betaFeedback

WorksPhysics and Astronomy

Bjorken sum rule down to the elastic region from the lattice

Abstract

We present a lattice determination of the Bjorken sum rule via a Feynman-Hellmann calculation of the polarised forward Compton amplitude. We evaluate the lowest moment of the nucleon's polarised structure function, g1g_1, which we calculate over the range 0.5≲Q2≲80.5\lesssim Q^2 \lesssim 8 GeV2^2, covering both non-perturbative and perturbative regions. We compare our calculated moments with the Bjorken sum rule, and discuss implications for the calculation of higher-twist effects, as well as evaluations of αs(Q2)α_s(Q^2) from hadronic quantities to complement phenomenological and other lattice methods.

Introduction

The polarised structure function g1g_1, accessible via polarised deep-inelastic scattering experiments, is central to our understanding of the spin structure of the nucleon. At the partonic level, the lowest moment of g1g_1 is related to the fraction of the total hadron spin carried by the partons,

∫01dx g1(x,Q2)→Q2→∞12∑qeq2[Δq(Q2)+Δqˉ(Q2)],(1)\int_0^1 dx\,g_1(x,Q^2) \xrightarrow{Q^2\to\infty} \frac{1}{2}\sum_q e_q^2[\Delta q(Q^2)+\Delta\bar{q}(Q^2)], \tag*{(1)}

where eqe_q is the electric charge and Δq\Delta q (Δqˉ\Delta\bar{q}) is the contribution to the spin from the (anti-)parton of flavour qq. Exploiting iso-spin symmetry, the Bjorken sum rule [1] is defined from the difference between the lowest moment of the proton and neutron g1g_1 structure functions,

Γ1p−n(Q2)=∫01dx [g1p(x,Q2)−g1n(x,Q2)],(2)\Gamma_1^{p-n}(Q^2)=\int_0^1 dx\,[g_1^p(x,Q^2)-g_1^n(x,Q^2)], \tag*{(2)}

which in the quark-parton model is equal to gAu−d/6g_A^{u-d}/6, with gAu−dg_A^{u-d} being the isovector axial charge. As the Bjorken sum rule is a flavour non-singlet quantity, radiative corrections can be calculated to a high degree of precision in perturbative QCD, therefore offering a robust avenue for precision studies of QCD dynamics. The Bjorken sum rule has been experimentally validated, most precisely by E155 at Q2=5 GeV2Q^2=5\ \mathrm{GeV}^2 [2]. The Q2Q^2 dependence of the sum rule has been mapped out at Jefferson Lab [3, 4, 5, 6], and shows agreement with theoretical predictions down to 1 GeV21\ \mathrm{GeV}^2. In the very low Q2Q^2 range, recent experiments [7] [8, 9] have mapped the Q2Q^2 dependence of the Bjorken sum rule to as low as 0.01 GeV20.01\ \mathrm{GeV}^2. In this Q2Q^2 range, the Q2Q^2 dependence of the integral in Equation (2) is instead governed by the Gerasimov-Drell-Hearn (GDH) sum rule [10–12]. The transition between these two regimes is still not well understood.

Lattice QCD offers a first-principles approach to the calculation of matrix elements of local operators. Calculations of the isovector axial charge have reached the sub-percent level [13–19], which is required for the Bjorken sum rule at the leading-twist level. Higher-twist contributions are suppressed by inverse powers of Q2Q^2 and therefore become relevant at low Q2Q^2. However, direct lattice calculations of higher-twist matrix elements are challenging. On the lattice, operators can mix with others transforming in the same irreducible representation of the hypercubic group [20–22], including lower-dimensional operators that induce power-divergent mixing. Twist-three matrix elements remain tractable in suitably chosen channels [23–27]; however, corresponding calculations of twist-four matrix elements and beyond have not been pursued, as the operator-mixing structure becomes substantially more complicated and is not fully known. These sub-leading twist contributions offer unique insight into hadron structure, going beyond the parton model by probing quark-gluon correlations [28] and are essential for understanding the internal forces that bind quarks into hadrons. Alternative methods for calculating these higher-twist matrix elements are hence highly desirable.

Beyond hadron structure implications, the Bjorken sum rule offers an excellent avenue for a non-perturbative definition of the strong coupling αs\alpha_s across all momentum scales. As the Bjorken sum rule is a non-singlet quantity, at leading order in QCD it has vanishing anomalous dimension. Additionally, the non-singlet Wilson coefficient is known to next-to-next-to-next leading order (N3LO\mathrm{N}^3\mathrm{LO}) [29]. These two facts make the Bjorken sum rule, as well as the Gross-Llewellyn Smith sum rule [30], ideal candidates for determining αs\alpha_s [31, 32], and for a non-perturbative definition of αs\alpha_s via an effective charge approach analogous to QED (see Refs. [33, 34] for reviews). First principles calculations of αs\alpha_s via these sum rules are highly desirable as they illuminate the boundaries between perturbative and non-perturbative regimes, offer insights into confinement and provide an alternative determination of αs\alpha_s based on a hadronic observable.

In this work, we use the Feynman–Hellmann technique to determine the Bjorken sum rule across a range of hard scales, Q2Q^2, and compare with predictions from the parton model and perturbative QCD. The technique has proven to be a valuable approach for evaluating the Compton amplitude in lattice QCD [35], with applications to moments of unpolarised nucleon structure functions [36, 37], generalised parton distributions [38, 39], and the Gross–Llewellyn Smith sum rule [40]. Alongside direct four-point-function calculations of the Compton amplitude [41, 42], it provides a complementary framework for evaluating moments of structure functions across a broad range of probing momenta, enabling studies of power corrections and access to higher-twist matrix elements without the complications of power-divergent operator mixing. Preliminary investigations of spin-dependent structure functions using this approach were reported in Ref. [43].

The rest of the paper is structured as follows: In Section II we present the tensor decomposition of the Compton amplitude and relate the polarised Compton structure function to the nucleon structure function g1g_1. Our Feynman-Hellmann approach is explained in Section III, followed by details of our lattice calculation and analysis in Section IV. We present and discuss our results in Section V, before summarising and offering some concluding remarks in Section VI.

The Spin-Dependent Compton Amplitude

The forward Compton amplitude is the time-ordered product of two vector currents inserted between two nucleon states,

[Tμν(p,q)]σ′σ=i∫d4zeiq⋅z⟨p,σ′∣T{Jμ(z)Jν(0)}∣p,σ⟩(3)[T_{\mu\nu}(p,q)]_{\sigma'\sigma}=i\int d^4z e^{iq\cdot z}\langle p,\sigma'|\mathcal{T}\{J_\mu(z)J_\nu(0)\}|p,\sigma\rangle \tag*{(3)}

where pp (qq) is the nucleon (current) momentum, σ′,σ\sigma',\sigma are the outgoing and incoming nucleon spins and the vector current is defined as

Jμ(z)=∑fQfψˉf(z)γμψf(z)(4)J_\mu(z)=\sum_f Q_f\bar{\psi}_f(z)\gamma_\mu\psi_f(z) \tag*{(4)}

with QfQ_f being the electric charge of the quark flavour ff. To access the structure functions, we consider the Lorentz decomposition of the Compton amplitude [44, 45] in Minkowski space,

[Tμν(p,q)]σ′σ=(−gμν+qμqνq2)F1(ω,Q2)+(pμ−p⋅qq2qμ)(pν−p⋅qq2qν)F2(ω,Q2)p⋅q+ip⋅qϵμναβqαsσ′σβG1(ω,Q2)−i(p⋅q)2ϵμναβqα(p⋅q sσ′σβ−sσ′σ⋅q pβ)G2(ω,Q2)(5)\begin{aligned} [T_{\mu\nu}(p,q)]_{\sigma'\sigma}={}&\left(-g_{\mu\nu}+\frac{q_\mu q_\nu}{q^2}\right)\mathcal{F}_1(\omega,Q^2)+\left(p_\mu-\frac{p\cdot q}{q^2}q_\mu\right)\left(p_\nu-\frac{p\cdot q}{q^2}q_\nu\right)\frac{\mathcal{F}_2(\omega,Q^2)}{p\cdot q}\\ &+\frac{i}{p\cdot q}\epsilon_{\mu\nu\alpha\beta}q^\alpha s^\beta_{\sigma'\sigma}\mathcal{G}_1(\omega,Q^2)-\frac{i}{(p\cdot q)^2}\epsilon_{\mu\nu\alpha\beta}q^\alpha(p\cdot q\,s^\beta_{\sigma'\sigma}-s_{\sigma'\sigma}\cdot q\,p^\beta)\mathcal{G}_2(\omega,Q^2) \tag*{(5)} \end{aligned}

where σ\sigma (σ′\sigma') is the spin index of the incoming (outgoing) nucleon, ϵ0123=+1\epsilon_{0123}=+1, ω=2p⋅q/Q2\omega=2p\cdot q/Q^2, Q2=−q2Q^2=-q^2 and the nucleon structure functions are F1,F2\mathcal{F}_1,\mathcal{F}_2 and G1,G2\mathcal{G}_1,\mathcal{G}_2 for the unpolarised and polarised cases respectively (the normalisations have been chosen to match the scaling functions of the Particle Data Group [46]). sσ′σμs^\mu_{\sigma'\sigma} is the spin matrix of the nucleon axial current bilinear, defined as

sσ′σμ=12uˉ(p,σ′)γ5γμu(p,σ)(6)s^\mu_{\sigma'\sigma}=\frac{1}{2}\bar{u}(p,\sigma')\gamma_5\gamma^\mu u(p,\sigma) \tag*{(6)}

and satisfies s2=−m2s^2=-m^2. In order to access the polarised structure functions, we must consider off-diagonal components of the Compton amplitude, TμνT_{\mu\nu} with μ≠ν\mu\ne\nu. From analytic properties of the Compton tensor, crossing-symmetry relations and the optical theorem, the connection between the Compton structure function and the physical structure function is given by a dispersion relation,

G1(ω,Q2)=4ω∫01dxg1(x,Q2)1−x2ω2.(7)\mathcal{G}_1(\omega,Q^2)=4\omega\int_0^1 dx\frac{g_1(x,Q^2)}{1-x^2\omega^2}. \tag*{(7)}

On the lattice, the Wick rotation restricts access to the unphysical region ∣ω∣<1|\omega|<1 [36]. In this region, the denominator may be expanded in a geometric series, allowing the Compton structure function to be expressed in terms of Cornwall–Norton moments of g1g_1:

G1(ω,Q2)ω=4∑n=1∞ω2n−2M~2n−1(1)(Q2).(8)\frac{\mathcal{G}_1(\omega,Q^2)}{\omega}=4\sum_{n=1}^{\infty}\omega^{2n-2}\widetilde{M}^{(1)}_{2n-1}(Q^2). \tag*{(8)}

The moments of g1g_1 are defined as

M~2n−1(1)(Q2)=∫01dx x2n−2g1(x,Q2),(9)\widetilde{M}^{(1)}_{2n-1}(Q^2)=\int_0^1 dx\,x^{2n-2}g_1(x,Q^2), \tag*{(9)}

for n=1,2,3,…n = 1, 2, 3, \ldots. The n=1n = 1 moment gives direct access to the Bjorken sum rule

Γ1p−n(Q2)=M~1,p(1)(Q2)−M~1,n(1)(Q2).(10)\Gamma_1^{p-n}(Q^2) = \widetilde{M}_{1,p}^{(1)}(Q^2) - \widetilde{M}_{1,n}^{(1)}(Q^2). \tag*{(10)}

Feynman-Hellmann Approach

The Feynman-Hellmann approach provides an alternative formulation of the Compton amplitude in terms of two-point correlation functions, avoiding the need for an explicit four-point-function calculation.

We proceed by introducing two perturbing fields to the bare QCD action,

S(λ)=S0+λ1∫d4z(eiq⋅z+e−iq⋅z)Jμ(z)−iλ2∫d4y(eiq⋅y−e−iq⋅y)Jν(y),(11)S(\boldsymbol{\lambda}) = S_0 + \lambda_1 \int d^4z \left(e^{iq\cdot z} + e^{-iq\cdot z}\right) \mathcal{J}_{\mu}(z) - i\lambda_2 \int d^4y \left(e^{iq\cdot y} - e^{-iq\cdot y}\right) \mathcal{J}_{\nu}(y), \tag*{(11)}

where S0S_0 is the unperturbed action and λ=(λ1,λ2)\boldsymbol{\lambda} = (\lambda_1,\lambda_2) denotes the coupling strength of the external perturbing fields and the quark fields. The current Jμ(z)=ZVq‾f(z)γμqf(z)\mathcal{J}_{\mu}(z) = Z_V\overline{q}_f(z)\gamma_{\mu}q_f(z) is the renormalised vector currents coupling to the quark fields of flavour ff along the direction μ\mu. The external current momentum q=(q,0)q = (q,0) has q4=0q_4 = 0 by construction from the Feynman-Hellmann approach [36]. ZVZ_V is the RI′-MOM renormalisation constant for the local vector current, previously calculated on this lattice in Ref. [47]. A second order Feynman-Hellmann relation for this perturbation has previously been derived in Ref. [40] and reads

∂2ENλ(p,q)∂λ1∂λ2∣λ=0=iTμν(p,q)−Tμν(p,−q)2EN(p),(12)\left.\frac{\partial^2 E_N^{\lambda}(\boldsymbol{p},q)}{\partial\lambda_1\partial\lambda_2}\right|_{\boldsymbol{\lambda}=0} = i\frac{T_{\mu\nu}(p,q)-T_{\mu\nu}(p,-q)}{2E_N(p)}, \tag*{(12)}

where TμνT_{\mu\nu} is the Compton amplitude,

Tμν=ρσ′σ[Tμν]σ′σ,(13)T_{\mu\nu} = \rho_{\sigma'\sigma}[T_{\mu\nu}]_{\sigma'\sigma}, \tag*{(13)}

with [Tμν]σ′σ[T_{\mu\nu}]_{\sigma'\sigma} from Eq. (3), ρ=12(1+σ⋅s/mN)\rho= \frac{1}{2}(1+\boldsymbol{\sigma}\cdot\boldsymbol{s}/m_N) being the spin-density matrix, ENλ(p,q)E_N^{\lambda}(p,q) (ENE_N) is the (un-)perturbed nucleon energy with momentum pp and is also a function of the momentum carried by the perturbing current, qq. In order to isolate the G1G_1 structure function, we make a certain choice of kinematics, as discussed in Section IV.

Degeneracy of Perturbed Energies

Equation (12) assumes that there is only a single low-lying perturbed energy, however the polarised Compton amplitude depends on the spin of the nucleon, and hence our perturbing potential will break the degeneracy between the two spin states. Therefore there will be two low-lying perturbed energy states, one corresponding to the spin-up state at λ=0\lambda= 0 and the other to the spin-down state. The spectral form of our perturbed two-point correlator hence has the form

Gλ(p,q,t)≈A+(p,q)e−Eλ(+)(p,q)t+A−(p,q)e−Eλ(−)(p,q)t(14)G_{\lambda}(p,q,t) \approx A_+(p,q)e^{-E_{\lambda}^{(+)}(p,q)t}+A_-(p,q)e^{-E_{\lambda}^{(-)}(p,q)t} \tag*{(14)}

with Eλ(±)E_{\lambda}^{(\pm)} being the eigenvalues of the perturbed Hamiltonian, and A±A_{\pm} being the corresponding overlap factors. As shown in Appendix A, the perturbed Hamiltonian in the space of the two low-lying states has the following form, up to order λ2\lambda^2,

H^λ=H^0+ΔEunpol(λ1,λ2)+λ1λ2v⋅σ,(15)\hat{H}_{\lambda} = \hat{H}_0 + \Delta E_{\mathrm{unpol}}(\lambda_1,\lambda_2) + \lambda_1\lambda_2\boldsymbol{v}\cdot\boldsymbol{\sigma}, \tag*{(15)}

where H^0∣p,s⟩=EN(p)∣p,s⟩\hat{H}_0|p,s\rangle= E_N(p)|p,s\rangle defines the eigenstates of the unperturbed QCD Hamiltonian, the coefficients v\boldsymbol{v} capture the kinematic factors and structure functions from the polarised Compton amplitude, and σ=(σ1,σ2,σ3)\boldsymbol{\sigma} = (\sigma_1,\sigma_2,\sigma_3) are the Pauli sigma matrices. ΔEunpol(λ1,λ2)\Delta E_{\mathrm{unpol}}(\lambda_1,\lambda_2) captures the perturbations from the diagonal, unpolarised components of the Compton amplitude, corresponding to repeated insertions of the Jμ\mathcal{J}_{\mu} or Jν\mathcal{J}_{\nu} currents. Both have the form

[Tμν(p,q)]σ′σ=[−gμνF1(ω,Q2)+(pμ+p⋅qq2qμ)2F2(ω,Q2)p⋅q]δσ′σ,(16)[T_{\mu\nu}(p,q)]_{\sigma'\sigma} = \left[ -g_{\mu\nu}\mathcal{F}_1(\omega,Q^2) +\left(p_{\mu}+\frac{p\cdot q}{q^2}q_{\mu}\right)^2\frac{\mathcal{F}_2(\omega,Q^2)}{p\cdot q} \right]\delta_{\sigma'\sigma}, \tag*{(16)}

where μ\mu is just a repeated index and not summed over. We consider a spatial-temporal component of the polarised Compton amplitude for a nucleon at rest,

[Ti4(p,q)]σ′σ=2mNiϵijkqkσj′σQ2(G1(ω,Q2)ω+G2(ω,Q2)ω)∣ω=0.(17)[T_{i4}(p,q)]_{\sigma'\sigma} = \left. \frac{2m_N i\epsilon_{ijk}q_k\sigma'_{j}\sigma}{Q^2} \left(\frac{\mathcal{G}_1(\omega,Q^2)}{\omega}+\frac{\mathcal{G}_2(\omega,Q^2)}{\omega}\right) \right|_{\omega=0}. \tag*{(17)}

Here, we note that the term G2(ω,Q2)/ω∣ω=0\mathcal{G}_2(\omega,Q^2)/\omega|_{\omega=0} is proportional to the lowest moment of the g2g_2 structure function. The Burkhardt-Cottingham sum rule [48] predicts that this moment vanishes at all orders in Q2Q^2. However, it is worth noting that the Burkhardt-Cottingham sum rule is not a first-principles relation. With this caveat in mind, for our analysis we will assume that this term is negligible and focus on G1(ω,Q2)/ω∣ω=0\mathcal{G}_1(\omega,Q^2)/\omega|_{\omega=0}. Comparing Equations (16) and (17), we see that the unpolarised Compton amplitude is diagonal in the nucleon spins, whereas the polarised Compton amplitude mixes spin eigenstates.

For the Feynman-Hellmann relation, Equation (12), to be valid, we require the perturbed Hamiltonian to be diagonal in spin-space. For certain kinematic choices, as outlined in Appendix A, the contribution from the polarised Compton amplitude remains diagonal in spin-space. As an example, for the spatial-temporal components of the polarised Compton amplitude, terms proportional to σ3\sigma_3 remain diagonal. We therefore select the component T14(p,q)T_{14}(p,q) of the Compton amplitude and work with transfer momentum of the form q=(0,q2,0)\mathbf{q}=(0,q_{2},0) to keep the perturbed Hamiltonian diagonal. This allows us to use a single exponential to extract the energy shift and therefore access the polarised Compton amplitude. As a proof of concept, we also consider one transfer momentum which introduces off-diagonal terms to the Hamiltonian and explore the use of a generalised eigenvalue problem (GEVP) to extract the two low-lying energy states. These two cases are discussed separately in Section IV.

Calculation and Analysis Details

The gauge configurations used in this work were generated by the QCDSF Collaboration with 2+12+1 flavours of dynamical light and strange quarks, using a stout-smeared, non-perturbatively O(a)\mathcal{O}(a)-improved Wilson action together with a tree-level Symanzik-improved gauge action [49]. A single ensemble is analysed, with volume V=483×96V=48^{3}\times96 and coupling β=5.65\beta=5.65, corresponding to a lattice spacing of a≃0.068 fma\simeq0.068\,\mathrm{fm} [50]. The ensemble lies at the SU(3)-flavour-symmetric point, with the degenerate uu, dd, and ss quark masses set equal to approximately the physical flavour-singlet mass, m‾=(2ms+ml)/3\overline{m}=(2m_{s}+m_{l})/3 [52] [51], yielding mπ≈412 MeVm_{\pi}\approx412\,\mathrm{MeV}. Further details of the ensemble are given in Table I.

The primary target of this investigation is the determination of the Bjorken sum rule over a broad range of momentum scales, 0.5<Q2<8 GeV20.5<Q^{2}<8\,\mathrm{GeV}^{2}, allowing its Q2Q^{2} dependence to be explored. The specific kinematics of the momentum transfer are displayed in Table II. Since the present analysis focuses on the lowest moment, only the Compton amplitude at ω=0\omega=0 is required. This allows all calculations to be performed with the nucleon at rest, p=(0,0,0)\mathbf{p}=(0,0,0).

At large Euclidean times, the perturbed two-point correlation function takes the form

Gλ(p,q,t)→t≫0Aλ(p,q)e−ENλ(p,q)t,(18)G_{\lambda}(\mathbf{p},\mathbf{q},t)\mathrel{\xrightarrow{t\gg0}} A_{\lambda}(\mathbf{p},\mathbf{q})e^{-E_{N}^{\lambda}(\mathbf{p},\mathbf{q})t}, \tag*{(18)}

where ENλE_{N}^{\lambda} is the perturbed energy of the ground state nucleon in the presence of a background field, and AλA_{\lambda} is the corresponding overlap factor. For Q2≲1 GeV2Q^{2}\lesssim1\,\mathrm{GeV}^{2}, excited states are not sufficiently suppressed in the time range over which we extract the perturbed energy shifts. We therefore use a two-exponential ansatz,

G(±)(p,t)=A0(±)e−E0(±)t+A1(±)e−E1(±)t,(19)G^{(\pm)}(\mathbf{p},t)=A_{0}^{(\pm)}e^{-E_{0}^{(\pm)}t}+A_{1}^{(\pm)}e^{-E_{1}^{(\pm)}t}, \tag*{(19)}

where E0E_{0}, E1E_{1} refer to the ground and first excited state energies, and A0A_{0} and A1A_{1} are the corresponding overlaps, with the superscripts (±)(\pm) denoting either the spin-up (+)(+) or spin-down (−)(-) correlator. The required energy splitting is then given by

2ΔEpol=E0(+)−E0(−).(20)2\Delta E_{\mathrm{pol}}=E_{0}^{(+)}-E_{0}^{(-)}. \tag*{(20)}

Further discussion of this approach and a comparison to the ratio method can be found in Appendix A of Ref. [40].

In order to isolate the perturbed energy shifts, we define ratios of components of the positive-parity spin matrix, corresponding to the upper left 2×22\times2 sub-block of the two-point correlation function in the Dirac representation. This matrix is denoted as

G(p,t)=(G↑↑(p,t)G↑↓(p,t)G↓↑(p,t)G↓↓(p,t)).(21)G(\mathbf{p},t)= \begin{pmatrix} G^{\uparrow\uparrow}(\mathbf{p},t) & G^{\uparrow\downarrow}(\mathbf{p},t) \\ G^{\downarrow\uparrow}(\mathbf{p},t) & G^{\downarrow\downarrow}(\mathbf{p},t) \end{pmatrix}. \tag*{(21)}

The treatment of this correlator matrix varies depending on whether the perturbed Hamiltonian is diagonal in the nucleon spins, or non-diagonal. We detail each of these cases below.

Diagonal Perturbations

Diagonal perturbations correspond to the case where the Compton amplitude is proportional to σ3\sigma_{3}. From Equation (17) and considering the component T14T_{14}, this is equivalent to the kinematic choice q3=0q_{3}=0. Therefore the spin-up and spin-down states along the z^\hat{z} axis form an eigenbasis of the perturbed Hamiltonian. In this case, the energy shift due to the polarised Compton amplitude can be isolated through the ratio

R(t)=Gλ↑↑(0,t)G0↓↓(0,t)G0↑↑(0,t)Gλ↓↓(0,t)→t≫0Aλe−2ΔEpolt,(22)\mathcal{R}(t)=\frac{G_{\lambda}^{\uparrow\uparrow}(0,t)G_{0}^{\downarrow\downarrow}(0,t)}{G_{0}^{\uparrow\uparrow}(0,t)G_{\lambda}^{\downarrow\downarrow}(0,t)} \xrightarrow{t\gg0} A_{\lambda}e^{-2\Delta E_{\mathrm{pol}}t}, \tag*{(22)}

where the subscripts λ\lambda and 00 indicate perturbed and unperturbed correlators respectively and the energy shift due to the polarised Compton amplitude, neglecting terms higher order in λ1,λ2\lambda_{1},\lambda_{2} is

ΔEpol=λ1λ22q2Q2G1(ω,Q2)ω∣ω=0.(23)\Delta E_{\mathrm{pol}}=\lambda_{1}\lambda_{2}\left.\frac{2q_{2}}{Q^{2}}\frac{G_{1}(\omega,Q^{2})}{\omega}\right|_{\omega=0}. \tag*{(23)}

The inclusion of the unperturbed correlators in the ratio (22) enhances the cancellation of correlated gauge noise between bootstrap samples. A representative plot of this ratio for a nucleon with q=(0,5,0)2πL\mathbf{q}=(0,5,0)\frac{2\pi}{L} is shown in Figure 1. The signal region appears around t=6t=6, however it is also lost to noise quickly, around t=14t=14. Figure 1 also shows the λ1=λ2=λ\lambda_{1}=\lambda_{2}=\lambda dependence of the energy shifts and the corresponding quadratic fit. From Equation (23), the quadratic fit parameter is directly proportional to the Compton structure function G1(ω,Q2)G_{1}(\omega,Q^{2}).

Figure 1. (Top) Effective energy plot analogue for the ratio, Eq. (22), with λ1=λ2=λ\lambda_1=\lambda_2=\lambda, for λ=0.0125,0.025\lambda=0.0125, 0.025. The calculated energy shifts, uncertainties and their corresponding fit regions are indicated by the horizontal lines and shaded regions, respectively. In this figure, we show results for the uuuu contribution for q=(0,5,0)2πL\mathbf{q}=(0,5,0)\frac{2\pi}{L} at Q2≈3.6 GeV2Q^2\approx3.6\ \mathrm{GeV}^2. (Bottom) Corresponding λ\lambda dependence of the polarised energy shift.

Non-Diagonal Perturbations

For the case where q3≠0q_{3}\ne0, the perturbation is no longer solely proportional to σ3\sigma_{3} and introduces a mixing between spin components. We consider one such transfer momentum, q=(0,3,5)2πL\mathbf{q}=(0,3,5)\frac{2\pi}{L}, as a proof of concept. Again, we use a nucleon momentum of p=(0,0,0)\mathbf{p}=(0,0,0) corresponding to ω=0\omega=0. To determine the energy eigenvalues of the perturbed Hamiltonian, we must first diagonalise the correlator matrix (21). We implement a generalised eigenvalue problem (GEVP) to determine the required left v(i)†v^{(i)\dagger} and right u(i)u^{(i)} eigenvectors for each value of λ\lambda,

Details of the gauge ensemble used in this work. NfN_f is the number of fermion flavours, β\beta is the inverse gauge coupling, aa is the lattice spacing, cSWc_{\mathrm{SW}} is the clover coefficient, κℓ\kappa_\ell (κs\kappa_s) is the hopping parameter for the light (strange) quarks, LL (TT) is the spatial (temporal) extent of the lattice, mπm_\pi (mKm_K) is the pion (kaon) mass, ZVZ_V is the vector current renormalisation constant on this lattice, and NcfgN_{\mathrm{cfg}} is the number of configurations used.

NfN_fβ\betaaa [fm]cSWc_{\mathrm{SW}}κℓ=κs\kappa_\ell=\kappa_sL3×TL^3 \times Tmπ=mKm_\pi=m_K [MeV]ZVZ_VNcfgN_{\mathrm{cfg}}
2+12+15.650.0682.480.122005483×9648^3 \times964120.8615535

Table 1.

Momentum insertions q=(q1,q2,q3)\mathbf{q}=(q_1,q_2,q_3) considered in this work, noting that all q\mathbf{q} values have q1=0q_1=0. Momenta for which we perform a multi-exponential fit are given in italics, and momenta requiring a GEVP have q3≠0q_3\ne0.

q\mathbf{q} [2π/L2\pi/L]Q2Q^2 [GeV2^2]
(0,2,0)(0,2,0)0.577
(0,3,0)(0,3,0)1.298
(0,4,0)(0,4,0)2.308
(0,5,0)(0,5,0)3.606
(0,3,5)(0,3,5)4.904
(0,6,0)(0,6,0)5.193
(0,7,0)(0,7,0)7.068

Table 2.

Gλ(p,t0+Δt)u(i)(t0,Δt)=c(i)(Δt)Gλ(p,t0)u(i)(t0,Δt)G_\lambda(\mathbf{p},t_0+\Delta t)u^{(i)}(t_0,\Delta t)=c^{(i)}(\Delta t)G_\lambda(\mathbf{p},t_0)u^{(i)}(t_0,\Delta t)
v(i)†(t0,Δt)Gλ(p,t0+Δt)=c(i)(Δt)v(i)†(t0,Δt)Gλ(p,t0),(24)v^{(i)\dagger}(t_0,\Delta t)G_\lambda(\mathbf{p},t_0+\Delta t)=c^{(i)}(\Delta t)v^{(i)\dagger}(t_0,\Delta t)G_\lambda(\mathbf{p},t_0), \tag*{(24)}

for the eigenstates i=±i=\pm and some choice of initial time t0t_0 and separation Δt\Delta t, yielding the eigenvalues

c(i)(Δt)=e−Eλ(i)Δt,(25)c^{(i)}(\Delta t)=e^{-E_\lambda^{(i)}\Delta t}, \tag*{(25)}

with Eλ(i)E_\lambda^{(i)} being the perturbed ground state energy of state ii. The correlator matrix is rotated to the eigenbasis using the left and right generalised eigenvectors,

Cλ(i)(p,t)=v(i)†Cλ(p,t)u(i),(26)C_\lambda^{(i)}(\mathbf{p},t)=v^{(i)\dagger}C_\lambda(\mathbf{p},t)u^{(i)}, \tag*{(26)}

projecting out the corresponding eigenstate i=±i=\pm. To isolate the energy shift from the polarised Compton amplitude, we construct the ratio

Rλ(t)=Cλ(+)(0,t)Cλ(−)(0,t)C0(+)(0,t)C0(−)(0,t)→t≫0Aλe−2ΔEpolt,(27)R_\lambda(t)=\frac{C_\lambda^{(+)}(0,t)C_\lambda^{(-)}(0,t)}{C_0^{(+)}(0,t)C_0^{(-)}(0,t)} \xrightarrow[t\gg0]{} A_\lambda e^{-2\Delta E_{\mathrm{pol}}t}, \tag*{(27)}

where C0(±)(0,t)C_0^{(\pm)}(0,t) is the rotated unperturbed correlator, using the same eigenvectors as Equation (26). As with the diagonal case, the inclusion of the unperturbed correlator aids in the cancellation of correlated gauge noise between bootstraps. For the non-diagonal case, the energy shift due to the polarised Compton amplitude, neglecting terms higher order in λ1,λ2\lambda_1,\lambda_2 is

ΔEpol=λ1λ22q22+q32Q2∣G1(ω,Q2)ω∣ω=0(28)\Delta E_{\mathrm{pol}}=\lambda_1\lambda_2\frac{2\sqrt{q_2^2+q_3^2}}{Q^2}\left|\frac{\mathcal{G}_1(\omega,Q^2)}{\omega}\right|_{\omega=0} \tag*{(28)}

A representative plot of this ratio is shown in Figure 2. As with the diagonal case, the signal region appears early in Euclidean time and is equally lost to noise quite early. It is also worth noting that uncertainty in the calculated ratio is larger than the diagonalised case. This is likely a result of propagating statistical uncertainty in the generalised eigenvectors to the calculation of the energy shifts.

Figure 2. (Top) Effective energy plot analogue for the ratio, Eq. (27), with λ1=λ2=λ\lambda_1=\lambda_2=\lambda, for λ=0.0125,0.025\lambda=0.0125,0.025. The calculated energy shifts, uncertainties and their corresponding fit regions are indicated by the horizontal lines and shaded regions respectively. In this figure, we show results for the uuuu contribution for q=(0,3,5)2πa\mathbf{q}=(0,3,5)\frac{2\pi}{a} at Q2∼4.9 GeV2Q^2\sim4.9\ \mathrm{GeV}^2. (Bottom) Corresponding λ\lambda dependence of the polarised energy shift.

Figure 2 also shows the λ\lambda dependence of the energy shifts and the corresponding quadratic fit. From Equation (28), the quadratic fit parameter is directly proportional to the Compton structure function G1(ω,Q2)G_1(\omega,Q^2). Correspondingly, it also has a larger uncertainty when compared to the diagonal case.

Elastic Contributions

In the mid Q2Q^2 region, it is important to note that both elastic and inelastic processes contribute to the Compton amplitude. Elastic processes are expected to dominate in the low Q2Q^2 region, where the higher-twist contributions are also present. Therefore, it is imperative to first quantify the size of the elastic contributions. The elastic pole of the g1g_1 structure function for a quark of flavour ff is given by [53]

g1,fel(x,Q2)=12F1(f)(Q2)GM(f)(Q2)δ(1−x),(29)g_{1,f}^{\mathrm{el}}(x,Q^2)=\frac{1}{2}F_1^{(f)}(Q^2)G_M^{(f)}(Q^2)\delta(1-x), \tag*{(29)}

where GM(Q2)G_M(Q^2) is the Sachs magnetic form factor,

GM(Q2)=F1(Q2)+F2(Q2),(30)G_M(Q^2)=F_1(Q^2)+F_2(Q^2), \tag*{(30)}

with F1(Q2)F_1(Q^2) and F2(Q2)F_2(Q^2) being the Dirac and Pauli electromagnetic form factors. The δ(1−x)\delta(1-x) factor ensures that the only contribution to the dispersion integral, Equation (7), comes from x=1x=1. For a nucleon at rest, we have ω=0\omega=0, hence Equation (29) gives the elastic contribution to the lowest moment, M~1(1),el.,qq(Q2)\widetilde{M}_1^{(1),\mathrm{el.}}{}_{,qq}(Q^2).

M~1,qq(1),el.(Q2)=12F1(q)(Q2)GM(q)(Q2).(31)\widetilde{M}_{1,qq}^{(1),\mathrm{el.}}(Q^2)=\frac{1}{2}F_1^{(q)}(Q^2)G_M^{(q)}(Q^2). \tag*{(31)}

The elastic contribution is subtracted from the calculated moments to give the inelastic part of the moment,

M~1,qq(1),inel.(Q2)=M~1,qq(1)(Q2)−M~1,qq(1),el.(Q2),(32)\widetilde{M}_{1,qq}^{(1),\mathrm{inel.}}(Q^2)=\widetilde{M}_{1,qq}^{(1)}(Q^2)-\widetilde{M}_{1,qq}^{(1),\mathrm{el.}}(Q^2), \tag*{(32)}

for qq=uu,ddqq=uu,dd. The electromagnetic form factors F1(Q2)F_1(Q^2) and F2(Q2)F_2(Q^2) are calculated from three-point function methods on the same set of gauge configurations. We fit both F1(Q2)F_1(Q^2) and GM(Q2)G_M(Q^2) with a dipole parameterisation, yielding a 1/Q81/Q^8 dependence on the elastic contributions. Further details on the calculation of the electromagnetic form factors and the elastic contributions can be found in Appendix B.

Results and Discussion

Before presenting our results, we first comment on the reduction of discretisation errors arising from broken rotational symmetry on the lattice. In Refs. [54, 55, 56], a lattice operator product expansion (LOPE) expression for the Compton amplitude was derived up to first order in the covariant derivative. Starting from the lowest order expression and following the procedure discussed in Section III and Appendix A, we obtain an equivalent, discretised expression for the Compton amplitude. Matching terms with the continuum version, the LOPE-improvement is achieved by replacing the kinematic factor

qiQ2⟶sin⁡qi∑i[sin⁡2qi+(1−cos⁡qi)2],(33)\frac{q_i}{Q^2}\longrightarrow\frac{\sin q_i}{\sum_i[\sin^2q_i+(1-\cos q_i)^2]}, \tag*{(33)}

in Equations (23) and (28). A similar expression appears in our calculation of the Gross-Llewellyn Smith sum rule [40], where an alternative derivation of this factor is shown.

Flavour Structure of the Moments

We first present our results for the uuuu and dddd contributions to the lowest moment of the g1g_1 structure function, calculated using the procedure outlined in Section IV

M~1,qq(1)(Q2)=G1qq(ω,Q2)4ω∣ω=0,(34)\widetilde{M}_{1,qq}^{(1)}(Q^2)=\left.\frac{G_1^{qq}(\omega,Q^2)}{4\omega}\right|_{\omega=0}, \tag*{(34)}

at each of the Q2Q^2 indicated in Table II, where qqqq labels the flavour-diagonal pieces, uuuu and dddd. We show the LOPE-improved moments in Figure 3 for the uuuu and dddd contributions. Our calculated moments cover a wide range of Q2Q^2, incorporating both perturbative and nonperturbative scales where different physical phenomena are apparent. We therefore divide the discussion to two regions, although the boundary between these two regions is not definitive and likely exhibit quark mass and lattice spacing dependence.

Figure 3. The uuuu and dddd contributions to the lowest moment of the g1g_1 structure function, as a function of Q2Q^2. We compare the raw lattice results (blue circles) with the QPM model prediction (grey band), perturbative QCD result at N3^3LO accuracy (green band) and perturbative QCD result with elastic contributions added (yellow band).

In the large Q2Q^2 region, Q2≳4 GeV2Q^2 \gtrsim4\ \mathrm{GeV}^2, the moments are dominated by the leading-twist parton distributions and their associated contributions from perturbative QCD. In the quark-parton model (QPM), the lowest moment of the g1g_1 structure function is given by

M~1,qq(1)(Q2)→Q2→∞Δq2,(35)\widetilde{M}^{(1)}_{1,qq}(Q^2) \xrightarrow{Q^2\to\infty} \frac{\Delta q}{2}, \tag*{(35)}

with Δq\Delta q being the fraction of the hadron’s spin carried by the quark qq. At the hadronic level, this defines the Ellis-Jaffe sum rule [57], where the hadronic structure function g1g_1 is the charge-weighted sum of the quark spin contributions. Hence to compare our results to the QPM, we also calculate Δq\Delta q for both uu and dd quarks on the same set of configurations using three-point function methods. Our values for Δq\Delta q and associated statistical uncertainty are shown as grey bands in Figure 3. We utilise three source-sink separations and a two-exponential ansatz for both the two- and three-point correlation functions, and renormalise the axial current in the RI′^\prime-MOM scheme [47]. As a rigorous evaluation of these matrix elements is not the focus of this work, we leave a detailed discussion of their calculation to Appendix § 1.3.3. We see at our largest Q2Q^2 point that our calculated moments for both uu and dd quarks still deviate from the QPM prediction.

We also compare our calculated moments to the most precise perturbative QCD calculations. In full QCD, the QPM receives perturbative corrections which have been calculated to N3^3LO [58, 59, 60],

M~1,qq(1)(Q2)=Δq2[1−∑i=14Ciasi(Q2)]+ΔHTQ2+⋯(36)\widetilde{M}^{(1)}_{1,qq}(Q^2) = \frac{\Delta q}{2}\left[1-\sum_{i=1}^{4} C_i a_s^i(Q^2)\right]+\frac{\Delta_{HT}}{Q^2}+\cdots \tag*{(36)}

where the Wilson coefficients are C1=1C_1 = 1, C2=5512−13nfC_2 = \frac{55}{12}-\frac{1}{3}n_f, C3=41.440−8.020nf+0.177nf2C_3 = 41.440-8.020n_f+0.177n_f^2 and C4=479.4−117.6nf+7.46nf2−0.1037nf3C_4 = 479.4-117.6n_f+7.46n_f^2-0.1037n_f^3, with nf=3n_f=3, as(Q2)≡αsMS‾(Q2)/πa_s(Q^2)\equiv\alpha_s^{\overline{\mathrm{MS}}}(Q^2)/\pi and ΔHT\Delta_{HT} denotes the sub-leading twist contributions to the sum rule. The solid green band in Figure 3 shows the leading-twist result with perturbative corrections, given by Eq. (36) without the higher-twist term, calculated using αs(Q2)\alpha_s(Q^2) estimated to 4-loop accuracy in the MS‾\overline{\mathrm{MS}} scheme following the PDG definition [46] and with the three flavour ΛMS‾(3)=338(10)\Lambda_{\overline{\mathrm{MS}}}^{(3)}=338(10) MeV quoted by the Flavour Lattice Averaging Group [61], based on determinations from Refs. [62, 63, 64, 65, 66, 67, 68, 69, 70]. The running of αs\alpha_s is performed using the CRunDec package [71, 72]. Our calculated moments are in good agreement with the perturbatively-evolved, leading-twist expectation in this high Q2Q^2 region, albeit with mild discretisation artifacts, as was seen in our similar study of the Gross-Llewellyn Smith sum rule [40]. The calculated moment at Q2∼5 GeV2Q^2\sim5\ \mathrm{GeV}^2 corresponds to the single non-diagonal kinematics point and has a correspondingly larger uncertainty in the moment.

The mid Q2Q^2 region, 0.5≲Q2≲4 GeV20.5\lesssim Q^2\lesssim4\ \mathrm{GeV}^2, is significant for investigating higher-twist contributions as the power corrections go as Q−2Q^{-2}. However simultaneously in this region, the g1g_1 structure function becomes dominated by the elastic pole, as discussed in Section IV C. Indeed, we see the lattice calculations deviate from the perturbative prediction for the up quark as Q2Q^2 decreases. Upon addition of the elastic contribution to the perturbative result, shown by the solid yellow bands in Figure 3, we begin to capture this departure from the pure perturbative result. The discrepancy between the calculated moments and yellow bands is indicative of higher-twist contributions, however given some deviation is also present at large Q2Q^2, full control over discretisation artifacts is necessary before a rigorous calculation of the higher-twist term. With our limited kinematic choices, we are only able to calculate moments with q=(0,q2,0)\mathbf{q}=(0, q_2, 0), with our lowest transfer momentum being q=(0,2,0)×2πL\mathbf{q}=(0,2,0)\times\frac{2\pi}{L}. Higher resolution in Q2Q^2, particularly in this mid Q2Q^2 region, would further assist in the extraction of the higher-twist term.

The Bjorken Sum Rule

Recall the Bjorken sum rule is given by

Γ1p−n(Q2)=∫01dx [g1p(x,Q2)−g1n(x,Q2)],(37)\Gamma_1^{p-n}(Q^2)=\int_0^1 dx\,[g_1^p(x,Q^2)-g_1^n(x,Q^2)], \tag*{(37)}

and is related to the calculated quark-level moments by

Γ1p−n(Q2)=13[M~1,uu(1)(Q2)−M~1,dd(1)(Q2)].(38)\Gamma_1^{p-n}(Q^2)=\frac{1}{3}\left[\widetilde{M}^{(1)}_{1,uu}(Q^2)-\widetilde{M}^{(1)}_{1,dd}(Q^2)\right]. \tag*{(38)}

We show our LOPE-improved lattice results in Figure 4, compared with the QPM prediction, perturbative QCD and elastic contributions. As with the quark-level results, we divide our discussion into two regions in Q2Q^2.

Figure 4. Lattice QCD calculation of the Bjorken sum rule, as a function of Q2Q^2. Also plotted is the twist-two prediction for the Bjorken sum rule, gAu−d6\frac{g_A^{u-d}}{6}, the N3^3LO perturbative corrections to the twist-two prediction and N3^3LO perturbative QCD result plus elastic contribution. gAu−dg_A^{u-d} is calculated on the same set of gauge configurations.

In the high Q2Q^2 region, Q2≳4 GeV2Q^2\gtrsim4\ \mathrm{GeV}^2, the moments are dominated by the leading-twist parton distributions and their associated corrections from perturbative QCD. Combining Equations (35) and (38), we have the quark parton model (QPM) prediction for the Bjorken sum rule,

Γ1p−n,QPM(Q2)→Q2→∞gAu−d6.(39)\Gamma_1^{p-n,\mathrm{QPM}}(Q^2)\xrightarrow{Q^2\to\infty}\frac{g_A^{u-d}}{6}. \tag*{(39)}

where we have identified the isovector axial charge gAu−d=Δu−Δdg_A^{u-d}=\Delta u-\Delta d. We plot our calculated gAu−dg_A^{u-d} along with its statistical uncertainty as the grey band in Figure 4. We note that our calculation of gAu−dg_A^{u-d} with three-point function methods is consistent with a previous Feynman-Hellmann calculation on the same lattice [16]. We see our calculated moments lie below the QPM prediction, even at our largest Q2Q^2 value. Applying the perturbative corrections to the QPM, we have the perturbative QCD prediction for the Bjorken sum rule,

Γ1p−n(Q2)=gAu−d6[1−∑i=14Ciasi(Q2)]+ΔHTQ2+⋯(40)\Gamma_1^{p-n}(Q^2) = \frac{g_A^{u-d}}{6}\left[1-\sum_{i=1}^{4} C_i a_s^i(Q^2)\right] + \frac{\Delta_{HT}}{Q^2} + \cdots \tag*{(40)}

where the CiC_i terms are the same as in Equation (36). The perturbative prediction is plotted as the solid green band in Figure 4. We find our moments follow the qualitative behaviour of the perturbative result at large Q2Q^2, however are offset due to discretisation effects. In principle, the calculated moments can be used to extract an effective strong coupling in this large Q2Q^2 region [32–34]. As both gAu−dg_A^{u-d} and Γp−n\Gamma^{p-n} are calculable with lattice QCD, the deviation between these two quantities can be used to define an effective strong coupling, αseff\alpha_s^{\mathrm{eff}}, derived from a hadronic quantity. We leave such a study for future work, after a continuum and chiral extrapolation of the lattice results.

In the mid Q2Q^2 region, 0.5 GeV2≲Q2≲4 GeV20.5\ \mathrm{GeV}^2 \lesssim Q^2 \lesssim4\ \mathrm{GeV}^2, we begin to see the impact of the elastic contributions, causing the lattice moments to increase rapidly with decreasing Q2Q^2. Upon addition of the elastic contribution to the perturbative curve, we see the lattice moments more closely follow this curve, however with some offset. The offset in this region indicates the presence of higher-twist corrections, however given the discrepancy at large Q2Q^2, discretisation artifacts likely also contribute. This motivates further investigation into the lattice-spacing dependence of our calculated moments, which we leave as future work. With discretisation artifacts under control, higher-twist corrections, ΔHT\Delta_{HT}, could then be reliably estimated. Model calculations [73, 74, 75, 76] and experimental results [77, 78, 79] suggest these higher-twist corrections are small. It is worth noting that the calculation of higher-twist effects is intimately linked to the divergence of the perturbative series, and that renormalon ambiguity resembling ΛQCD2/Q2\Lambda_{QCD}^2/Q^2 behaviour [80, 81, 82, 83] further complicates a rigorous estimate of ΔHT\Delta_{HT}.

Summary

In this work, we have performed a lattice QCD calculation of the first moment of the nucleon’s polarised g1g_1 structure function via a Feynman-Hellmann calculation of the polarised forward Compton amplitude. Our calculations were performed on a single lattice ensemble with volume 483×9648^3 \times96, lattice spacing a=0.068a = 0.068 fm and pion mass tuned to the SU(3) symmetric point, mπ≈412m_\pi\approx412 MeV. Our results cover the 0.5≲Q2≲80.5 \lesssim Q^2 \lesssim8 GeV2^2 range, which includes both perturbative and non-perturbative regions. We expand our Feynman-Hellmann analysis to account for perturbations which are not diagonal in the nucleon spins through the use of a GEVP. Additionally, we utilised a LOPE-based parameterisation of the Compton amplitude to reduce the impact of discretisation artifacts on our results. We assessed the agreement of our moments calculated non-perturbatively on the lattice with the perturbative evolution of the leading-twist predictions for the Bjorken sum rule. Our results show deviation from the perturbative QCD predictions in the low Q2Q^2 region, hinting at non-negligible higher-twist contributions. A calculation with improved precision, together with several lattice spacings and lighter quark masses, would permit a study of the 1/Q21/Q^2 power corrections after subtraction of the elastic contribution. A direct determination of αs\alpha_s from the Bjorken sum rule will require a continuum and quark-mass extrapolation together with a controlled treatment of higher-twist and other finite-Q2Q^2 effects. The method presented in this work is immediately generalisable to finite nucleon momentum, allowing for the full ω\omega dependence of the polarised structure functions to be calculated. From this ω\omega dependence, higher moments of the structure functions could be accessed.

Nevertheless, this work demonstrates the feasibility of determining a non-singlet DIS moment from the lattice using the Feynman–Hellmann method.

Acknowledgments

The numerical configuration generation (using the BQCD lattice QCD program [84]) and data analysis (using the Chroma software library [85]) was carried out on the Extreme Scaling Service (DiRAC, EPCC, Edinburgh, UK), the Data Intensive Service (DiRAC, CS3D, Cambridge, UK), the Gauss Centre for Supercomputing (NIC, Jülich, Germany), the NHR Alliance (Germany), and resources provided by the NCI National Facility in Canberra, Australia (supported by the Australian Commonwealth Government), the Pawsey Supercomputing Centre (supported by the Australian Commonwealth Government and the Government of Western Australia), and the Phoenix HPC service (Adelaide University). PELR is supported in part by the STFC under contract ST/G00062X/1. GS is supported by DFG Grant SCHI 179/8-1. JAC and TGS are supported by an Australian Government Research Training Program (RTP) Scholarship. JAC partially received funding from the European Research Council (ERC) via the project “LEEX” grant agreement 101170304, funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council Executive Agency (ERCEA). Neither the European Union nor the ERCEA can be held responsible for them. RDY and JMZ are supported by the Australian Research Council grants DP190100297, DP220103098, and DP240102839.

Spin Structure of the Compton Amplitude

In this appendix, we provide more details on the diagonalisation of the Compton amplitude and the required kinematic choices for the diagonal and non-diagonal cases. Starting from Hamiltonian perturbation theory, our perturbed action introduces λ\lambda-dependent perturbations to the Hamiltonian,

H^λ=H^0+λH^1+λ2H^2+O(λ3),(41)\hat{H}_{\lambda} = \hat{H}_{0} + \lambda\hat{H}_{1} + \lambda^{2}\hat{H}_{2} + O(\lambda^{3}), \tag*{(41)}

where H^0\hat{H}_{0} is the unperturbed QCD Hamiltonian, H^0∣p,s⟩=EN(p)∣p,s⟩\hat{H}_{0}|p,s\rangle= E_{N}(p)|p,s\rangle, and the terms higher order in λ\lambda are related to matrix elements of the perturbing current via the Feynman-Hellmann theorem. As we consider forward matrix elements in this work, terms proportional to λ\lambda will not contribute, as each insertion of the current carries momentum q\mathbf{q}. Therefore, our energy splitting appears only at second order in λ\lambda. Using the perturbation to the QCD action described in Equation (11), the perturbed energy levels are given by

ENλ(p,q)=EN(p)+λ122∂2ENλ(p,q)∂λ12+λ222∂2ENλ(p,q)∂λ22+λ1λ2∂2ENλ(p,q)∂λ1∂λ2(42)E_{N}^{\lambda}(\mathbf{p},\mathbf{q}) = E_{N}(\mathbf{p}) + \frac{\lambda_{1}^{2}}{2}\frac{\partial^{2}E_{N}^{\lambda}(\mathbf{p},\mathbf{q})}{\partial\lambda_{1}^{2}} + \frac{\lambda_{2}^{2}}{2}\frac{\partial^{2}E_{N}^{\lambda}(\mathbf{p},\mathbf{q})}{\partial\lambda_{2}^{2}} + \lambda_{1}\lambda_{2}\frac{\partial^{2}E_{N}^{\lambda}(\mathbf{p},\mathbf{q})}{\partial\lambda_{1}\partial\lambda_{2}} \tag*{(42)}

Using the Feynman-Hellmann relation for repeated insertions of the same current [36],

∂2ENλ(p,q)∂λ2=−Tμμ(p,q)−Tμμ(p,−q)2EN(p),(43)\frac{\partial^{2}E_{N}^{\lambda}(\mathbf{p},\mathbf{q})}{\partial\lambda^{2}} = -\frac{T_{\mu\mu}(\mathbf{p},\mathbf{q}) - T_{\mu\mu}(\mathbf{p},-\mathbf{q})}{2E_{N}(\mathbf{p})}, \tag*{(43)}

the first two perturbations are proportional to the unpolarised Compton amplitude and are hence diagonal in the space of nucleon spins. In contrast, using Equation (12), the mixed derivative term is proportional to the polarised Compton amplitude. To isolate G1(ω,Q2)G_{1}(\omega,Q^{2}), we choose the indices μ=1\mu= 1 and ν=4\nu= 4 of the Compton amplitude [Tμν(p,q)]σ′σ[T_{\mu\nu}(p,q)]_{\sigma'\sigma}. From Equation (5), with the additional constraint of considering a nucleon at rest p=(0,0,0)\mathbf{p}=(0,0,0) and noting q4=0q_{4}=0 in the Feynman-Hellmann approach, this component of the Compton amplitude in Euclidean space is

[T14(p,q)]σ′σ=−2iQ2ϵ1ijqiσjσ′σ(G1(ω,Q2)ω∣ω=0+G2(ω,Q2)ω∣ω=0),(44)[T_{14}(p,q)]_{\sigma'\sigma} = -\frac{2i}{Q^{2}}\epsilon_{1ij}q_{i}\sigma^{\sigma'\sigma}_{j}\left(\left.\frac{G_{1}(\omega,Q^{2})}{\omega}\right|_{\omega=0}+\left.\frac{G_{2}(\omega,Q^{2})}{\omega}\right|_{\omega=0}\right), \tag*{(44)}

where ϵ123=+1\epsilon_{123}=+1. As discussed in the main text, we assume the contribution from the G2(ω,Q2)/ω∣ω=0G_{2}(\omega,Q^{2})/\omega|_{\omega=0} term is zero due to the Burkhardt-Cottingham sum rule. Evidently, by the diagonal nature of σ3\sigma^{3}, this component of the Compton amplitude remains diagonal in the space of nucleon spins for current momentum of the form q=(0,q2,0)\mathbf{q}=(0,q_{2},0). For more general current momentum, the off-diagonal nature of the σ1\sigma^{1} and σ2\sigma^{2} matrices causes a non-trivial mixing of states. The final term of Equation (42) is, with added spin indices,

[∂2ENλ(p,q)∂λ1∂λ2]σ′σ=2Q2ϵ1ijqiσjσ′σG1(ω,Q2)ω∣ω=0.(45)\left[\frac{\partial^{2}E_{N}^{\lambda}(\mathbf{p},\mathbf{q})}{\partial\lambda_{1}\partial\lambda_{2}}\right]_{\sigma'\sigma} = \left.\frac{2}{Q^{2}}\epsilon_{1ij}q_{i}\sigma_{j}^{\sigma'\sigma}\frac{G_{1}(\omega,Q^{2})}{\omega}\right|_{\omega=0}. \tag*{(45)}

We package the kinematic factors and structure function into a coefficient vjv_{j},

vj=2Q2ϵ1ijqiG1(ω,Q2)ω∣ω=0,(46)v_{j} = \left.\frac{2}{Q^{2}}\epsilon_{1ij}q_{i}\frac{G_{1}(\omega,Q^{2})}{\omega}\right|_{\omega=0}, \tag*{(46)}

such that this energy shift can be compactly written as

[∂2ENλ(p,q)∂λ1∂λ2]σ′σ=v⋅σσ′σ.(47)\left[\frac{\partial^{2}E_{N}^{\lambda}(\mathbf{p},\mathbf{q})}{\partial\lambda_{1}\partial\lambda_{2}}\right]_{\sigma'\sigma} = \mathbf{v}\cdot\boldsymbol{\sigma}_{\sigma'\sigma}. \tag*{(47)}

The perturbed Hamiltonian acting on the space of nucleon spins is

H^λ=H^0+ΔEunpol(λ1,λ2)δσ′σ+λ1λ2v⋅σσ′σ,(48)\hat{H}_{\lambda} = \hat{H}_{0} + \Delta E_{\mathrm{unpol}}(\lambda_{1},\lambda_{2})\delta_{\sigma'\sigma} + \lambda_{1}\lambda_{2}\mathbf{v}\cdot\boldsymbol{\sigma}_{\sigma'\sigma}, \tag*{(48)}

where ΔEunpol(λ1,λ2)δσ′σ\Delta E_{\mathrm{unpol}}(\lambda_{1},\lambda_{2})\delta_{\sigma'\sigma} captures the energy shifts from the unpolarised Compton amplitude. Diagonalising this Hamiltonian to determine the energy eigenvalues, we find

Eλ(±)=EN(p)+ΔEunpol(λ1,λ2)±λ1λ2v⋅v.(49)E_{\lambda}^{(\pm)} = E_{N}(\mathbf{p}) + \Delta E_{\mathrm{unpol}}(\lambda_{1},\lambda_{2}) \pm\lambda_{1}\lambda_{2}\sqrt{\mathbf{v}\cdot\mathbf{v}}. \tag*{(49)}

Elastic Contributions

The elastic contribution to the nucleon polarised structure function g1g_{1} is given by

g1,fel(x,Q2)=12F1(f)(Q2)GM(f)(Q2)δ(1−x),(50)g_{1,f}^{\mathrm{el}}(x,Q^{2}) = \frac{1}{2}F_{1}^{(f)}(Q^{2})G_{M}^{(f)}(Q^{2})\delta(1-x), \tag*{(50)}

where GM(Q2)G_{M}(Q^{2}) is the Sachs magnetic form factor,

GM(Q2)=F1(Q2)+F2(Q2),(51)G_{M}(Q^{2}) = F_{1}(Q^{2}) + F_{2}(Q^{2}), \tag*{(51)}

with F1(Q2)F_{1}(Q^{2}) and F2(Q2)F_{2}(Q^{2}) being the electromagnetic form factors. Our calculation of the form factors follows the conventional three-point function approach, and we detail the main steps for completeness.

The nucleon matrix element of the electromagnetic current is decomposed in terms of the Dirac (F1F_{1}) and Pauli (F2F_{2}) form factors,

⟨p′,s′∣Jμ(0)∣p,s⟩=u‾(p′,s′)[γμF1(Q2)+σμνqν2mNF2(Q2)]u(p,s),(52)\langle p',s'|J_{\mu}(0)|p,s\rangle= \overline{u}(p',s')\left[\gamma_{\mu}F_{1}(Q^{2})+\frac{\sigma_{\mu\nu}q_{\nu}}{2m_{N}}F_{2}(Q^{2})\right]u(p,s), \tag*{(52)}

where u(p,s)u(p,s) is a Dirac spinor with 4-momentum pp and spin ss, q=p′−pq=p'-p is the transfer momentum, Q2=−q2Q^{2}=-q^{2}, and σμν=i2[γμ,γν]\sigma_{\mu\nu} = \frac{i}{2}[\gamma_\mu,\gamma_\nu]. The electromagnetic current is given by

Jμ=23uˉγμu−13dˉγμd+…(53)J_\mu= \frac{2}{3}\bar{u}\gamma_\mu u - \frac{1}{3}\bar{d}\gamma_\mu d + \ldots \tag*{(53)}

The elastic form factors are calculated from ratios of three- and two-point correlation functions. For the Sachs magnetic form factor, we insert the local operators,

jμ=ψˉγμψ,for μ=1,2,3,4.(54)j_\mu= \bar{\psi}\gamma_\mu\psi,\qquad\text{for }\mu=1,2,3,4. \tag*{(54)}

Matrix elements of these operators are calculated from three-point functions,

G(3)(p′,t′;q,τ;O)=∑x1,x2e−ip′⋅x2eiq⋅x1Γαβ×⟨χˉ(x2,t′)O(x1,τ)χ(0,0)⟩,(55)G^{(3)}(\mathbf{p}',t';\mathbf{q},\tau;\mathcal{O}) = \sum_{\mathbf{x}_1,\mathbf{x}_2} e^{-i\mathbf{p}'\cdot\mathbf{x}_2}e^{i\mathbf{q}\cdot\mathbf{x}_1}\Gamma_{\alpha\beta}\times \langle\bar{\chi}(\mathbf{x}_2,t')\mathcal{O}(\mathbf{x}_1,\tau)\chi(0,0)\rangle, \tag*{(55)}

where χ\chi are the standard nucleon interpolating operators. The local operator O\mathcal{O} is inserted at timeslice τ\tau such that 0<τ<t′0 < \tau< t', where t′t' is the sink time. Γ\Gamma is a spin-parity projector, with Γunpol=12(I+γ4)\Gamma_{\mathrm{unpol}} = \frac{1}{2}(I+\gamma_4) and Γi=−i2(1+γ4)γiγ5\Gamma_i = -\frac{i}{2}(1+\gamma_4)\gamma_i\gamma_5, with i=1,2,3i=1,2,3. For this analysis, we neglect disconnected contributions and only consider connected contributions to the three-point function. We construct the following ratio of three- and two-point functions,

R(p′,t′;q,τ;O)=G(3)(p′,t′;q,τ;O)G(2)(p′,t′)[G(2)(p′,t′)G(2)(p′,τ)G(2)(p,t′−τ)G(2)(p,t′)G(2)(p,τ)G(2)(p′,t′−τ)]12,(56)\mathcal{R}(\mathbf{p}',t';\mathbf{q},\tau;\mathcal{O}) = \frac{G^{(3)}(\mathbf{p}',t';\mathbf{q},\tau;\mathcal{O})}{G^{(2)}(\mathbf{p}',t')} \left[\frac{G^{(2)}(\mathbf{p}',t')G^{(2)}(\mathbf{p}',\tau)G^{(2)}(\mathbf{p},t'-\tau)}{G^{(2)}(\mathbf{p},t')G^{(2)}(\mathbf{p},\tau)G^{(2)}(\mathbf{p}',t'-\tau)}\right]^{\frac{1}{2}}, \tag*{(56)}

which, assuming ground-state dominance at large Euclidean times, is proportional to the matrix element of the inserted operator. To control excited-state contamination, we use a two-exponential ansatz to model the three-point function when fitting directly to the ratio. The two-point functions are fit beforehand, also using a two-exponential ansatz, to determine the ground-state energy ENE_N and energy gap ΔEN\Delta E_N,

G(2)(p,t)=A2,0(p)e−EN(p)t(1+A2,1(p)e−ΔEN(p)t),(57)G^{(2)}(\mathbf{p},t) = A_{2,0}(\mathbf{p})e^{-E_N(\mathbf{p})t}\left(1+A_{2,1}(\mathbf{p})e^{-\Delta E_N(\mathbf{p})t}\right), \tag*{(57)}

where A2,0A_{2,0} and A2,1A_{2,1} are overlap factors. The three-point function ansatz is

G(3)(p′,t′;q,τ;O)=A3,0e−EN(p′)(t′−τ)e−EN(p)τ[B00+B10e−ΔEN(p′)(t′−τ)+B01e−ΔEN(p)τ+B11e−ΔEN(p′)(t′−τ)e−ΔEN(p)τ],(58)G^{(3)}(\mathbf{p}',t';\mathbf{q},\tau;\mathcal{O}) = A_{3,0}e^{-E_N(\mathbf{p}')(t'-\tau)}e^{-E_N(\mathbf{p})\tau} \left[B_{00}+B_{10}e^{-\Delta E_N(\mathbf{p}')(t'-\tau)}+B_{01}e^{-\Delta E_N(\mathbf{p})\tau} +B_{11}e^{-\Delta E_N(\mathbf{p}')(t'-\tau)}e^{-\Delta E_N(\mathbf{p})\tau}\right], \tag*{(58)}

where A3,0A_{3,0} is an overlap factor which cancels in the ratio B7, and B00B_{00} is proportional to the desired matrix element,

B00=[EN(p)EN(p′)(EN(p)+mN)(EN(p′)+mN)]12⟨p′,s′∣jμ∣p,s⟩.(59)B_{00} = \left[\frac{E_N(\mathbf{p})E_N(\mathbf{p}')}{(E_N(\mathbf{p})+m_N)(E_N(\mathbf{p}')+m_N)}\right]^{\frac{1}{2}}\langle\mathbf{p}',s'|j_\mu|\mathbf{p},s\rangle. \tag*{(59)}

The terms B01B_{01} and B10B_{10} are proportional to transition matrix elements between the ground state and first excited state (and vice versa), while B11B_{11} is proportional to the matrix element between two nucleons in the first excited state. A representative plot of the ratio used to determine GMG_M is shown in Figure 5.

Figure 5. Euclidean time dependence of the ratio in Equation (56) for three source-sink separations. The inserted operator is ψˉγ3ψ\bar{\psi}\gamma_3\psi, with a sink momentum p′=(1,0,0)\mathbf{p}'=(1,0,0) and source momentum p=(2,0,0)\mathbf{p}=(2,0,0), yielding Q2=0.12 GeV2Q^2=0.12\ \mathrm{GeV}^2. The three-point function is projected using the projector Γ3\Gamma_3, and the resulting ratio is proportional to GMG_M.

Repeating this analysis across a range of Q2Q^2 values allows us to determine the Q2Q^2 dependence of F1F_1 and GMG_M. We renormalise our form factors in the RI′–MOM scheme by computing the renormalisation factor

ZV(u)=2F1(u)(0),orZV(d)=1F1(d)(0),(60)Z_V^{(u)} = \frac{2}{F_1^{(u)}(0)},\qquad\text{or}\qquad Z_V^{(d)} = \frac{1}{F_1^{(d)}(0)}, \tag*{(60)}

for the up and down quarks respectively. We plot the resulting points in Figure 6. We model the Q2Q^2 dependence of the form factors using a dipole parameterisation, given by

Figure 6. The Q2Q^2 dependence of the form factors F1(Q2)F_1(Q^2) and GM(Q2)G_M(Q^2), as well as the elastic contribution to the g1g_1 structure function (Eq. (B1)). We show results for the uu- (blue circles) and dd-quark (red triangles) form factors, and the uu- (blue solid line) and dd-quark (red dashed line) elastic contribution g1elg_1^{\mathrm{el}}. g1elg_1^{\mathrm{el}} is extrapolated over the whole Q2Q^2 range using the dipole parameterisation from Eq. (B12).

F(Q2)=F(0)(1−Q2Λ2)2,(61)F(Q^2) = \frac{F(0)}{\left(1-\frac{Q^2}{\Lambda^2}\right)^2}, \tag*{(61)}

where F(0)F(0) and Λ\Lambda are fit parameters. We make use of this parameterisation to extrapolate these form factors beyond the Q2Q^2 range accessible via our lattice calculation. In Figure 6, we plot the elastic g1g_1 structure function across the whole Q2Q^2 range used in our analysis. For Q2≳4 GeV2Q^2 \gtrsim4\ \mathrm{GeV}^2, the elastic contribution to the structure function is practically negligible.

Calculation of gAu−dg_A^{u-d}

The contribution of quark flavour q to the total hadron spin, Δq\Delta q, is related to matrix elements of the axial current as follows,

⟨p,s∣ψˉq(0)γμγ5ψq(0)∣p,s⟩=2sμΔq,(62)\langle p,s|\bar{\psi}_q(0)\gamma_\mu\gamma_5\psi_q(0)|p,s\rangle= 2s_\mu\Delta q, \tag*{(62)}

where sμs_\mu is the spin four-vector. The isovector matrix element gAu−dg_A^{u-d} is then gAu−d=Δu−Δdg_A^{u-d}=\Delta u-\Delta d.

As with the electromagnetic form factors, we compute gAu−dg_A^{u-d} using three-point function methods. We insert the local operators

jA=ψˉγμγ5ψ,μ=1,2,3,4,(63)j_A=\bar{\psi}\gamma_\mu\gamma_5\psi,\qquad\mu=1,2,3,4, \tag*{(63)}

and compute the three-point function via sequential inversion through the sink. We consider connected contributions to the three-point function only, and extract the ground state contribution as described in Appendix B. Representative plots of the ratios used to extract Δqf\Delta q_f for the up and down quarks are shown in Figure 7. This yields the bare results for each quark flavour of Δu=0.958(15)stat\Delta u=0.958(15)_{\mathrm{stat}} and Δd=−0.299(57)stat\Delta d=-0.299(57)_{\mathrm{stat}}. The bare isovector combination is obtained by taking the difference of the individual quark results, yielding gAu−d=1.258(15)statg_A^{u-d}=1.258(15)_{\mathrm{stat}}. The RI′-MOM renormalisation constant for the axial current has previously been calculated for this lattice and takes the value ZA=0.8754(19)statZ_A=0.8754(19)_{\mathrm{stat}} [47]. The renormalised value for gAu−dg_A^{u-d} is then gAu−d=1.101(13)statg_A^{u-d}=1.101(13)_{\mathrm{stat}}, which is the value appearing in the main text.

Figure 7. Calculation of Δq\Delta q for both the up (left) and down (right) quarks, using the operator ψˉq(0)γ1γ5ψq(0)\bar{\psi}_q(0)\gamma_1\gamma_5\psi_q(0) and projector Γproj=Γ1\Gamma_{\mathrm{proj}}=\Gamma_1. The horizontal axis is shown in units of aa.

References

  1. [1]J. D. Bjorken, Inelastic scattering of polarized leptons from polarized nucleons, Phys. Rev. D 1, 1376 (1970).
  2. [2]P. Anthony et al., Measurements of the q²-dependence of the proton and neutron spin structure functions g1p and g1n, Physics Letters B 493, 19 (2000), arXiv:hep-ph/0007248.DOI
  3. [3]A. Deur et al., Experimental determination of the evolution of the Bjorken integral at low Q², Phys. Rev. Lett. 93, 212001 (2004), arXiv:hep-ex/0407007.DOI
  4. [4]A. Deur, P. Bosted, V. Burkert, D. Crabb, V. Dharmawardane, G. E. Dodge, T. A. Forest, K. A. Griffioen, S. E. Kuhn, R. Minehart, and Y. Prok., Experimental study of isovector spin sum rules, Phys. Rev. D 78, 032001 (2008), arXiv:0802.3198 [nucl-ex].DOI
  5. [5]A. Deur, Y. Prok, V. Burkert, D. Crabb, F.-X. Girod, K. A. Griffioen, N. Guler, S. E. Kuhn, and N. Kval- tine, High precision determination of the Q² evolution of the bjorken sum, Phys. Rev. D 90, 012009 (2014), arXiv:1405.7854 [nucl-ex].DOI
  6. [6]R. G. Fersch et al. (CLAS Collaboration), Determination of the proton spin structure functions for 0.05 < Q² < 5 GeV² using CLAS, Phys. Rev. C 96, 065208 (2017), arXiv:1706.10289 [nucl-ex].DOI
  7. [7]K. P. Adhikari et al. (CLAS Collaboration), Measurement of the Q² dependence of the deuteron spin structure function g1 and its moments at low Q² with clas, Phys. Rev. Lett. 120, 062501 (2018), arXiv:1711.01974 [nucl-ex].DOI
  8. [52]D 84, 054509 (2011), arXiv:hep-lat/1102.5300.
  9. [53]F. Hagelstein, R. Miskimen, and V. Pascalutsa, Nucleon Polarizabilities: from Compton Scattering to Hydrogen Atom, Prog. Part. Nucl. Phys. 88, 29 (2016), arXiv:1512.03765 [nucl-th].DOI
  10. [54]M. Göckeler, R. Horsley, H. Perlt, P. E. L. Rakow, G. Schierholz, and A. Schiller (QCDSF), Operator product expansion on the lattice: Analytic Wilson coefficients, PoS LAT2006, 119 (2006), arXiv:hep-lat/0610064.arxiv.org/abs/hep-lat/0610064
  11. [55]K. U. Can, A. Hannaford-Gunn, R. Horsley, P. E. L. Rakow, T. Schar, G. Schierholz, H. Stüben, R. D. Young, and J. M. Zanotti (QCDSF), Lattice QCD calculation of the Compton amplitude subtraction function, Phys. Rev. D 111, 094513 (2025), arXiv:2501.08619 [hep-lat].DOI
  12. [56]T. Schar, Lattice Discretisation Artefacts in the Compton Amplitude, Ph.D. thesis, Adelaide U. (2025).
  13. [57]J. R. Ellis and R. L. Jaffe, A Sum Rule for Deep Inelastic Electroproduction from Polarized Protons, Phys. Rev. D 9, 1444 (1974), [Erratum: Phys.Rev.D 10, 1669 (1974)].DOI
  14. [58]A. L. Kataev, Ellis-jaffe sum rule: The estimates of the next-to-next-to-leading-order qcd corrections, Phys. Rev. D 50, R5469 (1994).pubmed.ncbi.nlm.nih.gov/10018261
  15. [59]A. L. Kataev, Deep inelastic sum rules at the boundaries between perturbative and nonperturbative QCD, Mod. Phys. Lett. A 20, 2007 (2005), arXiv:hep-ph/0505230.
  16. [60]P. A. Baikov, K. G. Chetyrkin, and J. H. Kühn, Adler Function, Bjorken Sum Rule, and the Crewther Relation to Order α_s^4 in a General Gauge Theory, Phys. Rev. Lett. 104, 132004 (2010), arXiv:1001.3606 [hep-ph].
  17. [61]Y. Aoki et al. (Flavour Lattice Averaging Group (FLAG)), FLAG review 2024, Phys. Rev. D 113, 014508 (2026), arXiv:2411.04268 [hep-lat].
  18. [62]K. Maltman, D. Leinweber, P. Moran, and A. Sternbeck, The Realistic Lattice Determination of alpha(s)(M(Z)) Revisited, Phys. Rev. D 78, 114504 (2008), arXiv:0807.2020 [hep-lat].DOI
  19. [63]S. Aoki et al. (PACS-CS), Precise determination of the strong coupling constant in N_f = 2+1 lattice QCD with the Schrödinger functional scheme, JHEP 10, 053, arXiv:0906.3906 [hep-lat].
  20. [64]C. McNeile, C. T. H. Davies, E. Follana, K. Hornbostel, and G. P. Lepage, High-Precision c and b Masses, and QCD Coupling from Current-Current Correlators in Lattice and Continuum QCD, Phys. Rev. D 82, 034512 (2010), arXiv:1004.4285 [hep-lat].
  21. [65]B. Chakraborty, C. T. H. Davies, B. Galloway, P. Knecht, J. Koponen, G. C. Donald, R. J. Dowdall, G. P. Lepage, and C. McNeile, High-precision quark masses and QCD coupling from n_f = 4 lattice QCD, Phys. Rev. D 91, 054508 (2015), arXiv:1408.4169 [hep-lat].
  22. [66]M. Bruno, M. Dalla Brida, P. Fritzsch, T. Korzec, A. Ramos, S. Schaefer, H. Simma, S. Sint, and R. Sommer (ALPHA), QCD Coupling from a Nonperturbative Determination of the Three-Flavor A Parameter, Phys. Rev. Lett. 119, 102001 (2017), arXiv:1706.03821 [hep-lat].
  23. [67]A. Bazavov, N. Brambilla, X. Garcia i Tormo, P. Petreczky, J. Soto, A. Vairo, and J. H. Weber (TUMQCD), Determination of the QCD coupling from the static energy and the free energy, Phys. Rev. D 100, 114511 (2019), arXiv:1907.11747 [hep-lat].DOI
  24. [68]C. Ayala, X. Lobregat, and A. Pineda, Determination of α(M_Z) from an hyperasymptotic approximation to the energy of a static quark-antiquark pair, JHEP 09, 016, arXiv:2005.12301 [hep-ph].
  25. [69]S. Cali, K. Cichy, P. Korcyl, and J. Simeth, Running coupling constant from position-space current-current correlation functions in three-flavor lattice QCD, Phys. Rev. Lett. 125, 242002 (2020), arXiv:2003.05781 [hep-lat].DOI
  26. [70]P. Petreczky and J. H. Weber, Strong coupling constant from moments of quarkonium correlators revisited, Eur. Phys. J. C 82, 64 (2022), arXiv:2012.06193 [hep-lat].DOI
  27. [71]K. G. Chetyrkin, J. H. Kühn, and M. Steinhauser, RunDec: A Mathematica package for running and decoupling of the strong coupling and quark masses, Comput. Phys. Commun. 133, 43 (2000), arXiv:hep-ph/0004189.DOI
  28. [72]F. Herren and M. Steinhauser, Version 3 of RunDec and CRunDec, Comput. Phys. Commun. 224, 333 (2018), arXiv:1703.03751 [hep-ph].DOI
  29. [73]J. Balla, M. V. Polyakov, and C. Weiss, Nucleon matrix elements of higher twist operators from the instanton vacuum, Nucl. Phys. B 510, 327 (1998), arXiv:hep-ph/9707515.DOI
  30. [74]I. I. Balitsky, V. M. Braun, and A. V. Kolesnichenko, Power corrections 1 / Q**2 to parton sum rules for deep inelastic scattering from polarized targets, Phys. Lett. B 242, 245 (1990), [Erratum: Phys.Lett.B 318, 648 (1993)], arXiv:hep-ph/9310316.
  31. [75]E. Stein, P. Gornicki, L. Mankiewicz, and A. Schafer, QCD sum rule calculation of twist four corrections to Bjorken and Ellis-Jaffe sum rules, Phys. Lett. B 353, 107 (1995), arXiv:hep-ph/9502323.
  32. [76]B. Ehrnsperger, A. Schafer, and L. Mankiewicz, OPE analysis for polarized deep inelastic scattering, Phys. Lett. B 323, 439 (1994), arXiv:hep-ph/9311285.DOI
  33. [77]X.-D. Ji and W. Melnitchouk, Spin dependent twist four matrix elements from g_1 data in the resonance region, Phys. Rev. D 56, R1 (1997), arXiv:hep-ph/9703363.
  34. [78]Q. Yu, X.-G. Wu, H. Zhou, and X.-D. Huang, A novel determination of non-perturbative contributions to Bjorken sum rule, Eur. Phys. J. C 81, 690 (2021), arXiv:2102.12771 [hep-ph].DOI
  35. [79]C. Cocuzza, N. T. Hunt-Smith, W. Melnitchouk, N. Sato, and A. W. Thomas (JAM Collaboration (Spin PDF Analysis Group)), Global QCD analysis of spin PDFs in the proton with high-x and lattice constraints, Phys. Rev. D 112, 114017 (2025), arXiv:2506.13616 [hep-ph].arxiv.org/abs/2506.13616
  36. [80]A. Mueller, Combining higher twist terms with finite order perturbative contributions, Physics Letters B 308, 355 (1993).DOI
  37. [81]G. Martinelli and C. T. Sachrajda, On the difficulty of computing higher twist corrections, Nucl. Phys. B 478, 660 (1996), arXiv:hep-ph/9605336.DOI
  38. [82]M. Beneke and V. M. Braun, Renormalons and power corrections, in At The Frontier of Particle Physics (World Scientific, 2001) p. 1719–1773, arXiv:hep-ph/0010208.DOI
  39. [83]V. M. Braun, Higher Twists, EPJ Web Conf. 274, 01012 (2022), arXiv:2212.02887 [hep-ph].DOI
  40. [84]T. R. Haar, Y. Nakamura, and H. Stüben, An update on the BQCD Hybrid Monte Carlo program, EPJ Web Conf. 175, 14011 (2018), arXiv:1711.03836 [hep-lat].DOI
  41. [85]R. G. Edwards and B. Joo (SciDAC, LHPC, UKQCD), The Chroma software system for lattice QCD, Nucl. Phys. B Proc. Suppl. 140, 832 (2005), arXiv:hep-lat/0409003.DOI

Paper details

vlnc:W_4fm37u8f
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

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