Its angular-only clustering naturally suppresses the magnitude of both clustering and non-global logarithms.
• Theoretical Insight: The work demonstrates that C/A encompasses the full set of clustering orderings, acting as a correction to $k_t$. The fact that these corrections are negative (reducing the total logarithmic enhancement) suggests a structural advantage of angular clustering in mitigating non-global effects.
• Future Directions: The paper highlights the need for an integro-differential equation (similar to the BMS equation) or a dedicated Monte Carlo code specifically for C/A to achieve all-orders resummation without relying on fixed-order approximations. The current results provide the necessary coefficients to motivate such developments.

In summary, this paper provides a rigorous, high-order fixed-order analysis of the C/A jet algorithm, proving its superiority in minimizing non-global logarithmic effects, which is crucial for precision QCD phenomenology at future colliders.