The hadronic contribution to the running of 𝜶 and the

electroweak mixing angle

Alessandro Conigli,𝑎,𝑏,∗ Georg von Hippel,𝑐 Simon Kuberski,𝑑 Harvey B.

Meyer,𝑎,𝑏,𝑐 Konstantin Ottnad𝑐 and Hartmut Wittig𝑎,𝑏,𝑐
arXiv:2502.10159v1 [hep-lat] 14 Feb 2025

𝑎 Helmholtz Institute Mainz, Johannes Gutenberg-Universität Mainz, 55099 Mainz, Germany

𝑏 GSI Helmholtz Centre for Heavy Ion Research, 64291 Darmstadt, Germany
𝑐 PRISMA+ Cluster of Excellence and Institut für Kernphysik, Johannes Gutenberg-Universität Mainz,

55099 Mainz, Germany

𝑑 Theoretical Physics Department, CERN, 1211 Geneva 23, Switzerland

E-mail: [email protected]

We report on our update to [1] on the hadronic running of electroweak couplings from 𝑂 (𝑎)-
improved Wilson fermions with 𝑁 𝑓 = 2 + 1 flavours. The inclusion of additional ensembles at
very fine lattice spacings together with a number of techniques to split the different contributions
for a better control of cutoff effects allows us to substantially improve the precision. We employ
two different discretizations of the vector current to compute the subtracted Hadronic Vacuum
Polarization (HVP) functions 𝛱¯ 𝛾𝛾 and 𝛱¯ 𝑍 𝛾 for Euclidean time momenta up to 𝑄 2 ≤ 9 GeV2 .
To reduce cutoff effects in the short distance region we apply a suitable subtraction to the TMR
kernel function, which cancels the leading 𝑥04 behaviour. The subtracted term is then computed
in perturbative QCD using the Adler function and added back to compensate for the subtraction.
Chiral-continuum extrapolations are performed with five values of the lattice spacing and several
pion masses, including its physical value, and several fit ansätze are explored to estimate the
systematics arising from model selection. Our results show excellent prospects for high-precision
(5)
estimates of Δ𝛼had (𝑀𝑍2 ) at the Z-pole.

MITP-25-007
CERN-TH-2025-022

The 41st International Symposium on Lattice Field Theory (LATTICE2024)

28 July - 3 August 2024
Liverpool, UK

∗ Speaker

© Copyright owned by the author(s) under the terms of the Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0). https://pos.sissa.it/
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

1. Introduction

Precision observables play a crucial role in the search for new physics beyond the Standard
Model (SM). Complementary to direct searches conducted at high-energy colliders, theoretical
predictions allow us to put constraints on the SM by demanding that both theory and experimental
measurements reach high precision. This can be especially challenging when the observable in
question is influenced by significant hadronic effects. A notable example is the muon’s anomalous
magnetic moment 𝑎 𝜇 , where sub-percent precision was achieved both at the experimental [2, 3] and
theoretical level [4–8]. In this context, Lattice QCD has emerged as one of the leading methods,
replacing the data-driven approach with a first-principles calculation.
Here we examine two closely related quantities that play a crucial role in SM tests: the energy
dependence of the electromagnetic coupling 𝛼 and the electroweak mixing angle sin2 𝜃 𝑊 . The
former relates the square of the electric charge in the Thomson limit to its value at the Z-pole,
which is an important input quantity in electroweak precision tests. Moreover, the running of the
electroweak mixing angle is a sensitive probe of beyond-SM physics, particularly at low energies
[9]. As in the case of 𝑎 𝜇 , the overall precision of both the running of 𝛼 and 𝑠𝑖𝑛2 𝜃 𝑊 is limited by
hadronic uncertainties.

2. Methodology

2.1 Lattice setup

We perform our calculation on a set of 26 𝑁 𝑓 = 2 + 1 ensembles generated by the Coordinated
Lattice Simulations (CLS) [10–14] with non-perturbatively 𝑂 (𝑎)-improved Wilson fermions and a
tree-level improved Lüscher-Weisz gauge action. The gauge ensembles employed in this work cover
five values of the lattice spacing in the range 0.0039 fm < 𝑎 < 0.087 fm while the pion masses lie
in the range 130 MeV ≤ 𝑚 𝜋 ≤ 420 MeV. We work with a subset of the CLS ensembles where the
sum of the bare quark masses is kept constant when approaching the physical point, such that the
bare coupling 𝑔˜ 0 is also held constant along the chiral trajectory. To account for a small mistuning
phys phys
in 𝑚 𝐾 when hitting 𝑚 𝜋 we include four additional ensembles on a different chiral trajectory
where the strange quark mass is fixed to its physical value.
To reduce cutoff effects, we perform a full 𝑂 (𝑎) improvement of the observables with two
independent sets of non-perturbatively determined improvement coefficients: by set 1 we denote
the results from large-volume simulations as computed in [15], while we refer to set 2 when using
𝑍𝑉 and 𝑐 𝑉 from [16] and 𝑏 𝑉 , 𝑏¯ 𝑉 [17], determined in the Schrödinger Functional (SF) setup. In
addition, to further constrain the chiral-continuum extrapolations we employ two discretizations of
the vector current, the local (L) and the point-split conserved (C) currents as defined in [18].

2.2 Time Momentum Representation

The running of the electromagnetic coupling and the weak mixing angle at any momentum
transfer 𝑄 2 can be expressed in terms of the effective couplings as
𝛼

𝛼(𝑄 2 ) = , sin2 𝜃 𝑊 (𝑄 2 ) = sin2 𝜃 𝑊 1 + Δ sin2 𝜃 𝑊 (𝑄 2 ) , (1)
1 − Δ𝛼(𝑄 2 )

2
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

where 𝛼 and sin2 𝜃 𝑊 represent their values in the Thomson limit 𝑄 2 = 0. The running Δ𝛼(𝑄 2 ) and
Δ sin2 𝜃 𝑊 (𝑄 2 ) receive contributions from both the hadronic and leptonic sectors. While the latter
can be computed reliably in perturbation theory, the contribution arising from low-energy quarks is
non-perturbative and at leading order it can be expressed in terms of the subtracted HVP functions
¯ (𝛾,𝛾) and 𝛱
𝛱 ¯ (𝑍 ,𝛾) as

¯ (𝛾,𝛾) (𝑄 2 ), 4𝜋𝛼 ¯ (𝑍 ,𝛾) (𝑄 2 ),

Δ𝛼had (𝑄 2 ) = 4𝜋𝛼𝛱 (Δ sin2 𝜃 𝑊 )had (𝑄 2 ) = − 𝛱 (2)
sin2 𝜃 𝑊

where 𝛱¯ (𝑄 2 ) = 𝛱 (𝑄 2 ) − 𝛱 (0). The terms 𝛱 ¯ (𝛾,𝛾) and 𝛱 ¯ (𝑍 ,𝛾) can be computed on the lattice for
any space-like momentum transfer through the Time-Momentum Representation (TMR) [19, 20],
given by the integral over Euclidean time
∫ ∞
( 𝛼,𝛾)
¯
𝛱 2
(𝑄 ) = d𝑥 0 𝐺 ( 𝛼,𝛾) (𝑡)𝐾 (𝑥 0 , 𝑄 2 ), 𝛼 = 𝑍, 𝛾, (3)
0

where 𝐺 ( 𝛼,𝛾) denotes the zero-momentum-projected vector correlator, while 𝐾 (𝑥 0 , 𝑄 2 ) is a 𝑄 2 -

dependent kernel function. Following the notation of [1, 15] and working in the 𝑆𝑈 (3)-flavor basis,
the vector correlators of interest read
1 4 4 (𝑐,𝑐) 2 (𝑐,8)
𝐺 (𝛾,𝛾) = 𝐺 (3,3) + 𝐺 (8,8) + 𝐺 (𝑐,𝑐) + 𝐺 disc + √ 𝐺 disc +..., (4)
3 9 9 3 3
 
(𝑍 ,𝛾) 1 1 1 1 (𝑐,𝑐)
𝐺 = − sin 𝜃 𝑊 𝐺 (𝛾,𝛾) − √ 𝐺 (0,8) − 𝐺 (𝑐,𝑐) − 𝐺 disc
2
+..., (5)
2 6 3 18 18

in terms of the building blocks as defined in [1], which can be computed separately.

2.3 Computational strategy

(5)
To precisely determine the value of Δ𝛼had (𝑀𝑍2 ) at the Z-pole it is essential to reach high values
of the momentum transfer 𝑄 2 on the lattice. This ensures that the threshold energy above which the
perturbative running becomes relevant is effectively shifted to higher values, therefore reducing the
uncertainty from perturbation theory. However, achieving such high 𝑄 2 values while maintaining
control over the systematics involved in the computation presents significant challenges. To address
this, we propose the following decomposition for the subtracted HVP

¯ (𝑄 2 ) = 𝛱 (𝑄 2 ) − 𝛱 (𝑄 2 /4) + 𝛱 (𝑄 2 /4) − 𝛱 (0) .

   
𝛱 (6)

This facilitates a clear separation of contributions arising from different Euclidean distance scales,
thereby improving the control over the chiral-continuum extrapolations associated with each term.
In the analysis presented here, we focus exclusively on the first term on the right hand side of Eq. (6),
corresponding to the high energy observable 𝛱 b (𝑄 2 ) = 𝛱 (𝑄 2 ) − 𝛱 (𝑄 2 /4).

Isovector contribution: following the approach detailed in [21], we extract the isovector contri-
b (3,3) (𝑄 2 ) using the decomposition
bution 𝛱

𝛱 b (3,3) (𝑄 2 ) + 𝑏 (3,3) (𝑄 2 , 𝑄 2𝑚 ),
b (3,3) (𝑄 2 ) = 𝛱 (7)
sub

3
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

0.0030
Q2m = ∞ G(3,3)
1.4
Q2m = 3 GeV 2 10 · (G(3,3) − G(8,8) )
K(t, Q2, Q2m)/t3 [fm−1]
0.0025
1.2 4 (c,c)
Q2m = 5 GeV2

G(d,e)(t) · K(t, Q2)

9 Gconn
Q2m = 7 GeV2 0.0020 10 · 13 G(0,8)
1.0
Q2m = 9 GeV2 (c,c)
1000 · 49 Gdisc
0.8 0.0015 2 (c,8)
1000 · √ G
3 3 disc

0.6 0.0010

0.4
0.0005
0.2
0.0000
0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0
t [fm] t [fm]

Figure 1: Left: Illustration of the subtracted kernel in Eq. (8) for different values of the virtualities 𝑄 𝑚 .
Right: integrands of the various contributions according to Eq. (12) for 𝑄 2 = 5 GeV2 . Results are shown for
physical point ensemble E250 with 𝑎 ≈ 0.064 fm.

b (3,3) (𝑄 2 ) is computed from the TMR integral Eq. (3) with the kernel function defined as
where 𝛱 sub
   
2 2 16 4 𝑄𝑥 0 𝑄2 4 𝑄 𝑚 𝑥0
𝐾sub (𝑥 0 , 𝑄 , 𝑄 𝑚 ) = 2 sin − 4 sin . (8)
𝑄 4 𝑄𝑚 2
The second term,
 
(3,3) 2 𝑄2 (3,3) (3,3)
𝑏 (𝑄 , 𝑄 2𝑚 ) = 𝛱 2
(4𝑄 𝑚 ) − 𝛱 2
(𝑄 𝑚 ) , (9)
4𝑄 2𝑚
can be computed reliably using the massless perturbative Adler function. The subtracted kernel
𝐾sub (𝑥 0 , 𝑄 2 , 𝑄 2𝑚 ) is designed to cancel the 𝑥 04 behaviour in the TMR integral. The neat effect is a
reduction of higher-order cutoff effects, including the potentially dangerous contributions such as
𝑂 (𝑎 2 log(𝑎)) in the very short Euclidean distance regime [22, 23]. As illustrated on the left-hand
side of Fig. 1, the subtracted kernel function is shown for various choices of the virtualities 𝑄 𝑚 .
Throughout our analysis, we adopt 𝑄 𝑚 = 3 GeV as the default value unless otherwise specified.
Further reduction of lattice artefacts originating from the Short-Distance (SD) region is achieved
by incorporating tree-level perturbative corrections. Specifically, cutoff effects are computed at
tree-level in the massless theory, and the non-perturbatively evaluated observables are replaced
with
O tl (0)
O (𝑎) → O (𝑎) tl , (10)
O (𝑎)
where O tl (𝑎) represents the tree-level evaluation of the observable. At our coarsest lattice spacing,
this approach leads to a significant reduction of cutoff effects, from approximately 20% to 7%.
b (3,3) , the isoscalar contribution can be determined
Isoscalar contribution: after computing 𝛱
through the decomposition [21]
b (8,8) (𝑄 2 ) = 𝛱
𝛱 b (3,3) (𝑄 2 ) + Δ𝑙𝑠 (𝑄 2 ), (11)

where only the term Δ𝑙𝑠 has to be computed non-perturbatively. Notably, Δ𝑙𝑠 is parametrically
suppressed at short distances, therefore no assistance from perturbation theory is required to compute

4
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

this observable. In practice, we evaluate Δ𝑙𝑠 from

∫ ∞  
16 4 𝑄𝑥 0
d𝑥 0 𝐺 (8,8) − 𝐺 (3,3) 𝐾 (𝑥 0 , 𝑄 2 ),
2
  2
Δ𝑙𝑠 (𝑄 ) = 𝐾 (𝑥 0 , 𝑄 ) = 2 sin , (12)
0 𝑄 4

using the non-subtracted kernel function 𝐾 (𝑥 0 , 𝑄 2 ). A similar strategy is employed to compute

b (0,8) contribution, which is relevant exclusively to the electroweak mixing angle. This term
the 𝛱
vanishes linearly towards the 𝑆𝑈 (3)-symmetric point, ensuring that no additional perturbative
calculations are required. Specifically, we fit the only available discretization (conserved-local) that
avoids reliance on the renormalised singlet local current.

Charm connected contribution: similar to the isovector case, the charm connected contribution
can be extracted using the subtracted kernel Eq. (8), employing the following decomposition

𝛱 b (𝑐,𝑐) (𝑄 2 ) + 𝑏 conn
b (𝑐,𝑐) (𝑄 2 ) = 𝛱 (𝑐,𝑐)
(𝑄 2 , 𝑄 2𝑚 ). (13)
sub

(𝑐,𝑐)
The subtraction function 𝑏 conn , defined analogously to 𝑏 (3,3) in Eq. (9), is computed according to
(𝑐,𝑐)
𝑏 conn (𝑄 2 , 𝑄 2𝑚 ) = 2𝑏 (3,3) (𝑄 2 , 𝑄 2𝑚 ) + Δ𝑙𝑐 𝑏(𝑄 2 , 𝑄 2𝑚 ), (14)

where the first term, 2𝑏 (3,3) (𝑄 2 , 𝑄 2𝑚 ), is derived from massless perturbation theory. The second
term, Δ𝑙𝑐 𝑏, represents the difference between the isovector and charm-connected contributions,
which we evaluate non-perturbatively. In the right panel of Fig. 1 we illustrate the behaviour and
the relative size of the various contributions to the HVP.

3. Chiral-continuum extrapolations

Having computed all the relevant contributions across five distinct lattice spacings and multiple
quark masses, we proceed with a reliable extrapolation to the continuum and physical point. The
latter is defined within the isospin-symmetric limit by fixing 𝑚 𝜋 = (𝑚 𝜋 0 )phys and 2𝑚 2𝐾 − 𝑚 2𝜋 =
(𝑚 2𝐾 + + 𝑚 2𝐾 0 − 𝑚 2𝜋 + )phys [24, 25], which yields the physical masses 𝑚 𝜋 = 134.9768(5) MeV and
𝑚 𝐾 = 495.011(10) MeV for the pion and kaon, respectively. In practice, to describe the chiral
dependence we employ the dimensionless hadronic combinations
 
2 2 1 2
Φ2 = 8𝑡0 𝑚 𝜋 , Φ4 = 8𝑡0 𝑚 𝐾 + 𝑚 𝜋 , (15)
2
phys
where the gradient flow scale 𝑡0 /𝑎 2 [26] is used for scale setting. Its physical value, 𝑡0 =
0
0.1449(7) fm, is adopted from the Ξ -baryon mass determination in [27].
We parametrise the lattice spacing dependence guided by Symanzik effective theory. Given the
relative short-distance nature of the observables and their very precise determination, higher-order
cutoff effects must be incorporated into our fits. Our most general fit ansatz reads
     
phys phys phys
O (𝑋𝑎 ) = 𝛽2 𝑋𝑎2 +𝛽3 𝑋𝑎3 +𝛽4 𝑋𝑎4 +𝛿2 𝑋𝑎2 Φ2 − Φ2 +𝛿3 𝑋𝑎3 Φ2 − Φ2 +𝜖 2 𝑋𝑎2 Φ4 − Φ4 , (16)

where 𝑋𝑎2 = 𝑎 2 /(8𝑡0 ). In practice, it is challenging to constrain all fit parameters simultaneously,
therefore we explore various functional forms by selectively dropping one or more of the terms

5
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

Set 2 LC
0.0105
LL, set 1 LC, set 1
0.0106
LL, set 2 LC, set 2 0.0104
0.0105
0.0103

0.0104 0.0102
sub

sub
(∆α3,3)SD

(∆α3,3)SD
0.0103 0.0101

0.0102 0.0100

0.0099
0.0101 β = 3.4 β = 3.7
0.0098 β = 3.46 β = 3.85
0.0100 β = 3.55
0.0097
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
a2 [fm] φ2

Figure 2: Illustration of fits to the isovector contribution in the Short Distance (SD) region. Left: continuum
limit behaviours for fours sets of data based on different improvement schemes and discretisations of the
vector current. Each line corresponds to a single fit, with the opacity associated to the weights as given by
our model average prescription. Right: chiral approach to the physical pion mass for one of the fits with
phys
the highest weight. Data points are projected to Φ4 . Coloured lines denote the chiral trajectories at finite
lattice spacing, while the grey band shows the dependence on Φ2 in the continuum. Results are shown for
𝑄 2 = 𝑄 2𝑚 = 9 GeV2 .

multiplied by the parameters 𝛽𝑖 , 𝛿𝑖 , 𝜖2 . To model the chiral dependence of observable O, we include

a linear term in Φ2 along with a higher-order correction,
   

phys
phys phys
O Φ2 = O Φ2 + 𝛾 1 Φ2 − Φ2 + 𝛾2 𝑓 𝜒 (Φ2 ) − 𝑓 𝜒 (Φ2 ) , (17)

where 𝑓 𝜒 ∈ {Φ2 log(Φ2 ); Φ22 }. The strange quark mass dependence, governed by Φ4 , is included
as a linear term Since Φ4 is close to its physical value on all ensembles, higher-order corrections
are unnecessary. To assess systematic uncertainties arising from the chiral-continuum extrapola-
tions, we apply cuts to the data sets, either excluding the coarsest lattice spacing or by removing
all ensembles with 𝑚 𝜋 > 400 MeV. For the final estimate and systematic error analysis, we per-
form a weighted model average following [28], with weights assigned according to the Takeuchi
Information Criterion (TIC) [29].
On the left-hand side of Fig. 2, we present the continuum limit approach of the isovector
contribution at physical quark masses for all the explored models. The right-hand side illustrates
the chiral dependence based on the best fit results obtained through our model average prescription.
The isoscalar contribution is computed by evaluating Δ𝑙𝑠 (Δ𝛼), as defined in Eq. (11). This
quantity vanishes at the 𝑆𝑈 (3)-symmetric point and at leading order scales proportionally to
𝑚 𝑠 − 𝑚 𝑙 . To capture this behaviour, we model the chiral-continuum dependence using the following
parametrisation
 
Δ𝑙𝑠 (Δ𝛼) (Φ 𝛿 , Φ4 , 𝑋𝑎 ) = Φ 𝛿 𝛾1 + 𝛾2 Φ 𝛿 + 𝛽2 𝑋𝑎2 + 𝛽3 𝑋𝑎3 + 𝛾0 Φ4 , (18)

where Φ 𝛿 = Φ4 − 23 Φ2 . This formulation ensures that all cutoff effects are suppressed by Φ 𝛿 in the
proximity of the 𝑆𝑈 (3)-symmetric point. An illustration of the continuum limit behaviour and the
b (0,8) contribution vanishes linearly as
chiral-dependence for Δ𝑙𝑠 is shown in Fig. 3. Similarly, the 𝛱
the 𝑆𝑈 (3)-symmetric point is approached. We therefore describe the chiral-continuum dependence
using a fit ansatz similar to Eq. (18).

6
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

Set 2 LC
0.0011 LL, set 1 LC, set 1 β = 3.4 β = 3.7
LL, set 2 LC, set 2 0.0010 β = 3.46 β = 3.85
0.0010 β = 3.55
0.0009 0.0008
−∆ls(∆α)

−∆ls(∆α)
0.0008 0.0006

0.0007
0.0004
0.0006
0.0002
0.0005

0.0004 0.0000

0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
a2 [fm] φ2

Figure 3: Illustration of fits to the Δ𝑙𝑠 contribution. Left: continuum limit behaviours for four sets of data
based on different improvement schemes and discretisations of the vector current. Right: chiral approach to
the physical point for one of the best fits for a reference set. Results are shown for 𝑄 2 = 9 GeV2 .

Finally, for the charm connected contribution we evaluate non-perturbatively both 𝛱b (𝑐,𝑐) and
sub
Δ𝑙𝑐 𝑏 as introduced in Eq. (13). The charm quark mass tuning to match the physical 𝐷 𝑠 meson mass
of 1968.47 MeV and the evaluation of the renormalization constant for the local vector current are
described in [18, 30]. The extrapolation to the physical point follows the approach used for the
isovector contribution, however we exclude higher-order corrections in Φ2 , as we expect a mild
dependence on the pion mass. For the final value we exclude the local-local current discretization
from the model average due to its significantly larger cutoff effects.

4. Running with energy

In our study, we sampled the HVP function at multiple values of 𝑄 2 within the range 0.1 GeV2 ≤
𝑄 2 ≤ 9 GeV2 . To provide a smooth and analytic representation of the HVP momentum dependence
we employ a Padé approximant, whose general expression reads [1, 31]
Í𝑀 2𝑗
2 𝑁 2 𝑗=0 𝑎 𝑗 𝑄
𝛱 (𝑄 ) ≈ 𝑅 𝑀 (𝑄 ) =
b Í𝑁 . (19)
1 + 𝑘=1 𝑏 𝑘 𝑄 2𝑘
Here, numerator and denominator are polynomials of degree 𝑀 and 𝑁, respectively. Since the
subtracted HVP function vanishes at 𝑄 2 = 0, we impose the constraint 𝑎 0 = 0 during the fitting
process to ensure consistency. We find that polynomials of degree 𝑀 = 2 and 𝑁 = 3 accurately
describe the data, while higher order coefficients are poorly determined. The results of this analysis
for the quantities Δ𝛼had and Δhad sin2 𝜃 𝑊 in the high-energy regime, specifically 𝑄 2 − 𝑄 2 /4, are
illustrated in Fig. 4.

5. Conclusion

We have presented preliminary results for the leading hadronic contribution to the running
of the electromagnetic coupling and the electroweak mixing angle in the high-energy regime by
computing the 𝛱b (𝑄 2 ) = 𝛱 (𝑄 2 ) − 𝛱 (𝑄 2 /4) HVP across the momentum range 𝑄 2 ≤ 9 GeV2 . This
decomposition allows us to perform a clear separation of contributions from different Euclidean

7
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

0.0030 I=1

∆ sin2 θW (−Q2 ) − ∆ sin2 θW (−Q2 /4)

0.0000
∆α(−Q2) − ∆α(−Q2/4) I=0
0.0025 charm
total −0.0005
0.0020
−0.0010
0.0015
−0.0015
0.0010 I=1
I=0
−0.0020
0.0005 charm
08
−0.0025 total
0.0000
0 2 4 6 8 0 2 4 6 8
Q2 [GeV2] Q2 [GeV2]

Figure 4: HVP contribution computed in the region 𝑄 2 − 𝑄 2 /4 to the running of 𝛼 (left) and sin2 𝜃 𝑊 (right)
as a function of 𝑄 2 . Different colours represent the isovector (I = 1), isoscalar (I = 0), charm and, for
sin2 𝜃 𝑊 , the mixed 𝑍𝛾 contributions. The dashed line on the left panel denotes the tree-level value for the
isovector contribution.

regions, thus giving access to higher values of the space-like momenta compared to our 2022 study
[1]. This method also provides improved control over the systematic uncertainties associated with
the extrapolation to the physical point. Looking ahead, we plan to compute the missing low-energy
component, 𝛱 (𝑄 2 /4) − 𝛱 (0), within the same momentum range. Combining our lattice results
(5)
with perturbative QCD, we aim to obtain a high-precision estimate of Δ𝛼had (𝑀𝑍2 ). Together with
additional statistics and improved noise reduction techniques employed in the calculation of the
vector correlator 𝐺 (𝑡), we expect a significantly improved accuracy for Δ𝛼had at the Z-pole with
respect to [1]. This improved precision will be crucial to corroborate the observed tension with
data-driven approaches for Δ𝛼had (𝑄 2 ). In addition, this will either further confirm or disprove
the observed consistency between lattice results and global electroweak fits in the estimation of
(5)
Δ𝛼had (𝑀𝑍2 ).

Acknowledgments

Calculations for this project have been performed on the HPC clusters Clover and HIMster-II
at Helmholtz Institute Mainz and Mogon-II and Mogon-NHR at Johannes Gutenberg-Universität
(JGU) Mainz, on the HPC systems JUQUEEN and JUWELS and on the GCS Supercomputers
HAZELHEN and HAWK at Höchstleistungsrechenzentrum Stuttgart (HLRS). The authors grate-
fully acknowledge the support of the Gauss Centre for Supercomputing (GCS) and the John von
Neumann-Institut für Computing (NIC) projects HMZ21, HMZ23 and HINTSPEC at JSC, as
well as projects GCS-HQCD and GCS-MCF300 at HLRS. We also gratefully acknowledge the
scientific support and HPC resources provided by NHR-SW of Johannes Gutenberg-Universität
Mainz (project NHR-Gitter). This work has been supported by Deutsche Forschungsgemeinschaft
(German Research Foundation, DFG) through Project HI 2048/1-2 (Project No. 399400745) and
through the Cluster of Excellence “Precision Physics, Fundamental Interactions and Structure of
Matter” (PRISMA+ EXC 2118/1), funded within the German Excellence strategy (Project No.
39083149). This project has received funding from the European Union’s Horizon Europe research
and innovation programme under the Marie Skłodowska-Curie grant agreement No. 101106243.

8
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

References

[1] M. Cè, A. Gérardin, G. von Hippel, H. B. Meyer, K. Miura, K. Ottnad et al., The hadronic
running of the electromagnetic coupling and the electroweak mixing angle from lattice QCD,
JHEP 08 (2022) 220, [2203.08676].

[2] (Muon g-2), B. Abi et al., Measurement of the Positive Muon Anomalous Magnetic Moment
to 0.46 ppm, Phys. Rev. Lett. 126 (2021) 141801, [2104.03281].

[3] (Muon g-2), D. P. Aguillard et al., Measurement of the Positive Muon Anomalous Magnetic
Moment to 0.20 ppm, Phys. Rev. Lett. 131 (2023) 161802, [2308.06230].

[4] S. Borsanyi et al., Leading hadronic contribution to the muon magnetic moment from lattice
QCD, Nature 593 (2021) 51, [2002.12347].

[5] D. Djukanovic, G. von Hippel, S. Kuberski, H. B. Meyer, N. Miller, K. Ottnad et al., The
hadronic vacuum polarization contribution to the muon 𝑔 − 2 at long distances, 2411.07969.

[6] A. Boccaletti et al., High precision calculation of the hadronic vacuum polarisation
contribution to the muon anomaly, 2407.10913.

[7] (RBC, UKQCD), T. Blum et al., The long-distance window of the hadronic vacuum
polarization for the muon g-2, 2410.20590.

[8] A. Bazavov et al., Hadronic vacuum polarization for the muon 𝑔 − 2 from lattice QCD:
Long-distance and full light-quark connected contribution, 2412.18491.

[9] D. Becker et al., The P2 experiment, Eur. Phys. J. A 54 (2018) 208, [1802.04759].

[10] M. Bruno, T. Korzec and S. Schaefer, Setting the scale for the CLS 2 + 1 flavor ensembles,
Phys. Rev. D 95 (2017) 074504, [1608.08900].

[11] D. Mohler, S. Schaefer and J. Simeth, CLS 2+1 flavor simulations at physical light- and
strange-quark masses, EPJ Web Conf. 175 (2018) 02010, [1712.04884].

[12] D. Mohler and S. Schaefer, Remarks on strange-quark simulations with Wilson fermions,
Phys. Rev. D 102 (2020) 074506, [2003.13359].

[13] S. Kuberski, Low-mode deflation for twisted-mass and RHMC reweighting in lattice QCD,
Comput. Phys. Commun. 300 (2024) 109173, [2306.02385].

[14] (RQCD), G. S. Bali, E. E. Scholz, J. Simeth and W. Söldner, Lattice simulations with
𝑁 𝑓 = 2 + 1 improved Wilson fermions at a fixed strange quark mass, Phys. Rev. D 94 (2016)
074501, [1606.09039].

[15] A. Gerardin, T. Harris and H. B. Meyer, Nonperturbative renormalization and

𝑂 (𝑎)-improvement of the nonsinglet vector current with 𝑁 𝑓 = 2 + 1 Wilson fermions and
tree-level Symanzik improved gauge action, Phys. Rev. D 99 (2019) 014519, [1811.08209].

9
The hadronic contribution to the running of 𝛼 and the electroweak mixing angle Alessandro Conigli

[16] (ALPHA), J. Heitger and F. Joswig, The renormalised O(𝑎) improved vector current in
three-flavour lattice QCD with Wilson quarks, Eur. Phys. J. C 81 (2021) 254, [2010.09539].

[17] P. Fritzsch, Mass-improvement of the vector current in three-flavor QCD, JHEP 06 (2018)
015, [1805.07401].

[18] M. Cè et al., Window observable for the hadronic vacuum polarization contribution to the
muon g-2 from lattice QCD, Phys. Rev. D 106 (2022) 114502, [2206.06582].

[19] D. Bernecker and H. B. Meyer, Vector Correlators in Lattice QCD: Methods and
applications, Eur. Phys. J. A 47 (2011) 148, [1107.4388].

[20] A. Francis, B. Jaeger, H. B. Meyer and H. Wittig, A new representation of the Adler function
for lattice QCD, Phys. Rev. D 88 (2013) 054502, [1306.2532].

[21] S. Kuberski, M. Cè, G. von Hippel, H. B. Meyer, K. Ottnad, A. Risch et al., Hadronic
vacuum polarization in the muon g − 2: the short-distance contribution from lattice QCD,
JHEP 03 (2024) 172, [2401.11895].

[22] M. Cè, T. Harris, H. B. Meyer, A. Toniato and C. Török, Vacuum correlators at short
distances from lattice QCD, JHEP 12 (2021) 215, [2106.15293].

[23] R. Sommer, L. Chimirri and N. Husung, Log-enhanced discretization errors in integrated

correlation functions, PoS LATTICE2022 (2023) 358, [2211.15750].

[24] R. Urech, Virtual photons in chiral perturbation theory, Nucl. Phys. B 433 (1995) 234,
[hep-ph/9405341].

[25] H. Neufeld and H. Rupertsberger, The Electromagnetic interaction in chiral perturbation

theory, Z. Phys. C 71 (1996) 131, [hep-ph/9506448].

[26] M. Lüscher, Properties and uses of the Wilson flow in lattice QCD, JHEP 08 (2010) 071,
[1006.4518]. [Erratum: JHEP 03 (2014) 092].

[27] (RQCD), G. S. Bali, S. Collins, P. Georg, D. Jenkins, P. Korcyl, A. Schäfer et al., Scale
setting and the light baryon spectrum in N 𝑓 = 2 + 1 QCD with Wilson fermions, JHEP 05
(2023) 035, [2211.03744].

[28] W. I. Jay and E. T. Neil, Bayesian model averaging for analysis of lattice field theory results,
Phys. Rev. D 103 (2021) 114502, [2008.01069].

[29] J. Frison, Towards fully bayesian analyses in Lattice QCD, 2302.06550.

[30] A. Gérardin, M. Cè, G. von Hippel, B. Hörz, H. B. Meyer, D. Mohler et al., The leading
hadronic contribution to (𝑔 − 2) 𝜇 from lattice QCD with 𝑁f = 2 + 1 flavours of O(𝑎)
improved Wilson quarks, Phys. Rev. D 100 (2019) 014510, [1904.03120].

[31] C. Aubin, T. Blum, M. Golterman and S. Peris, Model-independent parametrization of the

hadronic vacuum polarization and 𝑔-2 for the muon on the lattice, Phys. Rev. D 86 (2012)
054509, [1205.3695].

10