Threshold Resummation in QCD

• Threshold resummation is a QCD technique that sums logarithmically enhanced soft and collinear gluon emissions near the partonic threshold.
• It utilizes Mellin-space exponentiation and renormalization-group evolution to systematically organize dominant radiative corrections at all orders.
• This method improves predictions for collider observables such as invariant mass, transverse momentum, and rapidity by reducing theoretical uncertainties.

Threshold resummation is a key technique in the perturbative analysis of QCD hard-scattering processes, systematically summing logarithmically enhanced contributions from soft and collinear gluon emissions in the partonic threshold region. It achieves all-orders control over the dominant radiative corrections near kinematic boundaries, enabling improved reliability and precision in phenomenological predictions for observables such as invariant-mass, transverse-momentum, and rapidity distributions, single-inclusive spectra, and parton distribution extractions.

1. Fundamental Concepts and Physical Motivation

Threshold resummation targets the phase-space boundaries of QCD processes, characterized by the suppression of available energy for final-state radiation and a corresponding dominance of soft and collinear gluon emission. For a generic process with scaling variable $z$ (e.g., $z=Q^2/\hat s$) approaching unity, the fixed-order expansion in $\alpha_s$ generates distributions with strong singularities as $z\to1$, specifically plus-distributions $[\ln^k(1-z)/(1-z)]_+$. In Mellin moment space, these map to powers of $L=\ln N$, the conjugate variable to $1-z$, with $N\to\infty$ enhancing the logarithmic terms.

The rationale for threshold resummation arises from the failure of fixed-order perturbation theory near partonic threshold, where towers of terms $\alpha_s^n\ln^m N$ (with $m\leq2n$) appear at each order. Without resummation, these logarithms jeopardize the reliability of the expansion, especially for observables sensitive to extreme kinematic configurations (e.g., high-mass, high-transverse momentum, or large rapidity).

Resummation techniques have broad relevance across hadronic collisions (DY, Higgs, heavy quark, gaugino, squark, dark matter production), lepton-induced processes (DIS, SIA, SIDIS, DDVCS), global PDF fits, and lattice-QCD-based quasi-PDF extractions. For both color singlet and colored final states, and for inclusive, semi-inclusive, or differential observables, threshold resummation enhances accuracy and theoretical control.

2. Resummed Factorization and Sudakov Exponentiation

The core formalism of threshold resummation follows a factorization theorem, in which the partonic cross section in Mellin space exponentiates the large threshold logarithms. The resummed moment-space cross section typically takes the schematic form

$z=Q^2/\hat s$0

with $z=Q^2/\hat s$1. The structure of the exponent $z=Q^2/\hat s$2 is universal, reflecting the soft and collinear dynamics governed by gauge invariance and color flow. In explicit processes, the exponent is organized as

$z=Q^2/\hat s$3

where $z=Q^2/\hat s$4, and the $z=Q^2/\hat s$5 functions resum the LL, NLL, NNLL (and beyond) logarithms. The key anomalous dimensions entering the resummed exponent are:

• The universal cusp anomalous dimension $z=Q^2/\hat s$6 (controls double logarithms from soft-collinear emission).
• Process-dependent non-cusp anomalous dimensions, e.g., $z=Q^2/\hat s$7 (final-state collinear), $z=Q^2/\hat s$8 (soft wide-angle), and those associated to jet or beam functions.

The prefactor $z=Q^2/\hat s$9 encapsulates non-singular hard virtual corrections.

For transverse-momentum ($\alpha_s$0) spectra of colorless final states (e.g., $\alpha_s$1), the decomposition includes both incoming and outgoing legs’ contributions, leading to a Sudakov exponent with DY-like and DIS-like sectors—each associated with its own scale and factorized functions (Forte et al., 2021).

Explicitly, for Higgs or DY-type distributions, the NNLL hard-scattering coefficient in Mellin space is

$\alpha_s$2

with closed-form $\alpha_s$3 as detailed in App. A of (Forte et al., 2021). Extensions to higher logarithmic orders are immediate upon supplying the three-loop cusp and process-dependent soft anomalous dimensions.

3. Renormalization-Group (RG) Structure and Effective Field Theory

Renormalization-group analysis underpins the all-orders resummation, connecting the various scales in the problem (hard, soft, collinear) via evolution equations for the hard, jet, and soft functions. The RG flow is controlled by the cusp anomalous dimension and the process-dependent non-cusp terms. Effective field theory (SCET) provides an operator-based realization of this structure, enabling systematic derivations of the exponentiation properties and organizing power corrections.

The generic RG-improved coefficient function for a colorless-final-state transverse-momentum spectrum, for example, is given by

$\alpha_s$4

as presented in (Forte et al., 2021), with $\alpha_s$5 the cusp anomalous dimension, $\alpha_s$6 final-state collinear, $\alpha_s$7 soft wide-angle, and the prefactor $\alpha_s$8 containing hard-virtual matching.

Upon RG solution, this structure yields the classic Sudakov double-logarithmic suppression and systematically sums all logarithmic enhancements, ensuring RG invariance at each logarithmic order.

4. Matching to Fixed-Order and Prescription Dependence

To avoid double counting of terms present in both the fixed-order expansion and the resummed result, a consistent matching procedure is used. The matched cross section is constructed as

$\alpha_s$9

where the $z\to1$0 denotes the expansion of the resummed expression up to the order considered. This ensures that only genuinely higher-order logarithms are provided by the resummation, while fixed-order accuracy is maintained for constants and subleading terms (Forte et al., 2021). Analogous matched formulas are standard for invariant-mass and rapidity observables (Bonvini, 2010, Banerjee et al., 2017).

The actual inverse Mellin transform from moment space to physical observables requires a prescription to avoid the Landau pole in $z\to1$1. Common choices include the "Minimal Prescription" of Catani-Trentadue, which defines the Mellin inversion contour to avoid nonperturbative ambiguities, or the Borel prescription (Bonvini, 2010, Bonvini et al., 2012, Sterman et al., 2013). While all orders agree in the strict threshold limit, differences between these prescriptions manifest at subleading power and affect the treatment of PDFs and residual ambiguities at moderate $z\to1$2.

5. Applications: Differential Distributions and Phenomenological Impact

Threshold resummation is systematically developed for a wide variety of observables:

• Transverse-momentum spectra: For a color singlet produced inclusively (e.g., Higgs, DY), the threshold-resummed $z\to1$3 spectrum exhibits improved agreement with NLO at large $z\to1$4, and the formalism is valid through NNLL and beyond (Forte et al., 2021).
• Invariant mass and rapidity: Double Mellin-space resummation yields fully differential NNLL+NNLO accuracy for Higgs and DY rapidity spectra, leading to a notable enhancement of the central value and a reduction of scale uncertainty bands by roughly a factor of two (Banerjee et al., 2017, Banerjee et al., 2018, Ros et al., 7 Jan 2026).
• Heavy colored particle production: Pair production of squarks, gluinos, and heavy quarks requires simultaneous resummation of soft gluons and Coulombic effects; the resulting enhancement of cross sections and reduction in theoretical uncertainty is crucial for high-mass searches (Beneke et al., 2010).
• PDF determination: Inclusion of threshold resummation in the analysis of DIS and DY data leads to significant corrections (up to 10–20% at $z\to1$5), a shift in extracted PDFs at large