Equivalent Vector Boson Approximation (EVA)

• Equivalent Vector Boson Approximation is a method that models quasi-collinear emission of weak bosons from fermions to simplify scattering process calculations.
• It factorizes complex high-energy interactions by treating emitted W and Z bosons as effective beams, using both single and correlated boson luminosities.
• Modern implementations balance computational efficiency with accuracy, though limitations such as gauge dependence and collinear-region restrictions require careful validation.

Equivalent Vector Boson Approximation (EVA), also called EVBA and, in channel-specific usage, the Effective $W$ Approximation (EWA) or EZA for $Z$-initiated reactions, is the electroweak analogue of the equivalent-photon approximation and the parton picture: an incoming quark or lepton radiates a weak gauge boson quasi-collinearly, and the emitted $W^\pm$ or $Z$ is treated as an effective beam constituent entering a hard subprocess such as $VV\to X$. In modern usage the term covers a family of factorized approximations, from leading-log single-boson distributions to polarization-resolved and correlated two-boson luminosities, with an accuracy controlled by virtuality, polarization, gauge choice, and event selection (Borel et al., 2012, Bernreuther et al., 2015, Ruiz et al., 2021, Dahlén et al., 25 Jul 2025).

1. Terminology, physical picture, and scope

The literature uses closely related names for essentially the same approximation class. In like-sign $WW$ studies and precision VBS reviews, the approximation is described as an “old high-energy approximation” in which one extends the parton picture of hadrons to include weak vector bosons radiated from quarks, and the basic splitting process

$q \to qV$

is approximated by its asymptotic behaviour in the collinear limit, where it is logarithmically enhanced (Dittmaier, 20 Apr 2026, Dittmaier et al., 2024). In channel-specific nomenclature, $W$-initiated processes are often called EWA, $Z$-initiated ones EZA, while EVA/EVBA is the generic label (Dahlén et al., 25 Jul 2025).

At the level of collider kinematics, the approximation replaces a full process such as

$f_1 f_2 \to f_1' f_2' X$

by a sequence in which the incoming fermions emit virtual weak bosons and the hard reaction is the embedded subprocess

$Z$0

The operational approximation is then to evaluate the hard scattering as if the vector bosons were incoming on-shell states, while the emission factors are absorbed into effective weak-boson distributions or luminosities (Bernreuther et al., 2015). This picture has been used for vector boson scattering (VBS), vector boson fusion (VBF), heavy-particle production, and multi-TeV lepton-collider studies (Ballestrero et al., 2011, Ruiz et al., 2021, Dahlén et al., 25 Jul 2025).

A persistent misconception is that EVA is a single canonical formula. The literature instead distinguishes between ordinary EVBA based on single-boson distributions, improved EVBA based on correlated two-boson luminosities, threshold-corrected variants near $Z$1, and modern polarized implementations in Monte Carlo frameworks (Bernreuther et al., 2015, Alikhanov, 2018, Ruiz et al., 2021).

2. Factorization structure and canonical formulas

A field-theoretic formulation of the approximation analyzes when the scattering of equivalent massive spin-1 bosons factorizes from the complete process. For a generic weak process $Z$2, after integrating over the azimuth of the outgoing quark, the differential cross section takes the factorized form

$Z$3

where $Z$4 is the longitudinal momentum fraction carried by the emitted boson, $Z$5 is the transverse momentum of the recoiling forward fermion, and $Z$6 is the charge of the emitted weak boson (Borel et al., 2012).

The corresponding splitting functions are

$Z$7

$Z$8

$Z$9

For two emitting fermion lines, the approximation generalizes to a double convolution over the two boson momentum fractions and transverse momenta, with a sum over the polarization labels of the hard subprocess $W^\pm$0 (Borel et al., 2012).

In polarized PDF language for multi-TeV lepton collisions, the beam-level inclusive cross section is written as

$W^\pm$1

with $W^\pm$2 the boson light-cone energy fractions and $W^\pm$3 the hard boson-boson invariant mass (Ruiz et al., 2021). At leading order, the $W^\pm$4-evolved collinear weak-boson PDFs for a left-handed fermion are

$W^\pm$5

$W^\pm$6

$W^\pm$7

with the right-handed distributions related by $W^\pm$8 factors (Ruiz et al., 2021). A structurally important feature is that the transverse distributions carry the collinear logarithm, while the longitudinal distribution does not (Ruiz et al., 2021, Dahlén et al., 25 Jul 2025).

3. Polarization, off-shell continuation, and gauge dependence

For massive vector bosons, naive diagram-by-diagram factorization is gauge dependent. Off-shell subamplitudes and longitudinal polarization vectors can contain spurious $W^\pm$9 enhancements, so the validity of EVA is less trivial than for the equivalent-photon approximation (Borel et al., 2012). A derivation in a physical axial gauge,

$Z$0

shows that ghosts decouple, the propagator is physical and non-singular as $Z$1, and the singular $Z$2 structure associated with nearly on-shell boson exchange cleanly isolates the factorizing contribution (Borel et al., 2012).

The improved effective vector boson approximation (IEVBA) of Kuss and Spiesberger replaces simple products of single-boson distributions by correlated two-vector-boson luminosities $Z$3, keeping exact kinematics for boson emission and resolving the polarization structure of the boson pair (Bernreuther et al., 2015). Its characteristic cross-section formula is

$Z$4

The polarization basis contains nine combinations: $Z$5 with parity-odd luminosities becoming relevant for parity-violating subprocesses such as $Z$6 (Bernreuther et al., 2015).

The off-shell continuation of the hard subprocess is the central ambiguity. In unitary gauge, longitudinal polarization vectors contain kinematic singularities as $Z$7, and the off-shell cross sections are modeled by extrapolation factors

$Z$8

$Z$9

$VV\to X$0

In axial gauge, the longitudinal polarization vectors do not contain these unitary-gauge kinematic singularities, and the revisited IEVBA sets

$VV\to X$1

The approximation quality is nevertheless gauge dependent: in the tested $VV\to X$2 and $VV\to X$3 reactions, the unitary-gauge IEVBA performs better than the axial-gauge version (Bernreuther et al., 2015).

Modern LO EVA implementations for like-sign $VV\to X$4 scattering also expose non-uniqueness at the level of off-shell polarization vectors, transversality restoration, decay-current treatment, and relative sign conventions between transverse and longitudinal modes. In that context, all variants are evaluated with an on-shell projection of the $VV\to X$5 momenta in the VBS subprocess that forces the $VV\to X$6 momenta on-shell to guarantee gauge independence of the $VV\to X$7 matrix elements and preserves the locations of the photon poles in the $VV\to X$8- and $VV\to X$9-channel subdiagrams (Dittmaier et al., 2024).

4. Kinematic regimes: asymptotic, threshold, and multi-TeV limits

The standard asymptotic regime assumes a hard scale $WW$0 much larger than both the weak-boson mass and the boson virtuality, with

$WW$1

up to special corners of phase space (Borel et al., 2012). The approximation is therefore a statement about virtuality-scale separation rather than about a specific electroweak symmetry-breaking model (Borel et al., 2012).

Threshold behavior requires a different treatment. A focused threshold analysis showed that EVA is not restricted to $WW$2; near threshold the distribution functions acquire a natural shift

$WW$3

with support

$WW$4

For transverse $WW$5 emission from a fermion,

$WW$6

while the proposed threshold-corrected longitudinal distribution is

$WW$7

The corresponding $WW$8 distributions have the same shifted kinematic structure with the appropriate $WW$9 couplings (Alikhanov, 2018).

At the opposite extreme, collider processes at partonic center-of-mass energies of order $q \to qV$0 and above exhibit a regime in which electroweak gauge bosons mostly act as quasi-massless partons in vector boson fusion processes (Dahlén et al., 25 Jul 2025). In that regime the hard subprocess can often neglect weak-boson masses at leading power, while the masses still regulate the collinear splitting. A practical conclusion of recent Whizard studies is that dynamical scales around

$q \to qV$1

give the best overall agreement across many channels (Dahlén et al., 25 Jul 2025). In the muon-collider literature, the same logic is expressed by the statement that lepton colliders beyond a few TeV are effectively electroweak boson colliders (Ruiz et al., 2021).

5. Accuracy, validation, and known failure modes

The major limitation of EVA in realistic collider phenomenology is that the approximation is structurally tied to the collinear region, whereas experimental VBS selections often exclude that region. A precision review states the issue sharply: the event selection for VBS processes at hadron colliders via tagging two jets is the show-stopper of the EVA, because the transverse-momentum cuts on the tagging jets,

$q \to qV$2

exactly exclude the collinear splitting region for $q \to qV$3 which is the basis for the EVA construction. On that basis EVA is described as at best good at the qualitative level, and it cannot serve as basis for precise predictions (Dittmaier, 20 Apr 2026).

A complete NLO study of like-sign $q \to qV$4-boson scattering at the LHC confirmed this assessment. For

$q \to qV$5

different EVA versions based on the collinear emission of $q \to qV$6 bosons from incoming (anti)quarks were compared to the full LO result, while the paper’s VBS approximation reproduced the full NLO predictions within $q \to qV$7 in the most important regions of phase space (Dittmaier et al., 2024). Under typical VBS selection cuts, all EVA versions can only qualitatively describe the full VBS process. Even after inverting the jet-$q \to qV$8 cut to enforce softer jets, some EVA versions are good within $q \to qV$9 for distributions defined from the jet kinematics, but none are better than $W$0 for leptonic observables (Dittmaier et al., 2024).

Earlier LHC studies of strongly interacting electroweak symmetry breaking had used Equivalent Vector Boson Approximation as a baseline for vector boson scattering, but later work emphasized full six-fermion calculations. In studies of unitarized models, predictions were compared with analogous studies in Equivalent Vector Boson Approximation, while the final phenomenology was obtained from a complete calculation with six fermions in the final state implemented in the PHANTOM Monte Carlo (Ballestrero et al., 2011). This comparison established EVA/EVBA as a conceptual and historical baseline rather than the endpoint for realistic LHC counting experiments (Ballestrero et al., 2011).

The improved EVBA can be quantitatively useful in cleaner environments. At $W$1 TeV, for $W$2, the unitary-gauge IEVBA reaches about $W$3 accuracy with central and sufficiently hard $W$4 cuts, for example with $W$5 GeV and $W$6 (Bernreuther et al., 2015). For $W$7, accuracy of order $W$8 or better is reached for fairly loose cuts, while the axial-gauge IEVBA fails badly in the tested regimes because the longitudinal channels are much too large (Bernreuther et al., 2015).

6. Contemporary role in collider phenomenology

Recent highest-energy-collider studies give a more differentiated picture than the LHC VBS analyses. In Whizard-based comparisons of EVA with full matrix elements, di-Higgs production in $W$9 collisions is identified as a clean success case: the dominant contribution, about $Z$0, comes through the longitudinal structure function $Z$1, and for

$Z$2

EVA agrees very well with the full matrix element (Dahlén et al., 25 Jul 2025). Top-pair production is also found to be surprisingly well approximated for

$Z$3

whereas $Z$4 and $Z$5 production, being transverse-dominated and more exposed to weak bremsstrahlung or non-VBF topologies, are much less reliable (Dahlén et al., 25 Jul 2025). For vector-boson-pair channels some flattening of the EVA-to-full ratio appears only above roughly $Z$6, and for $Z$7 at $Z$8 TeV the EVA prediction is roughly $Z$9 below the inclusive full result (Dahlén et al., 25 Jul 2025).

Muon-collider studies reach a related conclusion from a PDF-based formulation. For benchmark processes with explicit hard-scale cuts,

$f_1 f_2 \to f_1' f_2' X$0

the agreement can be excellent: at $f_1 f_2 \to f_1' f_2' X$1 TeV, $f_1 f_2 \to f_1' f_2' X$2 gives full $f_1 f_2 \to f_1' f_2' X$3 fb and EWA $f_1 f_2 \to f_1' f_2' X$4 fb, while $f_1 f_2 \to f_1' f_2' X$5 gives full $f_1 f_2 \to f_1' f_2' X$6 fb and EWA $f_1 f_2 \to f_1' f_2' X$7 fb (Ruiz et al., 2021). The same work emphasizes that transverse channels carry large factorization-scale ambiguities from the bare PDFs, with an estimated $f_1 f_2 \to f_1' f_2' X$8 effect for $f_1 f_2 \to f_1' f_2' X$9 TeV, and recommends

$Z$00

for responsible use of the approximation (Ruiz et al., 2021).

In current practice EVA is therefore neither obsolete nor a precision replacement for full off-shell calculations. It remains a useful and computationally efficient approximation for identifying phase-space regions dominated by electroweak collinear splittings, especially in longitudinally dominated fusion and in heavy final states at very high energy (Dahlén et al., 25 Jul 2025, Ruiz et al., 2021). Modern implementations reflect that role. A public implementation of polarized and unpolarized electroweak boson PDFs was released in MadGraph5_aMC@NLO version 3.3.0 (Ruiz et al., 2021), while a corrected EVA implementation in Whizard, with default, log_pt, log, and legacy modes, is scheduled for Whizard v3.1.7 (Dahlén et al., 25 Jul 2025). The common methodological conclusion across the recent literature is that EVA should be validated against full matrix elements whenever transverse modes, recoil effects, non-VBF topologies, or realistic tagging-jet cuts are phenomenologically important (Dittmaier et al., 2024, Dittmaier, 20 Apr 2026).