Momentum-Space Resummation in QCD

• Momentum-space resummation is a QCD technique that directly sums logarithmic corrections in momentum space, avoiding complications from inverse transforms.
• It employs RG evolution, SCET-based factorization, and analytic methods to effectively predict transverse momentum and threshold observables.
• The approach enables seamless matching with fixed-order results while minimizing sensitivity to nonperturbative artifacts and spurious singularities.

Momentum-space resummation is a set of techniques in perturbative quantum field theory, notably in QCD, used to systematically sum large logarithmic corrections that appear in physical observables expressed directly in momentum space, rather than in their conjugate (e.g., impact-parameter, Mellin, or Laplace) spaces. This approach has become increasingly important in the calculation of observables with multiple kinematic hierarchies, particularly in soft and collinear regimes, such as transverse momentum distributions, threshold regions, event shapes, and their combinations. It yields physical predictions that can be used directly in Monte Carlo implementations, often avoids the need for ad hoc regularizations (e.g., Landau pole avoidance in impact-parameter space), and provides improved theoretical control over uncertainties and nonperturbative effects.

1. Overview and Motivation

Momentum-space resummation addresses the need to resum large logarithms that arise in the perturbative series for both inclusive and differential observables, especially in kinematic regimes characterized by widely separated scales—such as $q_T \ll Q$ in Drell-Yan, Higgs, or dijet production. Traditionally, resummation was performed in conjugate spaces like Mellin $N$-space (for threshold logs) or Fourier (impact-parameter) $b$-space (for transverse-momentum logs). However, such formulations require inverse transformations back to momentum space, which can introduce additional theoretical complications—notably the Landau pole at large $b$ and artificial scale prescriptions. Momentum-space resummation circumvents these issues by constructing RG-improved predictions directly in the physical observable space, providing both conceptual and computational advantages (Kang et al., 2017, Ebert et al., 2016, Monni et al., 2016, Bizon et al., 2017, Simonelli, 8 Sep 2025).

2. Core Formalisms and Factorization Structures

Momentum-space resummation builds on all-orders factorization theorems, often formulated within Soft-Collinear Effective Theory (SCET), that decompose cross sections into hard, soft, and (beam) collinear functions:

• For color-singlet observables such as Drell-Yan or Higgs $q_T$ spectra at $q_T\ll Q$, the double-differential cross section is factorized as

$$\frac{d\sigma}{dQ^2\,dy\,dq_T^2} = \sigma_0\,H(Q^2;\mu) \int d^2k_a\,d^2k_b\,d^2k_s\,\delta^{(2)}(q_T - k_a - k_b - k_s) B_a(\omega_a, k_a; \mu, \nu) B_b(\omega_b, k_b; \mu, \nu) S(k_s; \mu, \nu)$$

where $H$ is the hard function encoding virtual corrections; $B_{a,b}$ are TMD beam functions; and $S$ is the soft function. Renormalization group equations (RGEs) in both $N$0 (virtuality) and $N$1 (rapidity) are used to resum Sudakov double logarithms, with anomalous dimensions organized into cusp and non-cusp terms (Ebert et al., 2016).

• For color-charged final states (e.g., heavy quark pair production, hadronic jets), matrix-valued soft functions and color-evolution kernels enter, with additional complexity from wide-angle and final-state correlations (Catani et al., 2014, Sun et al., 2015).
• Joint resummation for observables sensitive to both threshold and transverse-momentum singularities employs multi-scale factorization theorems and profile scale interpolations, with each resummation kernel evolved from canonical scales to common values to avoid double counting and recover the fixed-order limit at large $N$2 or away from threshold (Lustermans et al., 2016).

3. Methods for Direct Momentum-Space Resummation

Several strategies exist for performing resummation directly in momentum space:

• Distributional RG Evolution and Scale Setting: Plus-distributional ($N$3) RGEs are solved directly in momentum space using "distributional scale setting." By assigning canonical scales as distributional arguments, all logarithmic plus distributions in the boundary terms are exactly canceled, leaving only regular distributions in the final expression. This treatment generalizes to any observable whose spectrum is a plus distribution, such as thrust or $N$4-jettiness (Ebert et al., 2016).
• Hybrid and Analytic Techniques for $N$5-Spectra: Kang, Lee, and Vaidya (Kang et al., 2017) introduced a hybrid scheme where the virtuality scale $N$6 is set in momentum space while the rapidity scale $N$7 remains $N$8-dependent, enabling the entire resummed cross section to be evaluated semi-analytically in $N$9-space. By transforming the Bessel integrals analytically via Mellin–Barnes representations and Hermite polynomial expansions, all dependence on kinematics and scales becomes analytic; only a small set of numerically-precomputed coefficients is required. This approach completely avoids the Landau pole and achieves significant speedup over $b$0-space resummation.
• Cumulative and Differential Methods: Direct analytic inversion of factorized cumulants using saddle-point approximations (as in (Aglietti et al., 23 Jun 2025)) can yield precise results for event-shape and $b$1 observables. These methods systematically improve upon classical (CTTW-type) Taylor expansions by centering the expansion at the actual saddle and fully resuming all logarithmic towers.
• Infrared Regularization and Analytic Continuation: Analytic prescriptions for the low-$b$2 region (e.g., (Simonelli, 8 Sep 2025)) replace arbitrary $b$3 or minimal prescriptions with physically-motivated models for the freezing of $b$4 and the infrared evolution of PDFs, thus achieving a well-defined continuation into the nonperturbative regime without affecting the perturbative prediction for $b$5.

4. Connections to Threshold, Joint, and High-Energy Resummation

Momentum-space techniques extend naturally to other classes of observables:

• Threshold Resummation: Near kinematic endpoints ($b$6 or $b$7), soft-gluon emission leads to large logarithms of $b$8 or $b$9. Momentum-space RG evolution of hard and soft functions, with analytic Laplace-inverse transforms, is systematically applied to resummation for colored and colorless final states, including single-inclusive direct top, $b$0, and stop-pair production (Li et al., 2019, Yang et al., 2014, Broggio et al., 2013).
• Combined Resummation: Observables sensitive to both $b$1 and $b$2 singularities require multi-scale matching and RG evolution. The regime-wise matching framework in (Lustermans et al., 2016) partitions the phase space into regions dominated by TMD, collinear-soft, or threshold-soft dynamics; each sector has its own factorized and resummed description, and smooth profile functions ensure seamless matching.
• High-Energy (Low-$b$3) Resummation: In the small-$b$4 regime, momentum-space factorization with off-shell incoming gluons and BFKL anomalous dimension $b$5 produces closed-form all-order resummed $b$6 spectra, as illustrated for Higgs production in gluon fusion (Forte et al., 2015). The $b$7 impact factor and Sudakov structures smoothly interpolate to the standard double-logarithmic (Sudakov) form in the $b$8 limit.

5. Practical Implementation, Matching, and Numerical Impact

Momentum-space resummation schemes have been implemented in several phenomenologically important contexts, often matched directly to next-to-leading order (NLO) or next-to-next-to-leading order (NNLO) fixed-order results:

• Matching Procedures: To avoid double counting, the fixed-order (e.g., NLO) expansion of the resummed formula is subtracted from the fixed-order result and added back to the full resummed prediction. This ensures that the resummation is active only where needed (small $b$9 or near threshold) and that the exact fixed order is preserved elsewhere (Kang et al., 2017, Monni et al., 2016, Bizon et al., 2017).
• Numerical Stability and Comparison with Data: The momentum-space approach produces smooth, stable $q_T$0 (and threshold) spectra with substantially reduced theoretical uncertainties—typically $q_T$1 at NNLL for $q_T$2 production (Li et al., 2019), or percent-level control for Higgs $q_T$3 at N$q_T$4LL+NNLO (Bizon et al., 2017). Compared with standard impact-parameter-space or Mellin-space resummation, the approach gives compatible results at moderate $q_T$5, but with improved computational efficiency and no sensitivity to spurious nonperturbative model artifacts.
• Infrared Treatment and Phenomenology: The explicit separation of perturbative and nonperturbative physics above and below a scale $q_T$6 allows genuine model-independent testing of TMD and PDF evolution only where needed. Data comparisons confirm that the perturbative resummation alone accounts for the bulk of observed spectral shapes in high-energy Drell-Yan and $q_T$7 production (Simonelli, 8 Sep 2025).

6. Extensions, Generalizations, and Applications

The momentum-space resummation program has seen wide-ranging generalization:

• Global Event Shapes & Recursive IRC-Safe Observables: The resummation has been applied to fully global observables such as thrust, $q_T$8-jettiness, and $q_T$9-like event shapes, with scalability up to N$q_T\ll Q$0LL accuracy in both cumulative and differential forms (Bizon et al., 2017).
• Multi-differential and Color Non-Singlet Observables: Extensions to multi-differential observables, color-correlated matrix structures, and azimuthal modulations have been realized, allowing predictions for $q_T\ll Q$1 production, dijet correlations, and gluon-fusion exclusive processes with explicit spin-interference (through tensor decompositions in the hard-collinear matching) (Catani et al., 2010, Catani et al., 2014, Sun et al., 2015).
• Monte Carlo and Event Generators: Owing to the formulation directly in momentum space, these resummations facilitate fully exclusive event generation, arbitrary observable cuts, and efficient interfacing with parton showers, as demonstrated by practical implementations such as the RadISH generator (Bizon et al., 2017).

7. Key Results, Theoretical Significance, and Limitations

Momentum-space resummation is now established as a theoretically robust and computationally efficient tool, systematically improvable to high logarithmic orders for a wide class of QCD observables. It provides:

• Systematic resummation for towers of logarithms $q_T\ll Q$2 (with $q_T\ll Q$3 or similar), up to N$q_T\ll Q$4LL as needed
• Direct analytic control over all physical variables, scale variations, and error estimates
• Transparent matching to fixed-order predictions
• Nonperturbative modeling only where strictly required
• Absence of unphysical singularities (e.g., Landau pole cutoffs) and no reliance on auxiliary $q_T\ll Q$5-space regulators

Known limitations include the current reliance on the quality of SCET factorization theorems, the knowledge of anomalous dimensions and matching corrections to the desired order, and the careful treatment required for multi-scale observables that border different kinematic regimes (requiring explicit regime matching and profile scales). For certain observables, analytic continuation into deep-infrared remains an area of ongoing development.

For reference to explicit master formulas, anomalous dimensions, and matching coefficients, see (Kang et al., 2017, Ebert et al., 2016, Bizon et al., 2017, Monni et al., 2016, Simonelli, 8 Sep 2025).