$V/H$+Jet Production with the Cambridge/Aachen Algorithm

Technical Summary: V/H+Jet Production with the Cambridge/Aachen Algorithm

Problem Statement
Precision calculations of jet observables at hadron colliders, particularly the Large Hadron Collider (LHC), require a refined understanding of Quantum Chromodynamics (QCD) radiation patterns. A critical challenge in this domain is the accurate treatment of large logarithms that arise in the limit of small normalized jet mass ($m_j^2/p_t^2 \ll 1$). These logarithms are categorized into three types:

1. Global logarithms: Arising from soft-collinear and hard-collinear radiation off all hard partons, which exponentiate into the Sudakov form factor.
2. Non-Global Logarithms (NGLs): Arising from soft-gluon correlations connecting spatially separated phase-space regions (e.g., emissions inside the jet induced by radiation outside). These resist straightforward exponentiation and are a manifestation of non-Abelian color dynamics.
3. Clustering Logarithms (CLs): Unique to sequential recombination jet algorithms (other than anti-$k_t$), these arise from the algorithm-induced recombination of independent primary gluon emissions, distorting real-virtual cancellations.

While fixed-order calculations for NGLs and CLs have been established for the anti-$k_t$ and $k_t$ algorithms up to four loops, the behavior of the Cambridge/Aachen (C/A) algorithm—a geometric, angular-ordered clustering scheme widely used in jet substructure and grooming—remained unexplored at higher orders in hadronic collisions. The specific problem addressed is the calculation of the invariant-mass distribution of the leading jet in Vector Boson (V) or Higgs (H) + jet production using the C/A algorithm, incorporating both NGLs and CLs through four-loop order.

Methodology
The authors perform fixed-order perturbative calculations up to four-loop order for the normalized invariant mass squared ($\rho = m_j^2/p_t^2$) of the leading jet. The study focuses on three partonic Born channels relevant to V/H+jet production:

• $\delta_1: q\bar{q} \to g V/H$
• $\delta_2: qg \to q V/H$
• $\delta_3: gg \to g H$ (Higgs production via effective vertex)

The calculations are conducted within the eikonal approximation with strong energy ordering among final-state soft gluons, ensuring single-logarithmic accuracy. The methodology involves:

• Algorithm Implementation: The C/A algorithm is defined by a purely geometric distance measure ($d_{ij} \propto \Delta R_{ij}^2$), independent of transverse momentum. This contrasts with the $k_t$ algorithm, where distance is weighted by $p_{ti}$. Consequently, while C/A and $k_t$ yield identical results at two loops, their clustering sequences diverge at three loops and beyond due to different ordering of multi-particle configurations.
• Analytical Framework: The authors utilize a measurement operator formalism to compute the contributions of soft gluons. The calculation involves enumerating all possible in-jet and out-of-jet gluon configurations and all pairwise clustering orderings.
• Automation: Due to the combinatorial complexity of the C/A algorithm at four loops (involving 64 initial configurations and 1957 distance orderings), the authors developed a Python code for bookkeeping and a Mathematica script to organize the output expressions.
• Integration: The resulting integrands, which involve eikonal amplitudes squared and step functions enforcing C/A clustering constraints, are evaluated numerically using the Cuba library.
• All-Orders Estimation: Since dedicated all-orders numerical codes for C/A clustering with collinear singularities are unavailable, the authors construct resummed form factors by exponentiating the fixed-order coefficients up to four loops. These are compared against the Dasgupta–Salam Monte Carlo code (which implements $k_t$ and anti-$k_t$) to benchmark the behavior.

Key Contributions

1. First Four-Loop C/A Results: This paper provides the first systematic fixed-order expansion of both CLs and NGLs for the C/A algorithm in a hadronic environment through four-loop order.
2. Separation of Contributions: At two and three loops, the authors successfully separate the CL and NGL coefficients. At four loops, the interplay between color factors and C/A constraints prevents the independent numerical determination of CLs and NGLs; however, their combined effect is calculated.
3. Geometric vs. Energy Ordering: The work explicitly derives and analyzes the distinct higher-loop clustering structures introduced by the purely geometric nature of the C/A distance measure, which first manifests at three loops.
4. Finite-$N_c$ Corrections: The calculations retain full dependence on color factors ($N_c$), allowing for an assessment of finite-$N_c$ corrections beyond the large-$N_c$ approximation.

Results

• Two-Loop Order: The C/A algorithm yields results identical to the $k_t$ algorithm, as expected, since only a single pair of soft gluons is considered.
• Three-Loop Order:
  • CLs: The pure C/A corrections to the CL coefficients are generally of opposite sign to the $k_t$ contributions and smaller in magnitude, effectively reducing the total CL impact. Notably, for channels $\delta_2$ and $\delta_3$, the total C/A CL coefficient vanishes at specific jet radii ($R \sim 0.5$ and $R \sim 0.78$, respectively), a feature not present in the $k_t$ case.
  • NGLs: The C/A algorithm suppresses NGLs more effectively than $k_t$ at small $R$ (reductions of $\sim 75-79\%$). Similar to CLs, the NGL coefficients for channels $\delta_2$ and $\delta_3$ vanish at $R \sim 0.55$.
  • Combined Effect: The sum of CL and NGL coefficients is significantly smaller for C/A than for $k_t$ and anti-$k_t$ over a wide range of $R$.
• Four-Loop Order:
  • The combined non-global effect (CL + NGL) for C/A remains smaller than for anti-$k_t$ across all channels.
  • Compared to $k_t$, C/A continues to show suppression for channels $\delta_1$ and $\delta_2$. However, for the gluon-fusion channel ($\delta_3$), the C/A effect exceeds that of $k_t$ for $R \lesssim 0.9$, indicating that $k_t$ may be preferable for this specific channel at lower loop orders.
  • The "optimal" radius where three-loop coefficients vanish does not persist at four loops, as the four-loop contributions are non-zero at those radii.
• All-Orders Behavior:
  • Resummed form factors constructed from the fixed-order coefficients suggest that at all orders, the C/A and $k_t$ algorithms perform comparably in mitigating non-global logarithms, particularly for $\rho \gtrsim 0.005$.
  • Both C/A and $k_t$ significantly outperform the anti-$k_t$ algorithm in reducing the overall impact of non-global logarithms on the jet-mass distribution.
  • Finite-$N_c$ corrections remain at the percent level ($<2\%$) for phenomenologically relevant values, consistent with previous findings in lepton and hadron processes.

Significance and Claims
The paper claims to complete the fixed-order study of the three major sequential recombination algorithms (anti-$k_t$, $k_t$, and C/A) on non-global observables up to four loops for processes involving two and three hard colored legs. The authors emphasize that while the C/A algorithm is theoretically attractive for its angular ordering and use in jet grooming, this work quantifies its specific perturbative structure regarding non-global logarithms.

The primary conclusion is that the C/A algorithm suppresses large non-global logarithms more effectively than the anti-$k_t$ algorithm and, at fixed order, often more effectively than the $k_t$ algorithm. However, the authors modestly note that at all orders, the performance of C/A and $k_t$ appears comparable. The work highlights that the "optimal" jet radius for eliminating non-global effects observed at three loops is an order-by-order cancellation that does not persist at four loops, cautioning against relying on fixed-order cancellations for physical predictions without all-orders resummation. The study underscores the complexity of extending such calculations to higher orders and the current limitations in available numerical tools for validating C/A clustering in hadronic environments with collinear singularities.