betaFeedback

WorksPhysics and Astronomy

Precision Measurements of the Effective Leptonic Weak Mixing Angle at Hadron Colliders

Abstract

We review precision determinations of the effective leptonic weak mixing angle, sin⁡2θeffℓ\sin^2θ^\ell_{\mathrm{eff}}, from Drell-Yan angular asymmetries at the Tevatron and the LHC. As statistical uncertainties have dramatically decreased, parton distribution function (PDF) uncertainty now dominates the most precise hadron-collider results. We describe the angular observables, electroweak calculation schemes, and advanced methods such as replica reweighting and Hessian profiling, used to constrain PDFs directly from measured angular distributions. To identify the origin of differences between modern PDF sets (e.g., CT18, MSHT20, NNPDF) in the extracted mixing angle, we employ a simplified leading-order Z-pole calculation that separates quark-flavor production weights from quark-direction dilution. We demonstrate how these PDF-induced differences can be mitigated by incorporating complementary constraints from W/Z cross-section ratios and W-boson decay asymmetries. Furthermore, we discuss the requirements for a common reinterpretation and future combination of LHC measurements, highlighting how correlated PDF uncertainties limit the ultimate combined precision. Finally, we outline future prospects, including complementary measurements in Z+bZ+b production and high-mass tests of the running MS‾\overline{\mathrm{MS}} weak mixing angle.

Introduction

The weak mixing angle is a fundamental cornerstone of the Standard Model (SM) of particle physics. It describes the mixing of the neutral electroweak gauge fields into the photon and the Z boson. Precision measurements of this parameter provide one of the most stringent consistency tests of the electroweak sector and serve as powerful, model-independent probes for physics beyond the Standard Model (BSM). Because the effective mixing angle absorbs higher-order electroweak radiative corrections, it is highly sensitive to the presence of new BSM particles running in quantum loops. For instance, theoretical extensions such as two-Higgs-doublet models or theories with an additional U(1)U(1) gauge symmetry (predicting a Z′Z' boson) can shift the electroweak radiative corrections and modify the observed leptonic Z couplings [1, 2]. Therefore, pushing the precision limits of this parameter serves as a highly sensitive probe. for these BSM scenarios. This sensitivity to new physics can be enhanced by combining measurements of the mixing angle with those of the W-boson mass, which depends on a different combination of electroweak radiative corrections.

At tree level, the weak mixing angle can be expressed in terms of either the SU(2)LSU(2)_L and U(1)YU(1)_Y gauge couplings (gg and g′g'), or the W- and Z-boson masses (mWm_W and mZm_Z):

sin⁡2θW=g′2g2+g′2=1−mW2mZ2(tree level).(1)\sin^2\theta_W = \frac{g'^2}{g^2+g'^2} = 1-\frac{m_W^2}{m_Z^2}\quad\text{(tree level).} \tag*{(1)}

Beyond tree level, radiative corrections modify the relation between these two expressions, making the definition of the mixing angle dependent on the chosen renormalization scheme. For example, the on-shell scheme retains the mixing angle using the physical boson masses (sin⁡2θWOS≡1−mW2/mZ2\sin^2\theta_W^{\mathrm{OS}}\equiv1-m_W^2/m_Z^2 [3]), while the MS scheme introduces a scale-dependent running parameter based on renormalized gauge couplings. To avoid these scheme-dependent theoretical conversions when reporting experimental data, precision asymmetry measurements are typically reported in terms of the effective leptonic weak mixing angle, sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. This pseudo-observable [4, 5] is directly related to the measurable ratio of the effective vector (gVℓg_V^\ell) and axial-vector (gAℓg_A^\ell) couplings of the Z boson to leptons at the Z pole.

sin⁡2θeffℓ=14(1−Re⁡[gVℓgAℓ]).(2)\sin^2\theta_{\mathrm{eff}}^\ell= \frac{1}{4}\left(1-\operatorname{Re}\left[\frac{g_V^\ell}{g_A^\ell}\right]\right). \tag*{(2)}

Historically, the most precise determinations of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell came from electron-positron (e+e−e^+e^-) collisions at the Large Electron-Positron collider (LEP) and the SLD experiment at the Stanford Linear Collider. By measuring various parity-violating observables, including the forward-backward asymmetry, the left-right asymmetry, and the tau polarization, these experiments provided six individual determinations, which are summarized in Figure 1. They achieved a combined value of sin⁡2θeffℓ=0.23153±0.00016\sin^2\theta_{\mathrm{eff}}^\ell=0.23153\pm0.00016. While this combined value is in excellent agreement with the SM prediction, the internal consistency of the six measurements exhibits χ2/ndf=11.8/5\chi^2/\mathrm{ndf}=11.8/5, corresponding to a p-value of approximately 4% [4]. This level of internal spread provides a motivation for independent determinations.

Figure 1. Measurements of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell [6, 7, 8, 9, 10, 11]. Error bars show total uncertainties. Bold entries, separated by horizontal lines, indicate the LEP+SLD [4] and CDF+D0 [12] combinations. The ATLAS 8 TeV result is preliminary. The orange band shows the 2026 PDG SM prediction, calculated without using weak-mixing-angle measurements as inputs [3].

Today, the landscape of precision electroweak physics has expanded to hadron colliders, which offer a compelling testing ground for this purpose. Measurements of the angular distributions in Drell–Yan dilepton production (qqˉ→γ∗/Z→ℓ+ℓ−q\bar{q}\to\gamma^*/Z\to\ell^+\ell^-) at the Tevatron and the LHC (also shown in Figure 1) utilize completely different initial states and involve experimental systematics that are largely orthogonal to those of e+e−e^+e^- colliders. Consequently, precision determinations at hadron colliders are not merely complementary; they are crucial for providing an independent perspective on this fundamental parameter.

However, extracting precision electroweak parameters at hadron colliders presents a unique and profound challenge: the initial state involves composite protons (or antiprotons). The predicted angular distributions heavily depend on the parton distribution functions (PDFs) of the incoming hadrons. Because the precise momentum fraction and direction of the interacting quark and antiquark are not known on an event-by-event basis, incorrect quark direction assignments dilute the measured forward-backward asymmetry. Furthermore, the relative contributions from different quark flavors (e.g., up, down, strange) strongly depend on the PDFs. Consequently, as the statistical uncertainties have been substantially reduced with the large LHC datasets, the uncertainty associated with PDFs has become the dominant barrier in the most precise hadron-collider determinations of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell.

This review provides a comprehensive overview of precision effective weak mixing angle measurements at hadron colliders, with a central focus on the methodologies developed to understand and constrain the associated PDF uncertainties. In Section 2, we describe the relevant angular observables, electroweak calculations, and modern techniques for constraining PDFs directly from measured angular distributions, including replica reweighting and Hessian profiling. Section 3 reviews the measurements and analysis strategies from the Tevatron (CDF and D0), and LHC (ATLAS, CMS, and LHCb) experiments. In Section 4, we employ a simplified leading-order Z-pole calculation to identify the origin of PDF-set dependence, separating quark-flavor production weights from quark-direction dilution. Section 5 discusses the requirements and challenges for a common reinterpretation and future combination of LHC measurements, highlighting the impact of correlated PDF uncertainties. Finally, in Section 6, we discuss future prospects, including complementary constraints from Z+bZ+b production and high-mass tests of the running weak mixing angle above the Z pole.

Angular observables and extraction methods

This section describes how Drell–Yan angular measurements are used to extract the effective leptonic weak mixing angle. We first introduce the angular observables, quark-direction dilution, and the treatment of acceptance and detector effects. We then discuss the electroweak corrections used in the extraction and the methods for constraining PDFs with the measured angular distributions.

The forward–backward asymmetry is measured from the polar-angle distribution of the negative lepton. In the Collins–Soper frame [13], this angle is denoted by θ\theta, with the forward direction assigned using an estimate of the incoming quark direction. After integration over the azimuthal angle ϕ\phi, the distribution takes the form

dσdcos⁡θ∝1+cos⁡2θ+A02(1−3cos⁡2θ)+A4cos⁡θ.(3)\frac{\mathrm{d}\sigma}{\mathrm{d}\cos\theta} \propto1+\cos^2\theta+\frac{A_0}{2}(1-3\cos^2\theta)+A_4\cos\theta. \tag*{(3)}

The term A4cos⁡θA_4\cos\theta is odd under cos⁡θ→−cos⁡θ\cos\theta\to-\cos\theta and determines the forward–backward difference. For full angular acceptance, integration over cos⁡θ\cos\theta yields AFBfull=(3/8)A4A_{\mathrm{FB}}^{\mathrm{full}}=(3/8)A_4. The term proportional to A0A_0 integrates to zero for full angular acceptance but can affect the asymmetry normalization after lepton cuts. Acceptance corrections therefore depend on the QCD modeling of the dilepton recoil and angular coefficients. The angular coefficients A4A_4 and A0A_0 depend on the dilepton invariant mass m≡mℓℓm \equiv m_{\ell\ell}, rapidity y≡yℓℓy \equiv y_{\ell\ell}, and transverse momentum. The angular coefficient A4(m,y)A_4(m,y) (integrated over transverse momentum) is used in the extraction of the weak mixing angle. Near the Z pole, the dependence of A4A_4 on the leptonic Z couplings provides sensitivity to sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell.

To predict A4A_4 or AFBA_{\mathrm{FB}}, the partonic angular distributions must be combined with the PDFs of the incoming hadrons. At leading order in QCD, the contributions from qqˉ→γ∗/Z→ℓ+ℓ−q\bar{q} \to\gamma^*/Z \to\ell^+\ell^- have the form

∑q[fq/h1(x1,μF2)fqˉ/h2(x2,μF2)σ^qqˉ+fqˉ/h1(x1,μF2)fq/h2(x2,μF2)σ^qˉq](4)\sum_q \left[f_{q/h_1}(x_1,\mu_F^2)f_{\bar q/h_2}(x_2,\mu_F^2)\hat{\sigma}_{q\bar q}+f_{\bar q/h_1}(x_1,\mu_F^2)f_{q/h_2}(x_2,\mu_F^2)\hat{\sigma}_{\bar q q}\right] \tag*{(4)}

Here h1h_1 and h2h_2 denote the incoming hadrons, and σ^\hat{\sigma} denotes the partonic cross section. The momentum fractions are x1,2=(m/s)e±yx_{1,2}=(m/\sqrt{s})e^{\pm y}, with factorization scale μF≃m\mu_F \simeq m. The quark flavors contribute with different electroweak couplings and partonic asymmetries and their PDF weights affect the predicted AFBA_{\mathrm{FB}}. The two terms correspond to the quark entering from opposite beams, whose relative contributions also affect the quark-direction assignment. Higher-order QCD predictions include radiation and additional channels such as the quark-gluon (qgqg) process. The displayed expression illustrates the flavor and direction effects at Born level.

In proton–antiproton collisions at the Tevatron, the incoming quark is more likely to come from the proton than from the antiproton. The proton contains valence quarks, while the antiproton contains valence antiquarks, favoring this assignment. The proton-beam direction is therefore used to define the forward direction. Events with an antiquark from the proton and a quark from the antiproton are assigned the opposite quark direction. Their contribution to the forward–backward difference changes sign and dilutes the asymmetry.

In proton–proton collisions at the LHC, the two beams have the same parton distributions, so neither beam is a preferred quark source. However, valence quarks typically carry a larger momentum fraction than sea antiquarks. The longitudinal boost direction of the dilepton system is therefore used to estimate the quark direction and define the forward direction. At small ∣y∣|y|, the incoming partons have similar momentum fractions. The boost direction then gives a less reliable estimate of the quark direction, and the asymmetry is strongly diluted. At larger ∣y∣|y|, the momentum imbalance generally improves the assignment and the dilution is reduced. Both the dilution and the flavor weights depend on PDFs and vary with dilepton mass and rapidity. Measurements in mass and rapidity bins therefore provide information on PDFs as well as sensitivity to sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell.

A forward–backward asymmetry can be measured as (NF−NB)/(NF+NB)(N_F-N_B)/(N_F+N_B), where NFN_F and NBN_B are the numbers of events with positive and negative cos⁡θ\cos\theta, respectively. This uses only the sign of cos⁡θ\cos\theta, although the sensitivity to A4A_4 varies with ∣cos⁡θ∣|\cos\theta|. Fitting the angular distribution or comparing yields in matched angular bins retains this additional information. At fixed dilepton kinematics, the local asymmetry in narrow bins at +c+c and −c-c, with c=∣cos⁡θ∣c=|\cos\theta|, follows

N(c)−N(−c)N(c)+N(−c)=A4c1+c2+A02(1−3c2)(5)\frac{N(c)-N(-c)}{N(c)+N(-c)}=\frac{A_4c}{1+c^2+\frac{A_0}{2}(1-3c^2)} \tag*{(5)}

Here NN denotes the signal yield. Acceptance and efficiency may vary substantially with cc, while remaining similar between the matched positive and negative bins. In the limit of equal responses and negligible bin migration, their common factor cancels between the numerator and denominator. Residual forward–backward differences are often much smaller and can be included, together with bin migration effects, in reconstructed-level templates or unfolding.

The angular dependence can also be used without explicit binning in cos⁡θ\cos\theta. Angular weighting uses the cc dependence in Eq. (5) to combine information on the common coefficient A4A_4 from different angles, with weights reflecting the statistical sensitivity at each angle [14]. The weighted event sums give a single asymmetry estimate, AFBwA_{\mathrm{FB}}^w, in each dilepton mass and rapidity bin. In the ideal limit discussed above, this estimates AFBfull=(3/8)A4A_{\mathrm{FB}}^{\mathrm{full}}=(3/8)A_4. The method can improve statistical sensitivity relative to simple counting. Avoiding an additional angular binning also reduces the number of analysis bins and the need to handle sparsely populated bins.

The measured AFBA_{\mathrm{FB}} or A4A_4 distributions provide the input to a fit of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. This fit can be performed at reconstructed level using templates that include detector effects and QED final-state radiation (FSR), or after unfolding to a specified lepton definition. For unfolded results, the acceptance and treatment of FSR define the quantity being fitted. In either approach, the mixing angle is varied in the theory prediction, whose electroweak and PDF dependence is discussed below [8, 10].

Electroweak corrections and input schemes

The effective leptonic mixing angle is defined by the leptonic Z-pole coupling ratio in Eq. (2). In hadron-collider measurements, the fitted angular distributions also depend on quark couplings and cover a range of dilepton masses. An electroweak prescription is therefore needed to relate the fitted mixing angle to these distributions, including flavor-dependent weak corrections, photon–Z interference, and the mass dependence of the form factors [4, 15].

Earlier extractions incorporated weak corrections through effective couplings or electroweak form factors while retaining the Born structure of the neutral-current amplitude. As experimental precision improved, more complete implementations accounted for differences between the effective couplings of leptons and quarks, complex and mass-dependent form factors, and photon vacuum polarization [12]. Such form-factor treatments are commonly described as improved or enhanced Born approximations (IBA/EBA). The form factors can be calculated with ZFITTER, which uses the Dizet library for the weak corrections [16, 17].

A further step was to define the fitted effective mixing angle as an independent renormalized input of the electroweak calculation. The (Gμ,mZ,sin⁡2θeffℓ)(G_{\mu},m_Z,\sin^2\theta_{\mathrm{eff}}^\ell) scheme was formulated for Drell–Yan calculations in 2019 [15]. In this scheme, the mixing angle can be varied directly while the quark couplings and off-pole corrections are calculated consistently. The renormalization conditions preserve the leptonic pole definition at each perturbative order. This scheme is implemented in Z\textsubscript{EW}-BMNNPV within POWHEG-BOX-V2, allowing the effective mixing angle to be fitted directly in a full NLO electroweak calculation of Drell–Yan production [15, 18, 19].

Using sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell as an input absorbs important pole corrections into its definition. Near the Z pole, published comparisons find smaller residual universal higher-order corrections and a weaker top-quark-mass dependence in AFBA_{\mathrm{FB}} than in other widely used input schemes, such as (Gμ,mW,mZ)(G_{\mu},m_W,m_Z) and (α(0),Gμ,mZ)(\alpha(0),G_{\mu},m_Z) [15, 19]. These features make the sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell input scheme well suited for precision mixing angle extractions, and it has been adopted in the recent CMS and LHCb measurements using 13 TeV Run 2 data samples [10, 11].

PDF constraints from angular asymmetries

PDFs are a leading source of systematic uncertainty in hadron-collider determinations of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. Their relative importance increases as larger data samples reduce the statistical uncertainty. Different PDF sets may yield different extracted mixing angles from the same measured asymmetry distributions [10]. Therefore, both the uncertainty within a PDF set and the dependence of the extracted mixing angle on the choice of PDF set need to be understood.

The dilepton mass and rapidity dependence of the measured AFBA_{\mathrm{FB}} or A4A_4 distributions can be used to constrain the PDFs within the same fit that is used in the extraction of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell [20]. Changing the mixing angle mainly affects the leptonic Z couplings, while PDF variations change the quark-flavor mixture and quark-direction dilution. These effects have different mass and rapidity dependences, as illustrated in Figure 2. A fit in both mass and rapidity bins can therefore distinguish some PDF variations from a mixing angle shift. This can reduce both the propagated PDF uncertainty and the spread in fitted mixing angles extracted with different PDF sets.

POWHEG predictions for the dilepton mass dependence of $A_{\mathrm{FB}}$

Figure 2. POWHEG predictions for the dilepton mass dependence of AFBA_{\mathrm{FB}} in the CMS 8 TeV analysis, taken from Fig. 5 of Ref. [8]. The left panels integrate over rapidity, while the right panels show six rapidity bins. Upper panels give the predictions at sin⁡2θeffℓ=0.23120\sin^2\theta_{\mathrm{eff}}^\ell= 0.23120. Lower panels show changes in AFBA_{\mathrm{FB}} from mixing angle variations (solid curves) and 100 NNPDF3.0 replicas (dashed curves). The shaded bands indicate the PDF standard deviation.

PDF uncertainties are commonly provided as Monte Carlo replicas or Hessian eigenvector variations [21, 22, 23]. In both representations, the published error ensemble supplies the prior PDF uncertainty, while the measured asymmetries enter through a common data likelihood fit. The mixing angle is fitted together with the PDF variations (profiling). Let seff2=sin⁡2θeffℓs_{\mathrm{eff}}^2=\sin^2\theta_{\mathrm{eff}}^\ell, let dd contain the measured asymmetries, and let t(seff2,f)t(s_{\mathrm{eff}}^2,f) denote the predictions for a PDF ff. The data contribution is

χA2(seff2,f)=[d−t(seff2,f)]TV−1[d−t(seff2,f)],(6)\chi_A^2(s_{\mathrm{eff}}^2,f)=\left[d-t(s_{\mathrm{eff}}^2,f)\right]^{\mathsf{T}}V^{-1}\left[d-t(s_{\mathrm{eff}}^2,f)\right], \tag*{(6)}

where VV is the covariance matrix of the measured distributions. Systematic effects can instead be treated with constrained nuisance parameters in the fit.

In the replica representation, the PDF prior is sampled by an ensemble of equally weighted members frf_r. For each member, one finds seff,r2s_{\mathrm{eff},r}^2, obtaining a best-fit value sˉeff2\bar{s}_{\mathrm{eff}}^2 and a minimum χA,r2\chi_{A,r}^2. The Giele–Keller (GK) prescription assigns normalized likelihood weights [24, 25]

wr=exp⁡(−χA,r2/2)∑kexp⁡(−χA,k2/2),sˉeff2=∑rwrseff,r2,σPDF2=∑rwr(seff,r2−sˉeff2)2.(7)\begin{aligned} w_r &= \frac{\exp(-\chi_{A,r}^2/2)}{\sum_k\exp(-\chi_{A,k}^2/2)}, \qquad&\bar{s}_{\mathrm{eff}}^2 &= \sum_r w_r s_{\mathrm{eff},r}^2,\\ \sigma_{\mathrm{PDF}}^2 &= \sum_r w_r\left(s_{\mathrm{eff},r}^2-\bar{s}_{\mathrm{eff}}^2\right)^2. \tag*{(7)} \end{aligned}

Replicas that give a poorer description of the data receive smaller weights. The weighted spread of seff,r2s_{\mathrm{eff},r}^2 in Eq. (7) represents the PDF component. The within-member fit uncertainty is also retained when determining the total uncertainty. The effective ensemble size, Neff=1/∑rwr2N_{\mathrm{eff}} = 1/\sum_r w_r^2 in this normalization, measures whether sufficiently many replicas contribute to the result.

In the Hessian representation, PDF variations are described by eigenvector directions about the central fit. Let βk\beta_k be PDF nuisance parameters normalized to the published one-standard-deviation variations, with a confidence-level conversion applied where necessary. To linear order, the prediction and the profiling objective can be written as

t(seff2,β)≃t0(seff2)+∑kDk(seff2)βk,χprof2(seff2,β)=χA2(seff2,β)+∑kβk2(8)t(s_{\mathrm{eff}}^2,\boldsymbol{\beta}) \simeq t_0(s_{\mathrm{eff}}^2) + \sum_k D_k(s_{\mathrm{eff}}^2)\beta_k, \qquad \chi^2_{\mathrm{prof}}(s_{\mathrm{eff}}^2,\boldsymbol{\beta}) = \chi_A^2(s_{\mathrm{eff}}^2,\boldsymbol{\beta}) + \sum_k \beta_k^2 \tag*{(8)}

where DkD_k is obtained from the PDF error members, for example from half their positive-minus-negative difference for a symmetric eigenvector pair. The quadratic penalty encodes the Gaussian PDF prior. Minimization shifts and constrains the PDF directions favored by the data. At each trial seff2s_{\mathrm{eff}}^2, the profiled PDF and other nuisance parameters are minimized, retaining their correlations with the mixing angle. A unit increase of the profiled objective gives the usual one-parameter interval in the Gaussian approximation. For linear predictions, a Gaussian prior, and a common data likelihood and uncertainty normalization, GK replica reweighting and Hessian profiling give equivalent central values and covariance with adequate replica sampling [26].

Some global Hessian PDF fits use a tolerance to account for tensions among datasets and limitations of the parametrization and theory [22, 23]. Instead of the usual Δχ2=1\Delta\chi^2 = 1 criterion, they allow a larger increase in the global-fit objective, Δχ2=T2\Delta\chi^2 = T^2, where T>1T > 1 can depend on the eigenvector direction. Profiling uses the resulting broader uncertainty as a PDF prior. The prior constraint is therefore weaker, while the asymmetry measurement retains its measured covariance. This treatment can give different results from a global refit including the same data.

In these precision electroweak extractions, because no additional tolerance is applied to the asymmetry data, their experimental covariance and the theory uncertainties must be reliably evaluated. In measurements of asymmetries and cross section ratios common normalization, efficiency and acceptance uncertainties mostly cancel, and in statistically dominated bins the covariance is largely determined by event counts. The fit quality, the shifts and constraints of individual nuisance parameters, and the stability of the extracted mixing angle with alternative PDF sets should be checked. Using the profiled PDFs outside the probed (x,μF)(x,\mu_F) region requires independent validation.

Section 3 describes how these methods are used in individual measurements. Table 1 summarizes the electroweak and PDF treatments and uncertainties of selected measurements.

MeasurementObservableEW treatmentPDF treatment
CDF 1.96 TeV [12, 29]<br>2016, 9.4 fb−1^{-1}<br>Uncertainty<br>Stat. 43<br>PDF 16<br>Total 46Observable<br>Angular-weighted AFB(m)A_{\mathrm{FB}}(m)<br>(unfolded)<br><br>Channel<br>eeee (central–central, central–forward)<br><br>μμ\mu\mu (central–central)EW treatment<br>ZFITTER/Dizet EW corrections<br><br>Lepton acceptance<br>Central 0.05<∣η∣<1.050.05 < |\eta| < 1.05<br>Forward 1.2<∣η∣<2.81.2 < |\eta| < 2.8<br>Trigger muon ∣η∣<1|\eta| < 1<br>Partner acceptance to ∣η∣≃1.5|\eta| \simeq1.5PDF treatment<br>NNPDF3.0 NNLO replicas<br>GK reweighting<br><br>Dilepton rapidity<br>∣y∣<1.7|y| < 1.7<br>∣y∣<1|y| < 1
D0 1.96 TeV [12, 30]<br>2018, 9.7 fb−1^{-1}<br>Uncertainty<br>Stat. 35<br>PDF 19<br>Total 40Observable<br>AFB(m)A_{\mathrm{FB}}(m)<br>(detector level)<br><br>Channel<br>eeee (central–central, central–forward, forward–forward)<br><br>μμ\mu\mu (central–central)EW treatment<br>ZFITTER/Dizet EW corrections<br><br>Lepton acceptance<br>Central ∣η∣<1.1|\eta| < 1.1<br>Forward 1.5<∣η∣<3.21.5 < |\eta| < 3.2<br>Both ∣η∣<1.8|\eta| < 1.8<br>At least one ∣η∣<1.6|\eta| < 1.6PDF treatment<br>NNPDF3.0 NNLO reference<br>No reweighting<br><br>Dilepton rapidity<br>No separate cut<br>No separate cut
ATLAS 8 TeV [9]<br>preliminary<br>2018, 20.2 fb−1^{-1}<br>Uncertainty<br>Stat. 21<br>PDF 24<br>Total 36Observable<br>A4(m,∣y∣)A_4(m, |y|)<br>(unfolded)<br><br>Channel<br>eeee, μμ\mu\mu (central–central)<br><br>eeee (central–forward)EW treatment<br>ZFITTER/Dizet EW corrections<br><br>Lepton acceptance<br>∣ηℓ∣<2.4|\eta_\ell| < 2.4<br>Central ∣ηe∣<2.4|\eta_e| < 2.4<br>Forward 2.5<∣ηe∣<4.92.5 < |\eta_e| < 4.9PDF treatment<br>MMHT14 NNLO Hessian<br>Profiling<br><br>Dilepton rapidity<br>∣y∣<2.5|y| < 2.5<br>1.6<∣y∣<3.61.6 < |y| < 3.6
CMS 8 TeV [8]<br>2018, 19.6 fb−1^{-1}<br>Uncertainty<br>Stat. 36<br>PDF 31<br>Total 53Observable<br>Angular-weighted AFB(m,∣y∣)A_{\mathrm{FB}}(m, |y|)<br>(detector level)<br><br>Channel<br>eeee, μμ\mu\mu (central–central)EW treatment<br>Effective Born approximation<br><br>Lepton acceptance<br>∣ηℓ∣<2.4|\eta_\ell| < 2.4PDF treatment<br>NNPDF3.0 NLO replicas<br>GK reweighting<br><br>Dilepton rapidity<br>∣y∣<2.4|y| < 2.4
CMS 13 TeV [10]<br>2025, 138 fb−1^{-1}<br>Uncertainty<br>Stat. 10<br>PDF 27<br>Total 31Observable<br>Angular-weighted AFB(m,∣y∣)A_{\mathrm{FB}}(m, |y|)1<br>(detector level)<br><br>Channel<br>eeee, μμ\mu\mu (central–central)<br><br>eeee (central–forward, ECAL)<br><br>eeee (central–forward, HCAL)EW treatment<br>sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell input scheme<br><br>Lepton acceptance<br>∣ηe∣<2.5|\eta_e| < 2.5, ∣ημ∣<2.4|\eta_\mu| < 2.4<br>Central ∣ηe∣<2.5|\eta_e| < 2.5<br>Forward 2.5<∣ηe∣<2.872.5 < |\eta_e| < 2.87<br>Central 1.57<∣ηe∣<2.51.57 < |\eta_e| < 2.5<br>Forward 3.14<∣ηe∣<4.363.14 < |\eta_e| < 4.36PDF treatment<br>CT18Z NNLO Hessian<br>Profiling<br><br>Dilepton rapidity<br>∣y∣<2.4|y| < 2.4<br>1.2<∣y∣<2.71.2 < |y| < 2.7<br>2.0<∣y∣<3.42.0 < |y| < 3.4
LHCb 13 TeV [11]<br>2024, 5.4 fb−1^{-1}<br>Uncertainty<br>Stat. 44<br>PDF 21<br>Total 50Observable<br>AFB(∣Δημμ∣)A_{\mathrm{FB}}(|\Delta\eta_{\mu\mu}|)<br>(unfolded)<br><br>Channel<br>μμ\mu\mu (forward–forward)EW treatment<br>sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell input scheme<br><br>Lepton acceptance<br>2<ημ<4.52 < \eta_\mu< 4.5PDF treatment<br>Average of CT18, MSHT20, NNPDF3.1 NLO<br>No profiling<br><br>Dilepton rapidity<br>No separate cut

Table 1. Selected Drell–Yan measurements of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell, with their analysis methods and approximate kinematic coverage. The publication year and integrated luminosity are shown for each measurement, using the larger luminosity when channels differ. Uncertainties are given in units of 10−510^{-5}. Total uncertainties include contributions beyond the statistical and PDF components and account for the reported correlations. The CDF and D0 entries refer to their combined electron and muon results.

Tevatron and LHC measurements

We review the CDF and D0 measurements of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell and their Tevatron combination, followed by the ATLAS, CMS, and LHCb results. For each experiment, we describe the angular observables and kinematic coverage, the electroweak and PDF treatments used in the extraction, and the main sources of uncertainty. Figure 1 compares the reported values, and Table 1 summarizes the analysis methods and uncertainties.

The Tevatron and LHC measurements have different strengths and limitations. At the Tevatron, quark-direction dilution is relatively small, as discussed in § 2, but the available Drell–Yan statistics limit the precision. ATLAS and CMS benefit from larger cross sections and the high integrated luminosities accumulated at the LHC. In their central regions, however, the asymmetry is more strongly diluted and the extraction is more sensitive to PDFs than at the Tevatron. Forward-lepton channels extend the rapidity coverage, while binning in rapidity retains the different sensitivities of central and forward events. LHCb benefits from smaller dilution and PDF uncertainties in the forward region, but its lower integrated luminosity and restricted acceptance leave the measurement limited by statistics [11]. Figure 3 compares the momentum-fraction ranges probed by these acceptances.

Figure 3. Parton momentum-fraction ranges near the Z pole for representative Tevatron and LHC rapidity intervals. The horizontal bars show xlox_{\mathrm{lo}} and xhix_{\mathrm{hi}} from the leading-order relation xhi,lo=(mZ/s)e±∣y∣x_{\mathrm{hi,lo}}=(m_Z/\sqrt{s})e^{\pm|y|}, using the collision energies and dilepton rapidity ranges indicated. The solid and dashed curves show CT18Z NNLO xuvx u_v and xuˉx\bar{u}, respectively, at μF=mZ\mu_F=m_Z, with uv=u−uˉu_v=u-\bar{u}.

Tau polarization provides a complementary probe of the leptonic Z couplings without estimating the incoming quark direction. ATLAS measured this observable at 8 TeV [27], and CMS extracted sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell from its 13 TeV measurement [28]. The following discussion focuses on electron and muon Drell–Yan asymmetries, which currently give more precise mixing-angle determinations.

Tevatron measurements

CDF and D0 measured the Drell–Yan forward–backward asymmetry in electron and muon final states at s=\sqrt{s} = 1.96 TeV. In proton–antiproton collisions, the incoming quark is more likely to originate from the proton, allowing the proton-beam direction to define forward. The resulting asymmetry is less diluted than in proton–proton collisions.

CDF used approximately 9 fb−1^{-1} of Tevatron Run II data in each lepton channel. Angular weighting retained information on the variation of the statistical sensitivity with the lepton angle. The resulting AFBw(m)A_{\mathrm{FB}}^{w}(m) distributions were unfolded for detector effects and QED final-state radiation, then compared with POWHEG templates incorporating ZFITTER electroweak form factors. NNPDF3.0 NNLO replicas were reweighted using the measured asymmetries, allowing the same data to constrain the PDFs used in the mixing angle extraction. The combined electron–muon result was sin⁡2θeffℓ=0.23221±0.00046\sin^{2}\theta_{\mathrm{eff}}^{\ell} = 0.23221 \pm0.00046 [12, 29].

D0 used 9.7 fb−1^{-1} of electron data and 8.6 fb−1^{-1} of muon data. It determined AFB(m)A_{\mathrm{FB}}(m) from forward and backward event counts, integrated over the accepted rapidity range. The mixing angle was extracted by comparison with reconstructed-level templates that included the detector response. Periodic reversals of the magnetic fields helped suppress detector-induced charge asymmetries [12, 30].

The final CDF and D0 central values were expressed using a common ZFITTER-based electroweak prescription and NNPDF3.0 NNLO PDFs. For D0, this involved corrections to the mixing angles extracted with Pythia templates, including the update from NNPDF2.3 to NNPDF3.0 for the electron channel. The PDF uncertainties retained the original replica evaluations, without reweighting to the D0 asymmetry data. The corrected electron and muon results gave sin⁡2θeffℓ=0.23095±0.00040\sin^{2}\theta_{\mathrm{eff}}^{\ell} = 0.23095 \pm0.00040 [12, 30].

The CDF and D0 results were then combined with BLUE, treating their PDF uncertainties as fully correlated. The Tevatron result was

sin⁡2θeffℓ=0.23148±0.00033(9)\sin^{2}\theta_{\mathrm{eff}}^{\ell} = 0.23148 \pm0.00033 \tag*{(9)}

with statistical and PDF contributions of 0.00027 and 0.00018, respectively. The combined precision remained mainly limited by statistics. The compatibility test of the CDF and D0 inputs gives a p-value of 2.6%, providing additional motivation for independent measurements at the LHC [12].

ATLAS measurements

The higher Drell–Yan cross section at the LHC provides ATLAS with more events per unit integrated luminosity than at the Tevatron. The larger quark-direction dilution in pppp collisions increases the PDF sensitivity of the mixing angle extraction. The 2015 ATLAS 7 TeV measurement [6] used 4.8 fb−1^{-1} of electron data and 4.6 fb−1^{-1} of muon data. Central–forward electron pairs were included alongside the central electron and muon channels, extending the dilepton rapidity coverage and reducing quark-direction dilution. Despite having the fewest selected events, the forward-electron channel gave the smallest statistical uncertainty on the mixing angle among the three channels [6].

The extraction used ATLAS-epWZ12 LO, a PDF fit prepared for this analysis using HERA inclusive DIS data and ATLAS 7 TeV W- and Z-boson measurements. The combined result was sin⁡2θeffℓ=0.23080±0.00120\sin^{2}\theta_{\mathrm{eff}}^{\ell} = 0.23080 \pm0.00120, with a PDF uncertainty of 0.00090 evaluated using this set [6].

The 2018 preliminary ATLAS 8 TeV analysis [9] used 20.2 fb−1^{-1} and retained the three channels, covering dilepton rapidities up to ∣y∣=3.6|y| = 3.6. The Born-level angular coefficients in the full lepton decay phase space were determined by fitting reconstructed (cos⁡θ,ϕ)(\cos\theta, \phi) distributions in mass and rapidity bins with templates that included the detector response. The mixing angle extraction used the predicted dependence of A4(m,∣y∣)A_{4}(m, |y|) on sin⁡2θeffℓ\sin^{2}\theta_{\mathrm{eff}}^{\ell}. The predictions were obtained with DYTurbo, including electroweak form factors from DIZET 6.21 through TauSpinner event weights [9, 17].

The mixing angle fit used MMHT14 NNLO and profiled its eigenvectors, allowing the angular data to constrain the PDFs [9]. The ATLAS 2018 8 TeV preliminary result is

sin⁡2θeffℓ=0.23140±0.00021stat±0.00024PDF±0.00016syst,(10)\sin^{2}\theta_{\mathrm{eff}}^{\ell} = 0.23140 \pm0.00021_{\mathrm{stat}} \pm0.00024_{\mathrm{PDF}} \pm0.00016_{\mathrm{syst}}, \tag*{(10)}

with a total uncertainty of 0.00036. The statistical and PDF contributions are comparable, and the PDF uncertainty is larger than those reported by CDF and D0.

CMS measurements

CMS also benefits from higher Drell–Yan event rates and faces similar limitations from quark-direction dilution and PDFs. Unlike ATLAS, the 2018 CMS 8 TeV measurement [8] used only central electron and muon channels, with 19.6 fb−1^{-1} of dielectron data and 18.8 fb−1^{-1} of dimuon data. The mixing angle was extracted from detector-level angular-weighted AFBℓ(m,∣y∣)A_{\mathrm{FB}}^{\ell}(m, |y|) distributions using POWHEG templates with NNPDF3.0 NLO replicas. Giele–Keller reweighting used the measured mass and rapidity dependence to constrain these replicas, reducing the PDF uncertainty from 0.00057 to 0.00031, including the finite-ensemble contribution [8]. The combined electron–muon result was

sin⁡2θeffℓ=0.23101±0.00036stat±0.00018syst±0.00016theo±0.00031PDF,(11)\sin^2\theta_{\mathrm{eff}}^{\ell} = 0.23101 \pm0.00036_{\mathrm{stat}} \pm0.00018_{\mathrm{syst}} \pm0.00016_{\mathrm{theo}} \pm0.00031_{\mathrm{PDF}}, \tag*{(11)}

with a total uncertainty of 0.00053. After reweighting, the statistical and PDF contributions were comparable.

The 2025 CMS 13 TeV analysis [10] used 138 fb−1^{-1} and added forward-electron channels, extending the dilepton rapidity coverage from ∣y∣<2.4|y| < 2.4 to ∣y∣<3.4|y| < 3.4. These channels provided sensitivity in a region with smaller quark-direction dilution. Electrons beyond the tracking acceptance were reconstructed using the forward calorimeters, including the forward hadron calorimeter beyond the crystal ECAL. Their use required dedicated identification, efficiency measurements, energy-response corrections, and control of jet backgrounds [10].

The mixing angle was extracted by fitting detector-level angular-weighted AFBℓA_{\mathrm{FB}}^{\ell} in mass and rapidity bins using POWHEG MiNNLOPS simulation. Weak corrections and mixing angle variations were evaluated with Z_ew-BMNNPV using the sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^{\ell} input scheme described in § 2.1. The analysis used CT18Z NNLO Hessian PDFs, whose eigenvectors were profiled together with the mixing angle and the other nuisance parameters [10].

The baseline result was

sin⁡2θeffℓ=0.23152±0.00010stat±0.00015exp±0.00008theo±0.00027PDF,(12)\sin^2\theta_{\mathrm{eff}}^{\ell} = 0.23152 \pm0.00010_{\mathrm{stat}} \pm0.00015_{\mathrm{exp}} \pm0.00008_{\mathrm{theo}} \pm0.00027_{\mathrm{PDF}}, \tag*{(12)}

with a total uncertainty of 0.00031 after accounting for correlations. The statistical uncertainty was more than a factor of three smaller than at 8 TeV, leaving the PDF contribution as the dominant uncertainty. Alternative PDF sets gave total uncertainties between 0.00024 and 0.00035 [10].

CMS also published unfolded A4(m,∣y∣)A_4(m, |y|) distributions before QED final-state radiation, together with their covariance matrices. Fits to these distributions and direct fits to the (m,∣y∣,cos⁡θ)(m, |y|, \cos\theta) distributions gave mixing angle results compatible with the angular-weighted AFBℓA_{\mathrm{FB}}^{\ell} extraction [10].

LHCb measurements

LHCb measured the dimuon asymmetry with both muons in 2.0<ημ<4.52.0 < \eta_{\mu} < 4.5. This forward acceptance reduces quark-direction dilution and gives good sensitivity to the mixing angle. The 7 and 8 TeV measurements used 1 and 2 fb−1^{-1}, respectively, to determine AFB(mμμ)A_{\mathrm{FB}}(m_{\mu\mu}) [7].

The 2024 LHCb 13 TeV analysis [11] used 5.4 fb−1^{-1} recorded in 2016–2018 and measured AFBA_{\mathrm{FB}} in bins of the absolute muon pseudorapidity difference, ∣Δημμ∣|\Delta\eta_{\mu\mu}|, integrated over 66<mμμ<11666 < m_{\mu\mu} < 116 GeV and the accepted dilepton rapidity range. At negligible dilepton transverse momentum, cos⁡θCS≃tanh⁡(∣Δημμ∣/2)\cos\theta_{\mathrm{CS}} \simeq\tanh(|\Delta\eta_{\mu\mu}|/2). This binning therefore separates events with different angular sensitivity to the mixing angle [11].

The measured asymmetries were corrected to the fiducial bare-muon level before comparison with theory. The predictions used PowHEG-EW NLO with NLO QCD and electroweak corrections in the sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^{\ell} input scheme, supplemented by PHOTOS and PYTHIA 8. Baseline generation used NNPDF3.1 NLO. The reported mixing angle was an arithmetic average of the extractions with NNPDF3.1, CT18, and MSHT20, all at NLO [11].

The result was

sin⁡2θeffℓ=0.23147±0.00044stat±0.00005syst±0.00023model.(13)\sin^2\theta_{\mathrm{eff}}^{\ell} = 0.23147 \pm0.00044_{\mathrm{stat}} \pm0.00005_{\mathrm{syst}} \pm0.00023_{\mathrm{model}}. \tag*{(13)}

The model uncertainty included a PDF contribution of 0.00021, smaller than the statistical uncertainty of 0.00044. The precision was therefore mainly limited by statistics, and PDF profiling was not used. At this level of precision, there was less need to use the mass and dilepton rapidity dependence for additional PDF constraints [11].

The uncertainty contributions in Table 1 show the different limitations of these measurements. The Tevatron combination and the LHCb 13 TeV measurement remain mainly limited by statistics. The ATLAS 8 TeV result has comparable statistical and PDF uncertainties, whereas PDFs dominate the CMS 13 TeV uncertainty. To provide an analytic understanding of this limitation, § 4 examines which PDF combinations produce differences between the predicted Drell–Yan asymmetries. § 5 discusses a common interpretation of the LHC measurements and estimates their PDF correlations and the expected precision of a combination.

PDF flavor structure and Drell–Yan asymmetries

As larger Drell–Yan datasets at the LHC reduce the statistical uncertainties, understanding PDF uncertainties becomes increasingly important for more precise weak-mixing-angle determinations. The CMS 13 TeV Run 2 analysis already illustrates this situation. With 138 fb−1^{-1} of data, the PDF uncertainty is 27×10−527 \times10^{-5}, compared with a total uncertainty of 31×10−531 \times10^{-5} [10]. In this section, we therefore examine the flavor structure of modern PDF sets using a simplified leading-order Z-pole calculation. By distinguishing the contributions of quark-flavor production weights and quark-direction dilution, we connect the PDF-set dependence of the predicted asymmetry to the underlying PDF combinations and identify momentum-fraction regions where additional constraints could improve future measurements.

At leading order in pppp collisions, let xhi,lo=(m/s)e±∣y∣x_{\mathrm{hi,lo}} = (m/\sqrt{s})e^{\pm|y|} for a dilepton pair with mass mm and rapidity yy, and write qi=fq(xi,μF2)q_i = f_{q}(x_i,\mu_F^2) and qˉi=fqˉ(xi,μF2)\bar{q}_i = f_{\bar{q}}(x_i,\mu_F^2) at factorization scale μF≃m\mu_F \simeq m. When the dilepton boost is used to estimate the quark direction, the combination qhiqˉloq_{\mathrm{hi}}\bar{q}_{\mathrm{lo}} corresponds to a correct assignment, whereas qˉhiqlo\bar{q}_{\mathrm{hi}}q_{\mathrm{lo}} corresponds to a reversed assignment. Both combinations contribute to the total rate, but the reversed assignment changes the sign of the forward–backward difference and therefore dilutes the asymmetry [31]. Accounting for both contributions, the pure-Z, leading-order asymmetry near the Z pole is

AFBLO(P,s02)=∑qwqA^FBqLA,q(P)∑qwqLS,q(P).(14)A_{\mathrm{FB}}^{\mathrm{LO}}(P,s_0^2)=\frac{\sum_q w_q\widehat{A}_{\mathrm{FB}}^q L_{A,q}(P)}{\sum_q w_q L_{S,q}(P)}. \tag*{(14)}

Here PP denotes a PDF set and the sum runs over q=u,d,s,c,bq=u,d,s,c,b. The luminosities are defined over the chosen ∣y∣|y| interval as

LS,q=∫d∣y∣[qhiqˉlo+qˉhiqlo],LA,q=∫d∣y∣[qhiqˉlo−qˉhiqlo].L_{S,q}=\int d|y|\left[q_{\mathrm{hi}}\bar{q}_{\mathrm{lo}}+\bar{q}_{\mathrm{hi}}q_{\mathrm{lo}}\right],\qquad L_{A,q}=\int d|y|\left[q_{\mathrm{hi}}\bar{q}_{\mathrm{lo}}-\bar{q}_{\mathrm{hi}}q_{\mathrm{lo}}\right].

The flavor weight is wq=vq2+aq2w_q=v_q^2+a_q^2, and the undiluted partonic asymmetry is A^FBq=(3/4)AℓAq\widehat{A}_{\mathrm{FB}}^q=(3/4)A_\ell A_q, with Af=2vfaf/(vf2+af2)A_f=2v_fa_f/(v_f^2+a_f^2). We use the tree-level couplings vf=T3f−2Qfs02v_f=T_3^f-2Q_fs_0^2 and af=T3fa_f=T_3^f, where T3fT_3^f and QfQ_f are the third component of weak isospin and the electric charge in units of the positron charge, and s02s_0^2 is the common reference value of seff2s_{\mathrm{eff}}^2. The discussion here focuses on pppp collisions, but the same analysis can be carried out for ppˉp\bar{p} collisions with appropriately defined LS,qL_{S,q} and LA,qL_{A,q}.

Despite its simplifying assumptions, the LO Z-pole calculation can help explain PDF-set differences in experimental extractions of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. To compare with the unprofiled CMS 13 TeV AFBA_{\mathrm{FB}}-fit results [10], we express the LO asymmetry difference from CT18Z as a mixing-angle shift at fixed measured asymmetry, Δseff2≃−ΔAFBLO/(∂AFBLO/∂seff2)\Delta s_{\mathrm{eff}}^2 \simeq-\Delta A_{\mathrm{FB}}^{\mathrm{LO}}/(\partial A_{\mathrm{FB}}^{\mathrm{LO}}/\partial s_{\mathrm{eff}}^2). The derivative is evaluated with CT18Z at the common input mixing angle s02s_0^2. Figure 4 shows reasonable agreement with the CMS shifts for ∣y∣<3.4|y|<3.4. This calculation omits experimental acceptance and fit-bin weighting. Results for ∣y∣<2.4|y|<2.4 illustrate rapidity-range dependence with unchanged PDF-set ordering.

Figure 4. Comparison of PDF-set differences in sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell estimated from LO ZZ-pole asymmetries (vertical axis) with those from the unprofiled CMS 13 TeV AFBWA_{\mathrm{FB}}^W fit (horizontal axis). The CMS values are digitized from Fig. 11 of Ref. [10]. All differences are defined relative to CT18Z and given in units of 10−510^{-5}. Filled circles and open squares correspond to LO integrations over ∣y∣<3.4|y| < 3.4 and ∣y∣<2.4|y| < 2.4, respectively. The connecting segments show the change between these rapidity ranges. The dashed line denotes equality between the LO and CMS results, and the lower panel shows LO minus CMS. The LO calculation uses m=μF=mZm = \mu_F = m_Z, s=13\sqrt{s} = 13 TeV, and s02=0.23152s_0^2 = 0.23152, with the CT18Z slope evaluated separately for each rapidity range.

The approximate agreement in Figure 4 motivates using the LO calculation to examine the origin of the PDF-set differences. We rewrite Eq. (14) as a weighted mean of the diluted asymmetries for the individual quark flavors,

AFBLO=∑qWqAFBq,dil∑qWq.(15)A_{\mathrm{FB}}^{\mathrm{LO}}=\frac{\sum_q W_q A_{\mathrm{FB}}^{q,\mathrm{dil}}}{\sum_q W_q}. \tag*{(15)}

where Wq=wqLS,qW_q=w_qL_{S,q} is the production weight and AFBq,dil=A^FBqLA,q/LS,qA_{\mathrm{FB}}^{q,\mathrm{dil}}=\widehat{A}_{\mathrm{FB}}^qL_{A,q}/L_{S,q} is the diluted asymmetry for flavor qq. Taking CT18Z as the reference, we evaluate the contribution of each component to the mixing-angle shift Δseff2\Delta s_{\mathrm{eff}}^2 for a given comparison PDF set. For each component, we calculate the shift caused by replacing its CT18Z value with the corresponding value from that set. We average this shift over all possible orders of replacing the ten components (the Shapley value) [32]. Figure 5 shows the contributions in mixing-angle units.

Figure 5. Flavor contributions to mixing-angle shifts Δseff2\Delta s_{\mathrm{eff}}^2 relative to CT18Z, in units of 10−510^{-5}. Contributions from production-weight (WqW_q) and diluted-asymmetry (AFBq,dilA_{\mathrm{FB}}^{q,\mathrm{dil}}) changes are shown separately. The asymmetry shifts ΔAFB\Delta A_{\mathrm{FB}} are converted to mixing-angle shifts Δseff2\Delta s_{\mathrm{eff}}^2 using the CT18Z slope ∂AFB/∂seff2≃−1.35\partial A_{\mathrm{FB}}/\partial s_{\mathrm{eff}}^2 \simeq-1.35 for ∣y∣<3.4|y| < 3.4, as in Fig. 4. Cells show central values and symmetric 68% CL Hessian PDF uncertainties. The Total column gives their sum with its correlated uncertainty. CT18-family strange-asymmetry contributions vanish because s=sˉs = \bar{s} is imposed.

The flavor decomposition in Figure 5 shows two distinct patterns. Within the CT18 family, the positive shifts of CT18 and CT18X relative to CT18Z result primarily from the strange-quark production weight. For CT18, this contribution is +61.6×10−5+61.6 \times10^{-5}, partly offset by other contributions to give a total shift of +41.7×10−5+41.7 \times10^{-5}. In these input CT sets, s=sˉs = \bar{s} and hence LA,s=0L_{A,s} = 0. Strange production adds to the denominator without contributing to the forward–backward numerator. A smaller strange luminosity therefore reduces this dilution and raises the total asymmetry.

This strangeness pattern is consistent with the treatment of the precision ATLAS 7 TeV W/ZW/Z data in the CT18 analysis [22]. These data are included in CT18A and CT18Z, but not in CT18 or CT18X. They favor enhanced strangeness at the electroweak scale, as reflected in the larger strange luminosities of CT18A/Z in the LO calculation.

For the MSHT20 and NNPDF sets shown, the largest shift relative to CT18Z comes from the diluted dd-quark asymmetry and is negative. The diluted uu-quark asymmetry gives a smaller shift in the same direction. Their nonzero strange-quark asymmetries give positive contributions that partly compensate the reduced light-quark asymmetry. Production-weight changes further modify the shifts, but the net LO asymmetries remain below CT18Z.

Even the LO calculation of AFBA_{\mathrm{FB}} is rather involved for a quick comparison of PDF sets, as it requires PDF products, flavor-dependent weights, and rapidity integration. Simple combinations of PDFs would allow the main differences to be examined more directly. To show how the quark–antiquark imbalance enters the asymmetry, we introduce the sum Sq=q+qˉS_q = q + \bar{q} and difference Vq=q−qˉV_q = q - \bar{q}. The luminosities can then be written as

LA,q=12∫d∣y∣[Vq,hiSq,lo−Sq,hiVq,lo],LS,q=12∫d∣y∣[Sq,hiSq,lo−Vq,hiVq,lo].(16)\begin{aligned} L_{A,q} &= \frac{1}{2}\int d|y| \left[V_{q,\mathrm{hi}}S_{q,\mathrm{lo}} - S_{q,\mathrm{hi}}V_{q,\mathrm{lo}}\right],\\ L_{S,q} &= \frac{1}{2}\int d|y| \left[S_{q,\mathrm{hi}}S_{q,\mathrm{lo}} - V_{q,\mathrm{hi}}V_{q,\mathrm{lo}}\right]. \tag*{(16)} \end{aligned}

The strange density s+sˉs + \bar{s} provides a simple way to examine the production-weight differences within the CT18 family. These sets impose s=sˉs = \bar{s}, so Vs=0V_s = 0. Equation (16) then gives LA,s=0L_{A,s} = 0, while LS,sL_{S,s} depends only on the product of s+sˉs + \bar{s} on the two incoming legs. Strange production therefore affects only the denominator and dilutes the asymmetry. Comparing s+sˉs + \bar{s} over the relevant xx range helps identify the PDF differences responsible for this effect.

The summed quark–antiquark difference Q−QˉQ - \bar{Q}, where Q=u+d+s+c+bQ = u + d + s + c + b, provides a simple quantity for examining the PDF differences behind the diluted quark asymmetries. To connect it to AFBA_{\mathrm{FB}}, we first neglect the quark–antiquark difference on the low-xx leg, Vq,lo≃0V_{q,\mathrm{lo}} \simeq0. The differential asymmetry then becomes

AFBLO(y)≃∑qwqq^ qSFBqSq,loVq,hi∑qwqSq,loSq,hi.(17)A_{\mathrm{FB}}^{\mathrm{LO}}(y) \simeq\frac{\sum_q w_q \widehat{q}^{\,q} S_{\mathrm{FB}}^{q} S_{q,\mathrm{lo}} V_{q,\mathrm{hi}}}{\sum_q w_q S_{q,\mathrm{lo}} S_{q,\mathrm{hi}}}. \tag*{(17)}

Further ignoring flavor differences in the low-xx sea and electroweak weights gives the very rough relation AFBLO(y)∝(Q−Qˉ)hi/(Q+Qˉ)hiA_{\mathrm{FB}}^{\mathrm{LO}}(y) \propto(Q - \bar{Q})_{\mathrm{hi}}/(Q + \bar{Q})_{\mathrm{hi}}. This simplification motivates using Q−QˉQ - \bar{Q} as a simple diagnostic of the high-xx PDF differences entering the asymmetry numerator. The low-xx approximation in Eq. (17) improves at forward rapidity and fails at y=0y = 0, where the two terms in LA,qL_{A,q} cancel exactly. For CT18Z, setting Vq,lo=0V_{q,\mathrm{lo}} = 0 in both luminosities overestimates the integrated LO asymmetry by 7.8% for ∣y∣<3.4|y| < 3.4, and by 1.0% for 2<∣y∣<3.42 < |y| < 3.4.

Figure 6 compares Q−QˉQ-\bar{Q} and s+sˉs+\bar{s} over the momentum-fraction range sampled by the LO calculation. In the left panel, the CT18-family sets have similar Q−QˉQ-\bar{Q} distributions on the high-xx branch, whereas MSHT20 and the NNPDF sets give smaller values over much of x∼10−2x\sim10^{-2}–10−110^{-1}. This pattern is consistent with the reduced light-quark asymmetry found in the flavor decomposition. The ordering reverses at the lowest xx, but the quark–antiquark imbalance relative to the total density is much smaller there than on the high-xx leg. This suppresses the low-xx valence contribution to the asymmetry.

Figure 6. PDF combinations used to examine the quark–antiquark imbalance and strange density entering the LO asymmetry. The upper panels show x(Q−Qˉ)x(Q-\bar{Q}) (left), where Q=u+d+s+c+bQ=u+d+s+c+b, and x(s+sˉ)x(s+\bar{s}) (right) at μF=mZ\mu_F=m_Z. Lower panels show ratios to the CT18Z prediction. Curves correspond to the NNLO PDF sets in the legend, with symmetric 68% CL uncertainty bands. The left-hand insets enlarge the peak region near x=0.1x=0.1 and the corresponding ratios. The plotted range covers both momentum-fraction branches for LO Z-pole production at s=13\sqrt{s}=13 TeV and ∣y∣<3.4|y|<3.4, separated by the dotted line at x=mZ/sx=m_Z/\sqrt{s}.

In the right panel, CT18 and CT18X have smaller s+sˉs+\bar{s} than CT18Z on both the low- and high-xx branches. Together with their similar high-xx Q−QˉQ-\bar{Q} distributions, this supports the interpretation that the positive CT18 and CT18X shifts are mainly due to the strange production weight.

These results motivate better constraints on the strange density over both momentum-fraction branches and on the high-xx quark–antiquark differences at x∼10−2x\sim10^{-2}–10−110^{-1}. The measured mass and rapidity dependence of AFBA_{\mathrm{FB}} or A4A_4 can constrain PDFs through Hessian profiling or replica reweighting. These constraints reduce both the PDF uncertainty and the spread of extracted mixing angles, although some differences between PDF sets remain [10].

W charge asymmetries and W/Z cross-section ratios can provide complementary PDF constraints. An LHC simulation study showed that adding the charged-lepton asymmetry in W decays to an AFBA_{\mathrm{FB}} analysis reduces the PDF uncertainty in sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell [20]. In a recent, independent phenomenological reinterpretation [33], the published CMS 13 TeV13\ \mathrm{TeV} A4A_4 measurement was fitted together with the CMS W-decay lepton charge asymmetry [34] and W/Z cross-section ratios [35]. This separate study demonstrated that the additional constraints successfully reduced both the PDF uncertainty and the spread of extracted mixing angles among PDF sets. For CT18Z, this study obtained sin⁡2θeffℓ=0.23154±0.00024\sin^2\theta_{\mathrm{eff}}^\ell= 0.23154 \pm0.00024, compared with the published CMS result of 0.23152±0.000310.23152 \pm0.00031 [10, 33].

Toward a common reinterpretation of LHC measurements

ATLAS, CMS, and LHCb measure the same quantity, sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell, but extract it using different electroweak calculations, PDF sets, and PDF constraints. A difference between the published values can therefore include a shift from these inputs. Testing statistical compatibility requires these shifts and the correlated PDF uncertainties to be evaluated. A common reinterpretation of the angular measurements would provide the basis for this comparison and for a combination. In this section, we discuss the requirements for such a reinterpretation and use a simple estimate to illustrate how PDF correlations limit the gain in precision from combining the current measurements.

The Tevatron combination provides an example of a common reinterpretation. The input central values were expressed using common NNPDF3.0 NNLO PDFs and a ZFITTER-based electroweak prescription. Dedicated template comparisons were used to evaluate the required mixing angle shifts and the uncertainties of the corrections. The corrected mixing angle results were then combined with BLUE, assuming fully correlated PDF uncertainties. The combination was performed at the level of the extracted mixing angles, without a joint fit to the CDF and D0 angular distributions [12].

Combining only the extracted mixing angles and their covariance does not automatically use all the PDF information in the measured angular distributions [36]. The CDF data, for example, can constrain PDF parameters that also affect the D0 mixing angle and its uncertainty. Changing the PDF set or electroweak calculation also requires the central values and uncertainties to be reevaluated. These refits require unfolded asymmetry measurements and their covariance matrices. D0 did not provide tabulated unfolded AFBA_{\mathrm{FB}} values or their covariance [12, 30], while CDF provided fully corrected AFB(m)A_{\mathrm{FB}}(m) values with bin-by-bin errors but not the full covariance used in the fit [29].

A more complete approach is to publish unfolded AFBA_{\mathrm{FB}} or A4A_4 measurements with covariance matrices and use them in a joint fit. The fit uses a common electroweak calculation and PDF set, with shared PDF and theory nuisance parameters. Constraints from each measurement are shared across experiments, and the PDF correlations follow from their responses to the same nuisance parameters. The fit can test the consistency of the data and determine a common sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. The same measurements can be refitted when the PDFs or electroweak calculations are updated.

Early LHC measurements did not publish unfolded angular distributions with full covariance matrices, making a common reinterpretation difficult. CMS and LHCb provide these inputs for their 13 TeV13\ \mathrm{TeV} measurements [10, 11]. For ATLAS, the preliminary 8 TeV8\ \mathrm{TeV} mixing angle result does not provide the distributions and covariance needed for a refit [9]. However, the published unfolded triple-differential Drell–Yan cross sections and their correlated uncertainties can be used to extract the mixing angle [37]. A joint analysis of the CMS 13 TeV13\ \mathrm{TeV}, LHCb 13 TeV13\ \mathrm{TeV}, and ATLAS 8 TeV8\ \mathrm{TeV} datasets is therefore a useful first step toward a common LHC reinterpretation and combination. The omitted earlier measurements are less precise and are expected to have little effect on the combined precision.

A joint fit can retain the observable and lepton definition of each published measurement. For example, the CMS A4(m,∣y∣)A_4(m, |y|) values refer to leptons before QED FSR, while the LHCb AFB(∣Δημμ∣)A_{\mathrm{FB}}(|\Delta\eta_{\mu\mu}|) values use bare muons, without recombining radiated photons, within the fiducial acceptance [10, 11]. Predictions are evaluated in the corresponding bins and phase space, with photon radiation treated according to each definition. The fit uses these predictions and the experimental covariance matrices to determine a common mixing angle and shared PDF nuisance parameters.

To illustrate the possible gain in precision from combining the current LHC measurements, we make a simple estimate of the expected combined uncertainty. The PDF correlations are evaluated using the simplified Z-pole calculation of § 4, including the main lepton transverse-momentum and pseudorapidity cuts for CMS and LHCb at 13 TeV13\ \mathrm{TeV} and ATLAS at 8 TeV8\ \mathrm{TeV} [9, 10, 11]. The acceptance integration assumes massless leptons and zero Z-boson transverse momentum. For each PDF set, the three inclusive fiducial asymmetries are evaluated with the same Hessian error members, and their variations are converted to mixing angle shifts using the corresponding central slopes. The five unprofiled NNLO PDF sets CT18, CT18Z, NNPDF3.1, NNPDF4.0, and MSHT20 give correlations in the range of

ρCMS,ATLAS≃0.835–0.962,ρCMS,LHCb≃0.506–0.642,ρATLAS,LHCb≃0.717–0.768.(18)\rho_{\mathrm{CMS,ATLAS}} \simeq0.835\text{--}0.962, \\ \rho_{\mathrm{CMS,LHCb}} \simeq0.506\text{--}0.642, \\ \rho_{\mathrm{ATLAS,LHCb}} \simeq0.717\text{--}0.768. \tag*{(18)}

The ranges show the spread across the five PDF sets. The stronger CMS–ATLAS correlation reflects their more similar kinematic coverage, while the forward LHCb acceptance gives a more distinct PDF response. We use these LO correlations with the published PDF and total uncertainties and treat the remaining uncertainty as uncorrelated between experiments. A simple estimate using the best linear unbiased estimator (BLUE) [38] then gives an expected combined uncertainty of (26.8–27.5)×10−5(26.8\text{--}27.5) \times10^{-5}, compared with 31×10−531 \times10^{-5} for CMS alone [10]. We emphasize that this BLUE calculation is an illustrative projection to demonstrate the impact of PDF correlations, rather than an official experimental combination. The improvement of 11–14% illustrates the limited gain when the dominant PDF uncertainties are strongly correlated.

The W charge asymmetries and W/Z ratios discussed in § 4 could provide additional PDF constraints and improve the precision beyond this estimate. These measurements can be included together with the unfolded angular distributions in a joint fit. The fit should account for correlations among these data and any overlap with measurements already included in the input PDF fit. The resulting PDF constraints can then be shared consistently across experiments.

Outlook

The full LHC Run 3 dataset at 13.6 TeV and the future HL-LHC dataset will provide larger Drell–Yan samples than those used in the current measurements. These samples will reduce the statistical uncertainty in sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. However, PDF uncertainty already dominates the most precise hadron-collider determination. If this uncertainty remains at its present level, additional statistics will give only a limited improvement in the total precision.

Preserving unfolded AFBA_{\mathrm{FB}} or A4A_4 distributions with their experimental covariance matrices and observable definitions is therefore important for future mixing angle extractions. These distributions can be refitted with updated PDFs or electroweak predictions. They can also be used in joint fits across experiments and with the W charge asymmetries [34] or W/Z ratios [35] discussed in § 4.

Larger data samples are particularly useful for measurements with smaller event yields. One example is a determination of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell from the lepton forward–backward asymmetry in Z + bb production. The relevant Compton subprocesses are gb→Zbgb \to Zb and gbˉ→Zbˉg\bar{b} \to Z\bar{b}, followed by Z→ℓ+ℓ−Z \to\ell^+\ell^- [39, 40]. Their contributions to the Collins–Soper asymmetry are nearly equal when the bottom and antibottom PDFs are nearly equal. A charge-sensitive asymmetry can instead be defined using the recoiling jet in the dilepton rest frame. The forward axis is chosen opposite to the jet for bb events and along the jet for bˉ\bar{b} events, allowing their asymmetries to be combined [39].

Using the jet direction avoids estimating the quark direction from the longitudinal boost, while bottom tagging selects a specific quark flavor. These features can reduce the PDF sensitivity associated with direction dilution and the light-flavor mixture, providing a complementary determination of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell. Residual dependence on the bottom and gluon PDFs and on additional production channels remains. The smaller event yield, together with losses from bottom tagging and charge identification, makes large data samples essential. Charge misidentification dilutes the asymmetry, so reliable bottom-charge determination and calibration of its efficiency and misidentification rate are also required [40].

Larger data samples could also enable a test of the running of the weak mixing angle above the Z pole through high-mass Drell–Yan production [41]. The relevant quantity is the MS‾\overline{\mathrm{MS}} parameter

s^W2(μ)≡sin⁡2θWMS‾(μ)=g^′2(μ)g^2(μ)+g^′2(μ),(19)\hat{s}_W^2(\mu) \equiv\sin^2\theta_W^{\overline{\mathrm{MS}}}(\mu) = \frac{\hat{g}'^2(\mu)}{\hat{g}^2(\mu) + \hat{g}'^2(\mu)}, \tag*{(19)}

where g^\hat{g} and g^′\hat{g}' are the renormalized SU(2)LSU(2)_L and U(1)YU(1)_Y couplings at scale μ\mu. This parameter differs from sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell measured near the Z pole. At the Z scale, the two are related by a finite electroweak form factor, sin⁡2θeffℓ=Re⁡[κ^ℓ(mZ)]s^W2(mZ)\sin^2\theta_{\mathrm{eff}}^\ell= \operatorname{Re}[\hat{\kappa}_\ell(m_Z)]\hat{s}_W^2(m_Z) [3].

The running parameter can be extracted at scales μ∼m\mu\sim m using predictions calculated in the same renormalization scheme. Figure 7 shows projected sensitivities for Run 3 and the HL-LHC from a fit to cross sections binned in dilepton mass, rapidity, and forward or backward lepton direction [41]. The fit uses both rate and asymmetry information to determine a mixing angle in each mass interval. These measurements complement low-energy weak-charge and parity-violating-asymmetry measurements, interpreted in the same scheme with the required radiative corrections [42, 43]. Together, they test the running of the weak mixing angle over a wider range of scales.

Figure 7. Projected tests of the running MS‾\overline{\mathrm{MS}} parameter s^W2(μ)\hat{s}_W^2(\mu) in high-mass Drell–Yan production, from Ref. [41]. The blue curve shows SM running, and the violet point gives the Z-scale experimental reference expressed in the same scheme. The study assumes 13.6 TeV collisions with 300 fb−1^{-1} (Run 3) and 3000 fb−1^{-1} (HL-LHC), and SM running of αEMMS‾\alpha_{\mathrm{EM}}^{\overline{\mathrm{MS}}}. The two scenarios are displaced horizontally for clarity. Outer errors are total projected uncertainties. Inner errors include statistical and experimental contributions only.

Hadron-collider measurements provide determinations of sin⁡2θeffℓ\sin^2\theta_{\mathrm{eff}}^\ell that are experimentally independent of those from e+e−e^+e^- collisions at LEP and SLD. Their precision now makes the PDF dependence of the extraction a central issue. Better constraints on the relevant PDF combinations are needed to improve the precision, while a common treatment of PDFs and electroweak predictions is needed for consistent comparisons between measurements. Preserving unfolded angular distributions allows these comparisons to be updated as the predictions improve. Complementary Z-pole observables and future high-mass measurements can extend the experimental tests of electroweak couplings and their scale dependence.

Acknowledgements

This research was supported by the U.S. Department of Energy under University of Rochester grant number DE-SC0008475, and by National Research Foundation grants RS-2024-00350406 and RS-2008-NR007227 in Korea.

References

  1. [1]W. Grimus, L. Lavoura, O. M. Ogreid, P. Osland, The oblique parameters in multi-Higgs-doublet models, Nucl. Phys. B 801 (2008) 81–96. arXiv:0802.4353, doi:10.1016/j.nuclphysb.2008.04.019.DOI
  2. [2]P. Langacker, The Physics of Heavy Z′ Gauge Bosons, Rev. Mod. Phys. 81 (2009) 1199–1228. arXiv:0801.1345, doi:10.1103/RevModPhys.81.1199.DOI
  3. [3]J. de Blas, D. Dittmaier, R. Kogler, Electroweak model and constraints on new physics, Review of Particle Physics, Particle Data Group, table 10.2, Standard Model prediction from directly measured input parameters (2026). URL https://pdg.lbl.gov/2026/reviews/rpp2026-rev-standard-model.pdf
  4. [4]S. Schael, et al., Precision electroweak measurements on the Z resonance, Phys. Rept. 427 (2006) 257–454. arXiv:hep-ex/0509008, doi:10.1016/j.physrep.2005.12.006.DOI
  5. [5]I. Dubovyk, A. Freitas, J. Gluza, T. Riemann, J. Usovitsch, Electroweak pseudo-observables and Z-boson form factors at two-loop accuracy, JHEP 08 (2019) 113. arXiv:1906.08815, doi:10.1007/JHEP08(2019)113.DOI
  6. [6]G. Aad, et al., Measurement of the forward-backward asymmetry of electron and muon pair-production in pp collisions at √s = 7 TeV with the ATLAS detector, JHEP 09 (2015) 049. arXiv:1503.03709, doi:10.1007/JHEP09(2015)049.DOI
  7. [7]R. Aaij, et al., Measurement of the forward-backward asymmetry in Z/γ* → μ+μ− decays and determination of the effective weak mixing angle, JHEP 11 (2015) 190. arXiv:1509.07645, doi:10.1007/JHEP11(2015)190.DOI
  8. [8]A. M. Sirunyan, et al., Measurement of the effective weak mixing angle using the forward-backward asymmetry of Drell-Yan events in pp collisions at √s = 8 TeV, Eur. Phys. J. C 78 (9) (2018) 701. arXiv:1806.00863, doi:10.1140/epjc/s10052-018-6148-7.DOI
  9. [9]Measurement of the effective leptonic weak mixing angle using electron and muon pairs from Z-boson decay in the ATLAS experiment at √s = 8 TeV, ATLAS-CONF-2018-037 (unpublished) (7 2018).
  10. [10]A. Hayrapetyan, et al., Measurement of the Drell-Yan forward-backward asymmetry and of the effective leptonic weak mixing angle in proton-proton collisions at √s = 13 TeV, Phys. Lett. B 866 (2025) 139526. arXiv:2408.07622, doi:10.1016/j.physletb.2025.139526.DOI
  11. [11]R. Aaij, et al., Measurement of the effective leptonic weak mixing angle, JHEP 12 (2024) 026. arXiv:2410.02502, doi:10.1007/JHEP12(2024)026.DOI
  12. [12]T. Aaltonen, et al., Tevatron combination of effective leptonic weak mixing angle measurements, Phys. Rev. D 97 (11) (2018) 112007. arXiv:1801.06283, doi:10.1103/PhysRevD.97.112007.
  13. [13]J. C. Collins, D. E. Soper, Angular Distribution of Dileptons in High-Energy Hadron Collisions, Phys. Rev. D 16 (1977) 2219. doi:10.1103/PhysRevD.16.2219.DOI
  14. [14]A. Bodek, A simple event weighting technique for optimizing the measurement of the forward-backward asymmetry of Drell-Yan dilepton pairs at hadron colliders, Eur. Phys. J. C 67 (2010) 321–334. arXiv:0911.2850, doi:10.1140/epjc/s10052-010-1287-5.DOI
  15. [15]M. Chiesa, F. Piccinini, A. Vicini, Direct determination of sin² θ_eff^l at hadron colliders, Phys. Rev. D 100 (7) (2019) 071302. arXiv:1906.11569, doi:10.1103/PhysRevD.100.071302.
  16. [16]A. B. Arbuzov, M. Awramik, M. Czakon, A. Freitas, M. W. Grunewald, K. Monig, S. Riemann, T. Riemann, ZFITTER: A semi-analytical program for fermion pair production in e+e− annihilation, from version 6.21 to version 6.42, Comput. Phys. Commun. 174 (2006) 728–758. arXiv:hep-ph/0507146, doi:10.1016/j.cpc.2005.12.009.DOI
  17. [17]E. Richter-Was, Z. Was, Documentation of TauSpinner approach for electroweak corrections in LHC Z → ll observables, Eur. Phys. J. C 79 (6) (2019) 480. arXiv:1808.08616, doi:10.1140/epjc/s10052-019-6987-x.DOI
  18. [18]L. Barze, G. Montagna, P. Nason, O. Nicrosini, F. Piccinini, A. Vicini, Neutral current Drell-Yan with combined QCD and electroweak corrections in the POWHEG BOX, Eur. Phys. J. C 73 (6) (2013) 2474. arXiv:1302.4606, doi:10.1140/epjc/s10052-013-2474-y.DOI
  19. [19]M. Chiesa, C. L. Del Pio, F. Piccinini, On electroweak corrections to neutral current Drell–Yan with the POWHEG BOX, Eur. Phys. J. C 84 (2024) 539. arXiv:2402.14659, doi:10.1140/epjc/s10052-024-12908-1.DOI
  20. [20]A. Bodek, J. Han, A. Khukhunaishvili, W. Sakumoto, Using Drell-Yan forward-backward asymmetry to reduce PDF uncertainties in the measurement of electroweak parameters, Eur. Phys. J. C 76 (3) (2016) 115. arXiv:1507.02470, doi:10.1140/epjc/s10052-016-3958-3.DOI
  21. [21]R. D. Ball, et al., The path to proton structure at 1% accuracy, Eur. Phys. J. C 82 (5) (2022) 428. arXiv:2109.02653, doi:10.1140/epjc/s10052-022-10328-7.DOI
  22. [22]T.-J. Hou, et al., New CTEQ global analysis of quantum chromodynamics with high-precision data from the LHC, Phys. Rev. D 103 (1) (2021) 014013. arXiv:1912.10053, doi:10.1103/PhysRevD.103.014013.DOI
  23. [23]S. Bailey, T. Cridge, L. A. Harland-Lang, A. D. Martin, R. S. Thorne, Parton distributions from LHC, HERA, Tevatron and fixed target data: MSHT20 PDFs, Eur. Phys. J. C 81 (4) (2021) 341. arXiv:2012.04684, doi:10.1140/epjc/s10052-021-09057-0.DOI
  24. [24]W. T. Giele, S. Keller, Implications of hadron collider observables on parton distribution function uncertainties, Phys. Rev. D 58 (1998) 094023. arXiv:hep-ph/9803393, doi:10.1103/PhysRevD.58.094023.DOI
  25. [25]N. Sato, J. F. Owens, H. Prosper, Bayesian reweighting for global fits, Phys. Rev. D 89 (2014) 114020. arXiv:1310.1089, doi:10.1103/PhysRevD.89.114020.DOI
  26. [26]H. Hou, J. X. P. Zurita, PDF reweighting in the hessian matrix approach, JHEP 12 (2014) 100. arXiv:1402.6623, doi:10.1007/JHEP12(2014)100.DOI
  27. [27]ATLAS Collaboration, Measurement of τ polarisation in Z/γ* → ττ decays in proton–proton collisions at √s = 8 TeV with the ATLAS detector, Eur. Phys. J. C 78 (2018) 163. arXiv:1709.03490, doi:10.1140/epjc/s10052-018-5619-1.DOI
  28. [28]CMS Collaboration, Measurement of the τ lepton polarization in Z boson decays in proton-proton collisions at √s = 13 TeV, JHEP 01 (2024) 101. arXiv:2302.12408, doi:10.1007/JHEP01(2024)101.
  29. [29]T. A. Aaltonen, et al., Measurement of sin² θ_eff^lept using e+e− pairs from γ*/Z bosons produced in p p̄ collisions at a center-of-momentum energy of 1.96 TeV, Phys. Rev. D 93 (11) (2016) 112016, [Addendum: Phys.Rev.D 95, 119901 (2017)]. arXiv:1605.02719, doi:10.1103/PhysRevD.93.112016.DOI
  30. [30]V. M. Abazov, et al., Measurement of the Effective Weak Mixing Angle in p p̄ → Z/γ* → ℓ+ℓ− Events, Phys. Rev. Lett. 120 (24) (2018) 241802. arXiv:1710.03951, doi:10.1103/PhysRevLett.120.241802.DOI
  31. [31]R. D. Ball, A. Candido, S. Forte, F. Hekhorn, E. R. Nocera, J. Rojo, C. Schwan, Parton distributions and new physics searches: the Drell–Yan forward–backward asymmetry as a case study, Eur. Phys. J. C 82 (2022) 1160. doi:10.1140/epjc/s10052-022-11133-y.DOI
  32. [32]L. S. Shapley, A value for n-person games, in: H. W. Kuhn, A. W. Tucker (Eds.), Contributions to the Theory of Games, Volume II, Princeton University Press, Princeton, 1953, pp. 307–318. doi:10.1515/9781400881970-018.
  33. [33]A. Bodek, H.-S. Seo, U. K. Yang, Precision Measurements of the Electroweak Mixing Angle in the Region of the Z pole. arXiv:2508.18022.arxiv.org/abs/2508.18022
  34. [34]A. M. Sirunyan, et al., Measurements of the W boson rapidity, helicity, double-differential cross sections, and charge asymmetry in pp collisions at √s = 13 TeV, Phys. Rev. D 102 (9) (2020) 092012. arXiv:2008.04174, doi:10.1103/PhysRevD.102.092012.DOI
  35. [35]A. Hayrapetyan, et al., Measurement of the inclusive cross sections for W and Z boson production in proton-proton collisions at √s = 5.02 and 13 TeV, JHEP 04 (2025) 162. arXiv:2408.03744, doi:10.1007/JHEP04(2025)162.DOI
  36. [36]T. Dado, M. Owen, M. Pinamonti, Combinations of measurements that are simultaneous fits of parameters of interest and systematic uncertainties using the BLUE method (2026). arXiv:2609.11461.arxiv.org/abs/2609.11461
  37. [37]M. Aboud, et al., Measurement of the Drell–Yan triple-differential cross section in pp collisions at √s = 8 TeV, JHEP 12 (2017) 059. arXiv:1710.05167, doi:10.1007/JHEP12(2017)059.DOI
  38. [38]L. Lyons, D. Gibaut, P. Clifford, How to combine correlated estimates of a single physical quantity, Nucl. Instrum. Meth. A 270 (1988) 110–117. doi:10.1016/0168-9002(88)90018-6.DOI
  39. [39]F. del Aguila, L. Ametller, P. Talavera, Forward backward asymmetries of lepton pairs in events with a large transverse momentum jet at hadron colliders, Phys. Rev. Lett. 89 (2002) 161802. arXiv:hep-ph/0206160, doi:10.1103/PhysRevLett.89.161802.DOI
  40. [40]F. del Aguila, L. Ametller, R. Pittau, NLO forward-backward charge asymmetries in pp (pp̄) → ℓ⁻ℓ⁺ production at large hadron colliders, Phys. Lett. B 628 (2005) 40–48. arXiv:hep-ph/0507262, doi:10.1016/j.physletb.2005.09.034.DOI
  41. [41]S. Amoroso, M. Chiesa, C. L. Del Pio, K. Lipka, F. Piccinini, F. Vazzoler, A. Vicini, Probing the weak mixing angle at high energies at the LHC and HL-LHC, Phys. Lett. B 844 (2023) 138103. arXiv:2302.10782, doi:10.1016/j.physletb.2023.138103.DOI
  42. [42]K. S. Kumar, S. Mantry, W. J. Marciano, P. A. Souder, Low Energy Measurements of the Weak Mixing Angle, Ann. Rev. Nucl. Part. Sci. 63 (2013) 237–267. arXiv:1302.6263, doi:10.1146/annurev-nucl-102211-175056.DOI
  43. [43]J. Erler, K. S. Kumar, Low Energy Measurements of the Weak Mixing Angle, to be published in Physics Reports, Collection of papers on Electroweak Precision Measurements, edited by Costas Vellidis and others. (2025).

Paper details

vlnc:W_jd9ctn8n
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents