betaFeedback

WorksPhysics and Astronomy

Twist-2 relations for the twist-3 tensor-polarized distribution function fLTf_{LT} of a spin-1 hadron by the operator-product-expansion method

Abstract

In a spin-1 hadron, tensor-polarized parton distribution functions (PDFs) exist. The twist-2 function is f1LLf_{1LL} and a twist-3 one is fLTf_{LT}. Because an experiment is under preparation at the Thomas Jefferson National Accelerator Facility (JLab) to measure the cross section of electron-deuteron deep inelastic scattering with the tensor-polarized deuteron target, these PDFs need to be understood theoretically. Especially, measurements will be done in a relatively low-Q2Q^2 region at JLab, so that twist-3 contributions could become sizable in the cross section. In a previous work, a twist-2 relation was derived for fLTf_{LT} in terms of f1LLf_{1LL} by using a nonlocal operator, and it corresponds to the Wandzura-Wilczek (WW) relation between g1g_1 and g2g_2. In addition, another relation similar to the Burkhardt-Cottingham (BC) sum rule was obtained. It is known that a formal way to derive the WW relation and the BC sum rule is to use the operator product expansion (OPE) with local operators. In this work, the WW-like relation and the BC-like sum rule for fLTf_{LT} are derived by using the local OPE method as a reliable independent way to establish these relations.

Introduction

Spin structure of the spin-1/2 nucleon has been investigated in details at high energies for finding the origin of nucleon spin. Hadrons with higher spin ≥1\ge1 have much richer spin structure because of the existence of tensor polarizations. There are recent theoretical developments on polarized parton distribution functions (PDFs) of spin-1 [1] and spin-3/2 [2] hadrons. This additional spin structure has not been investigated extensively in comparison with the one of the spin-1/2 nucleon partly because of a lack of an experimental project for a long time. However, time has come to investigate this topic seriously because an electron deep inelastic scattering (DIS) experiment is under preparation now at the Thomas Jefferson National Accelerator Facility (JLab) with the tensor-polarized deuteron [3, 4]. There are other experimental facilities where tensor-polarized deuteron experiments will be possible in several years, such as the Fermi National Accelerator Laboratory (Fermilab) [5, 6], the Nuclotron-based Ion Collider fAcility (NICA) [7], the Large Hadron Collider (LHC) [8], and Electron-Ion Colliders (EICs) [9, 10]. In order to understand such upcoming experimental measurements on the spin-1 deuteron, it is necessary to investigate the tensor-polarized structure functions and PDFs theoretically. In particular, JLab data are expected to contain significant higher-twist effects because they are obtained typically in a few GeV2^2 Q2Q^2 region. In addition to the twist-2 functions [11], possible twist-3 and 4 PDFs and transverse-momentum-dependent parton distribution functions (TMDs) were obtained for spin-1 hadrons in Ref. [12]. Among them, there are the twist-2 PDF f1LLf_{1LL} and the twist-3 one fLTf_{LT}. It was shown in Ref. [13] that there exists a twist-2 relation between f1LLf_{1LL} and fLTf_{LT} in the similar way to the Wandzura-Wilczek (WW) relation [14] between the polarized structure functions g1g_1 and g2g_2. It was also shown that the deviation from this WW-like relation indicates twist-3 multipartonic distribution functions, and this deviation part is called the dynamical (or genuine) twist-3 term. In addition, a sum rule which is similar to the Burkhardt-Cottingham (BC) sum rule [15] was also obtained [13]. Based on these twist-2 relations, the function fLTf_{LT} of the deuteron was numerically estimated for the first time [16].

In the derivation of the BC-like sum rule and WW-like relation [13], a nonlocal operator was used with lightcone correlation functions, whereas a formal way to show such relations is to use the operator product expansion (OPE) with local operators. In addition, it is known that the original WW relation is related to the rotational invariance [17, 18, 19]. It is desirable to show the WW-like relation and the BC-like sum rule by the traditional local OPE method, in which the rotational invariance is manifestly satisfied, as a reliable independent method to confirm the relations for establishing them.

In this work, the WW-like relation and the BC-like sum rule are derived for fLTf_{LT} by using the local OPE method. This article consists of the following. In Section 2, the tensor-polarized PDFs f1LLf_{1LL} and fLTf_{LT} are introduced for spin-1 hadrons. The WW-like relation and BC-like sum rule are derived by the local OPE method in Section 3, and our studies are summarized in Section 4.

Tensor-polarized parton distribution functions in a spin-1 hadron

The PDFs of a hadron are generally defined in the correlation function with a nonlocal operator as

Φij[c](k,P,T)=∫d4ξ(2π)4eik⋅ξ⟨P,T∣ψˉj(0)W[c](0,ξ)ψi(ξ)∣P,T⟩,(1)\Phi_{ij}^{[c]}(k,P,T)=\int\frac{d^4\xi}{(2\pi)^4}e^{ik\cdot\xi}\langle P,T|\bar{\psi}_j(0)W^{[c]}(0,\xi)\psi_i(\xi)|P,T\rangle, \tag*{(1)}

where ψ\psi is the quark field, PP and TT indicate the hadron momentum and tensor polarization, ii and jj are components of ψ\psi, kk is the quark momentum, and W[c]W^{[c]} is the gauge link for the color gauge invariance with the integral path cc. The correlation function is the amplitude to extract a parton from the hadron and to insert it at a different space-time point ξ\xi. We do not write the spin vector polarization SS explicitly in Eq. (1) because only the tensor polarization TT, which is specific to hadrons with spin≥1\geq1, is investigated in this paper. The vector polarization part is essentially the same as the one for the spin-1/2 nucleons. The collinear correlation function is obtained by integrating Eq. (1) over the parton momentum as

Φij(x,P,T)=∫d2kT dk+dk− Φij[c](k,P,T)δ(k+−xP+)=∫dξ−2πeixP+ξ−⟨P,T∣ψˉj(0)W(0,ξ∣n)ψi(ξ)∣P,T⟩ξ+=0,ξ⃗T=0,(2)\begin{aligned} \Phi_{ij}(x,P,T)=\int d^2k_T\,dk^+dk^-\,\Phi_{ij}^{[c]}(k,P,T)\delta(k^+-xP^+) \\ &=\int\frac{d\xi^-}{2\pi}e^{ixP^+\xi^-}\langle P,T|\bar{\psi}_j(0)W(0,\xi|n)\psi_i(\xi)|P,T\rangle_{\xi^+=0,\vec{\xi}_T=0}, \tag*{(2)} \end{aligned}

where the lightcone ±\pm momenta are defined by a±=(a0±a3)/2a^\pm=(a^0\pm a^3)/\sqrt{2}, and xx is the lightcone momentum fraction given by k+=xP+k^+=xP^+. The gauge link W(0,ξ∣n)W(0,\xi|n) indicates the line connecting 0 to ξ=(ξ+,ξ−,ξ⃗T)\xi=(\xi^+,\xi^-,\vec{\xi}_T) along the straight lightcone direction of ξ−\xi^-. The lightcone vectors nμn^\mu and nˉμ\bar{n}^\mu are defined by

nμ=12(1,0,0,−1),nˉμ=12(1,0,0,1).(3)n^\mu=\frac{1}{\sqrt{2}}(1,0,0,-1),\qquad\bar{n}^\mu=\frac{1}{\sqrt{2}}(1,0,0,1). \tag*{(3)}

The tensor polarization TμνT^{\mu\nu} is expressed in terms of the tensor-polarization parameters SLLS_{LL}, SLTμS_{LT}^{\mu}, and STTμνS_{TT}^{\mu\nu} as [11, 12]

Tμν=12[43SLL(P⋅n)2M2nˉμnˉν−23SLL(nˉμnν+nˉνnμ−gTμν)+13SLLM2(P⋅n)2nμnν+P⋅nM(nˉμSLTν+nˉνSLTμ)−M2P⋅n(nμSLTν+nνSLTμ)+STTμν].(4)T^{\mu\nu}=\frac{1}{2}\left[\frac{4}{3}S_{LL}\frac{(P\cdot n)^2}{M^2}\bar{n}^{\mu}\bar{n}^{\nu}-\frac{2}{3}S_{LL}\left(\bar{n}^{\mu}n^{\nu}+\bar{n}^{\nu}n^{\mu}-g_T^{\mu\nu}\right)+\frac{1}{3}S_{LL}\frac{M^2}{(P\cdot n)^2}n^{\mu}n^{\nu}+\frac{P\cdot n}{M}\left(\bar{n}^{\mu}S_{LT}^{\nu}+\bar{n}^{\nu}S_{LT}^{\mu}\right)-\frac{M}{2P\cdot n}\left(n^{\mu}S_{LT}^{\nu}+n^{\nu}S_{LT}^{\mu}\right)+S_{TT}^{\mu\nu}\right]. \tag*{(4)}

where gTμνg_T^{\mu\nu} is given by gTμν=gμν−(nˉμnν+nˉνnμ)g_T^{\mu\nu}=g^{\mu\nu}-(\bar{n}^{\mu}n^{\nu}+\bar{n}^{\nu}n^{\mu}) and MM is the hadron mass. The tensor TμνT^{\mu\nu} is symmetric and traceless, and it satisfies PμTμν=0P_{\mu}T^{\mu\nu}=0. The parameter SLLS_{LL} is the tensor polarization along the longitudinal axis, and SLTμS_{LT}^{\mu} and STTμνS_{TT}^{\mu\nu} indicate polarization differences along the axes between the longitudinal and transverse directions and along the transverse axes, respectively [11, 12]. They satisfy the relations SLT⋅nˉ=SLT⋅n=0S_{LT}\cdot\bar{n}=S_{LT}\cdot n=0 and STTμνnˉμ=STTμνnμ=STTμνgTμν=0S_{TT}^{\mu\nu}\bar{n}_{\mu}=S_{TT}^{\mu\nu}n_{\mu}=S_{TT}^{\mu\nu}g_{T\mu\nu}=0. The collinear correlation function is expressed in terms of these parameters and tensor-polarized PDFs f1LL(x)f_{1LL}(x), eLL(x)e_{LL}(x), fLT(x)f_{LT}(x), and f3LL(x)f_{3LL}(x) as [13]

\Phi(x,P,T)=\frac{1}{2}\left[S_{LL}\slashed{n}f_{1LL}(x)+\frac{M}{P\cdot n}S_{LL}e_{LL}(x)+\frac{M}{P\cdot n}\slashed{P}f_{LT}(x)+\frac{M^2}{(P\cdot n)^2}S_{LL}\slashed{n}f_{3LL}(x)\right], \tag*{(5)}

where ϕ=γμαμ\phi=\gamma^\mu\alpha_\mu. The quark distribution function f1LL(x)f_{1LL}(x) is a twist-2 function, while eLL(x)e_{LL}(x) and fLT(x)f_{LT}(x) are twist-3 functions, and f3LL(x)f_{3LL}(x) is a twist-4 function. The eLL(x)e_{LL}(x) is a chiral-odd function, and the others are chiral-even ones.

In the previous derivation of the WW-like relation and the BC-like sum rule in 2021 [13], the nonlocal operator ψˉ(0)γμψ(ξ)\bar{\psi}(0)\gamma^\mu\psi(\xi) was used. Its matrix element was expressed by the available Lorentz vectors PμP^\mu, ξμ\xi^\mu, and TμνξνT^{\mu\nu}\xi_\nu as

⟨P,T∣ψˉ(0)γμψ(ξ)∣P,T⟩=∫−11dx e−ixP⋅ξ[ξ⋅T⋅ξ{A(x)Pμ+B(x)ξμ}+C(x)Tμνξν].(6)\langle P,T|\bar{\psi}(0)\gamma^\mu\psi(\xi)|P,T\rangle=\int_{-1}^{1}dx\,e^{-ixP\cdot\xi}\left[\xi\cdot T\cdot\xi\{A(x)P^\mu+B(x)\xi^\mu\}+C(x)T^{\mu\nu}\xi_\nu\right]. \tag*{(6)}

The coefficients AA, BB, and CC can be obtained in arbitrary frame, so that they were determined on the lightcone. Then, the twist-3 combination ψˉ(0)(γμγν−γνγμ)ψ(ξ)\bar{\psi}(0)(\gamma^\mu\gamma^\nu-\gamma^\nu\gamma^\mu)\psi(\xi) was used for finding the WW-like relation and also the twist-3 multi-parton distribution functions, which are deviation terms from the WW-like relation. Then, the BC-like sum rule was obtained. In § 3, we derive the WW-like twist-2 relation and the BC-like sum rule by a formal OPE method with local operators as an independent way.

Operator-product-expansion method for deriving the twist-2 relations

The OPE is used for deriving the twist-2 relations in this section. The OPE method for deriving the WW relation and BC sum rule is summarized in Refs. [17, 18, 20]. The polarized structure function g2g_2 of the spin-1/2 nucleon is separated into WW and dynamical twist-3 parts as

g2(x)=g2WW(x)+g2HT(x),(7)g_2(x)=g_2^{\mathrm{WW}}(x)+g_2^{\mathrm{HT}}(x), \tag*{(7)}

where HT indicates higher twist, and then the WW part is expressed by the twist-2 structure function g1g_1 as

g2WW(x)=−g1(x)+∫x1dyyg1(y).(8)g_2^{\mathrm{WW}}(x)=-g_1(x)+\int_x^1\frac{dy}{y}g_1(y). \tag*{(8)}

This is the WW relation. If this WW relation is integrated over xx, it becomes

∫01dx g2WW(x)=0,(9)\int_0^1 dx\,g_2^{\mathrm{WW}}(x)=0, \tag*{(9)}

which is the BC sum rule.

In the following derivations, we follow the explanations in Ref. [20] on the OPE method for deriving the WW relation and the BC sum rule to obtain the corresponding twist-2 relations for fLTf_{LT}. The tensor-polarization part of the matrix element of the nonlocal vector operator ψˉ(0)γσψ(ξ)\bar{\psi}(0)\gamma^\sigma\psi(\xi) is given by [13]

∫d(P⋅ξ)4πP⋅neixP⋅ξ⟨P,T∣ψˉ(0)γσψ(ξ)∣P,T⟩∣ξ+=0,ξ⃗T=0=SLLnˉσf1LL(x)+MP⋅nSLTσfLT(x),(10)\int\frac{d(P\cdot\xi)}{4\pi P\cdot n}e^{i xP\cdot\xi}\langle P,T|\bar{\psi}(0)\gamma^\sigma\psi(\xi)|P,T\rangle\bigg|_{\xi^+=0,\vec{\xi}_T=0}=S_{LL}\bar{n}^{\sigma}f_{1LL}(x)+\frac{M}{P\cdot n}S_{LT}^{\sigma}f_{LT}(x), \tag*{(10)}

in terms of the tensor-polarized PDFs f1LLf_{1LL} and fLTf_{LT}. The twist-4 distribution f3LLf_{3LL} is neglected in this work up to twist 3. Taking nn-th moment of Eq. (10), we obtain

∫−11dx xn−1∫d(P⋅ξ)4πP⋅neixP⋅ξ⟨P,T∣ψˉ(0)γσψ(ξ)∣P,T⟩∣ξ+=0,ξ⃗T=0=SLLnˉσ∫−11dx xn−1f1LL(x)+MP⋅nSLTσ∫−11dx xn−1fLT(x).(11)\int_{-1}^{1}dx\,x^{n-1}\int\frac{d(P\cdot\xi)}{4\pi P\cdot n}e^{ixP\cdot\xi}\langle P,T|\bar{\psi}(0)\gamma^\sigma\psi(\xi)|P,T\rangle\bigg|_{\xi^+=0,\vec{\xi}_T=0} =S_{LL}\bar{n}^{\sigma}\int_{-1}^{1}dx\,x^{n-1}f_{1LL}(x)+\frac{M}{P\cdot n}S_{LT}^{\sigma}\int_{-1}^{1}dx\,x^{n-1}f_{LT}(x). \tag*{(11)}

The quark field ψ(ξ)\psi(\xi) is expanded in a Taylor series as

ψ(ξ)=∑m=0∞1m!(ξ⋅∂)mψ(ξ)∣ξ=0.(12)\psi(\xi)=\sum_{m=0}^{\infty}\frac{1}{m!}(\xi\cdot\partial)^m\psi(\xi)\big|_{\xi=0}. \tag*{(12)}

The factor xn−1x^{n-1} on the left-hand side of Eq. (11) is obtained by taking derivatives of eixP⋅ξe^{ixP\cdot\xi} with respect to P⋅ξP\cdot\xi. Then, the left-hand side of Eq. (11) becomes

∫−11dx xn−1∫d(P⋅ξ)4πP⋅neixP⋅ξ⟨P,T∣ψˉ(0)γσψ(ξ)∣P,T⟩∣ξ+=0,ξ⃗T=0=∫−11dx∫d(P⋅ξ)4πP⋅n1in−1[∂n−1∂(P⋅ξ)n−1eixP⋅ξ]×⟨P,T∣ψˉ(0)γσ∑m=0∞1m!(ξ⋅∂)mψ(ξ)∣ξ=0∣P,T⟩∣ξ+=0,ξ⃗T=0.(13)\begin{aligned} &\int_{-1}^{1}dx\,x^{n-1}\int\frac{d(P\cdot\xi)}{4\pi P\cdot n}e^{ixP\cdot\xi}\langle P,T|\bar{\psi}(0)\gamma^\sigma\psi(\xi)|P,T\rangle\bigg|_{\xi^+=0,\vec{\xi}_T=0}\\ ={}&\int_{-1}^{1}dx\int\frac{d(P\cdot\xi)}{4\pi P\cdot n}\frac{1}{i^{n-1}}\left[\frac{\partial^{n-1}}{\partial(P\cdot\xi)^{n-1}}e^{ixP\cdot\xi}\right]\\ &\qquad\times\langle P,T|\bar{\psi}(0)\gamma^\sigma\sum_{m=0}^{\infty}\frac{1}{m!}(\xi\cdot\partial)^m\psi(\xi)\big|_{\xi=0}|P,T\rangle\bigg|_{\xi^+=0,\vec{\xi}_T=0}. \tag*{(13)} \end{aligned}

Shifting the derivative operation to the matrix-element part by the partial integration and noting the relation

∂μ1⋯∂μn−1∑m=0∞1m!(ξ⋅∂)m=∂{μ1⋯∂μn−1}(1+∑m=1∞1m!(ξ⋅∂)m),(14)\partial^{\mu_1}\cdots\partial^{\mu_{n-1}}\sum_{m=0}^{\infty}\frac{1}{m!}(\xi\cdot\partial)^m=\partial^{\{\mu_1}\cdots\partial^{\mu_{n-1}\}}\left(1+\sum_{m=1}^{\infty}\frac{1}{m!}(\xi\cdot\partial)^m\right), \tag*{(14)}

we obtain

∫−11dx xn−1∫d(P⋅ξ)4πP⋅neixP⋅ξ⟨P,T∣ψˉ(0)γσψ(ξ)∣P,T⟩∣ξ+=0,ξ⃗T=0=nμ1⋯nμn−12(P⋅n)n⟨P,T∣in−1ψˉ(0)γσ∂{μ1⋯∂μn−1}ψ(0)∣P,T⟩.(15)\begin{aligned} &\int_{-1}^{1}dx\,x^{n-1}\int\frac{d(P\cdot\xi)}{4\pi P\cdot n}e^{ixP\cdot\xi}\langle P,T|\bar{\psi}(0)\gamma^\sigma\psi(\xi)|P,T\rangle\bigg|_{\xi^+=0,\vec{\xi}_T=0}\\ ={}&\frac{n_{\mu_1}\cdots n_{\mu_{n-1}}}{2(P\cdot n)^{n}}\langle P,T|i^{n-1}\bar{\psi}(0)\gamma^\sigma\partial^{\{\mu_1}\cdots\partial^{\mu_{n-1}\}}\psi(0)|P,T\rangle. \tag*{(15)} \end{aligned}

for n≥2n\ge2. The curly bracket {⋯ }\{\cdots\} indicates the symmetrization of indices defined by ∂{μ1⋯∂μn−1}=(∂μ1⋯∂μn−1+permutations)/(n−1)!\partial^{\{\mu_1}\cdots\partial^{\mu_{n-1}\}}=(\partial^{\mu_1}\cdots\partial^{\mu_{n-1}}+\text{permutations})/(n-1)!. The derivatives are replaced by the covariant ones by the minimal substitution, then the local operator becomes

R{σμ1⋯μn−1}=in−1ψˉ(0)γσD{μ1⋯Dμn−1}ψ(0)−traces.(16)R^{\{\sigma\mu_1\cdots\mu_{n-1}\}}=i^{n-1}\bar{\psi}(0)\gamma^\sigma D^{\{\mu_1}\cdots D^{\mu_{n-1}\}}\psi(0)-\text{traces}. \tag*{(16)}

Here, the traces indicate that the higher-twist (≥4\ge4) trace terms gμiμjR{σμ1⋯μn−1}g_{\mu_i\mu_j}R^{\{\sigma\mu_1\cdots\mu_{n-1}\}} and gσμiR{μ1⋯μn−1}g_{\sigma\mu_i}R^{\{\mu_1\cdots\mu_{n-1}\}} are subtracted to make the operator traceless, and it becomes the operator up to twist 3. Therefore, the twist-4 PDF f3LL(x)f_{3LL}(x) does not appear in this work. From Eqs. (11), (15), and (16), we obtain

nμ1⋯nμn−12(P⋅n)n⟨P,T∣R{σμ1⋯μn−1}∣P,T⟩=SLLnˉσ∫−11dx xn−1f1LL(x)+MP⋅nSLTσ∫−11dx xn−1fLT(x).(17)\frac{n_{\mu_1}\cdots n_{\mu_{n-1}}}{2(P\cdot n)^n}\langle P,T|R^{\{\sigma\mu_1\cdots\mu_{n-1}\}}|P,T\rangle=S_{LL}\bar{n}^{\sigma}\int_{-1}^{1}dx\,x^{n-1}f_{1LL}(x)+\frac{M}{P\cdot n}S_{LT}^{\sigma}\int_{-1}^{1}dx\,x^{n-1}f_{LT}(x). \tag*{(17)}

Then, the operator is decomposed into symmetric and mixed symmetric terms as

R{σμ1⋯μn−1}=R{σμ1⋯μn−1}+R[σ{μ1]⋯μn−1},R^{\{\sigma\mu_1\cdots\mu_{n-1}\}}=R^{\{\sigma\mu_1\cdots\mu_{n-1}\}}+R^{[\sigma\{\mu_1]\cdots\mu_{n-1}\}},
R[σ{μ1]⋯μn−1}=1n[R[σμ1μ2⋯μn−1]+Rμ1[σμ2⋯μn−1]+Rμ2[σμ1⋯μn−1]+⋯ ],R^{[\sigma\{\mu_1]\cdots\mu_{n-1}\}}=\frac{1}{n}\left[R^{[\sigma\mu_1\mu_2\cdots\mu_{n-1}]}+R^{\mu_1[\sigma\mu_2\cdots\mu_{n-1}]}+R^{\mu_2[\sigma\mu_1\cdots\mu_{n-1}]}+\cdots\right],
R[σ{μ1]⋯μn−1}=1n[(n−1)Rσ[μ1μ2⋯μn−1]−Rμ1[σμ2⋯μn−1]−Rμ2[μ1σ⋯μn−1]−⋯ ].(18)R^{[\sigma\{\mu_1]\cdots\mu_{n-1}\}}=\frac{1}{n}\left[(n-1)R^{\sigma[\mu_1\mu_2\cdots\mu_{n-1}]}-R^{\mu_1[\sigma\mu_2\cdots\mu_{n-1}]}-R^{\mu_2[\mu_1\sigma\cdots\mu_{n-1}]}-\cdots\right]. \tag*{(18)}

where the square bracket [⋯ ][\cdots] indicates antisymmetric combination. One may note that the antisymmetric combinations are taken between σ\sigma and all the other indices μ1,⋯ ,μn−1\mu_1,\cdots,\mu_{n-1} on the right-hand side of Eq. (18), although the square bracket is written only between σ\sigma and μ1\mu_1 on the left hand side. The twist is defined by the operator mass dimension minus its spin. The symmetric operator R{σμ1⋯μn−1}R^{\{\sigma\mu_1\cdots\mu_{n-1}\}} and the mixed symmetric one R[σ{μ1]⋯μn−1}R^{[\sigma\{\mu_1]\cdots\mu_{n-1}\}} are twist-2 and 3 operators, respectively.

In general, these matrix elements of the local operators should be expressed in terms of the available Lorentz vector PμP^\mu and the tensor TμνT^{\mu\nu} as

⟨P,T∣R{σμ1⋯μn−1}∣P,T⟩=2nanM2[∑i=1n−1(Tσμi+Tμiσ)∏j=1j≠in−1Pμj+∑i=1n−1∑j=1j≠in−1TμiμjPσ∏k=1k≠i,jn−1Pμk],(19)\begin{aligned} \langle P,T|R^{\{\sigma\mu_1\cdots\mu_{n-1}\}}|P,T\rangle={}&\frac{2}{n}a_nM^2\left[\sum_{i=1}^{n-1}(T^{\sigma\mu_i}+T^{\mu_i\sigma})\prod_{\substack{j=1\\j\ne i}}^{n-1}P^{\mu_j}\right.\\ &\left.+\sum_{i=1}^{n-1}\sum_{\substack{j=1\\j\ne i}}^{n-1}T^{\mu_i\mu_j}P^\sigma\prod_{\substack{k=1\\k\ne i,j}}^{n-1}P^{\mu_k}\right], \tag*{(19)} \end{aligned}
⟨P,T∣R[σ{μ1]⋯μn−1}∣P,T⟩=2ndnM2∑i=1n−1∑j=1j≠in−1(PσTμiμj−PμiTσμj)∏k=1k≠i,jn−1Pμk,(20)\langle P,T|R^{[\sigma\{\mu_1]\cdots\mu_{n-1}\}}|P,T\rangle=\frac{2}{n}d_nM^2\sum_{i=1}^{n-1}\sum_{\substack{j=1\\j\ne i}}^{n-1}(P^\sigma T^{\mu_i\mu_j}-P^{\mu_i}T^{\sigma\mu_j})\prod_{\substack{k=1\\k\ne i,j}}^{n-1}P^{\mu_k}, \tag*{(20)}

where ana_n and dnd_n are overall constants. In order to calculate the left-hand side of Eq. (17) by using Eq. (20), we first need to calculate nμTμνn_\mu T^{\mu\nu} and nμnνTμνn_\mu n_\nu T^{\mu\nu}. From Eqs. (4) and (3), they are obtained as

nμTμν=23SLL(P⋅n)2M2nˉν+P⋅n2MSLTν−13SLLnν,n_\mu T^{\mu\nu}=\frac{2}{3}S_{LL}\frac{(P\cdot n)^2}{M^2}\bar{n}^{\nu}+\frac{P\cdot n}{2M}S_{LT}^{\nu}-\frac{1}{3}S_{LL}n^\nu,
nμnνTμν=23SLL(P⋅n)2M2.(21)n_\mu n_\nu T^{\mu\nu}=\frac{2}{3}S_{LL}\frac{(P\cdot n)^2}{M^2}. \tag*{(21)}

In addition, the hadron momentum is expressed as [12]

Pμ=(P⋅n)nˉμ+M22P⋅nnμ.(22)P^\mu=(P\cdot n)\bar{n}^\mu+\frac{M^2}{2P\cdot n}n^\mu. \tag*{(22)}

By using these relations, the left-hand side of Eq. (17) becomes

nμ1⋯nμn−12(P⋅n)⟨P,T∣Rσ{μ1⋯μn−1}∣P,T⟩=SLLnσ[2(n−1)3an]+MP⋅nSLTσ[n−1nan−(n−1)(n−2)2ndn],(23)\frac{n_{\mu_1}\cdots n_{\mu_{n-1}}}{2(P\cdot n)}\langle P,T\vert R^{\sigma\{\mu_1\cdots\mu_{n-1}\}}\vert P,T\rangle = S_{LL}n^\sigma\left[\frac{2(n-1)}{3}a_n\right]+\frac{M}{P\cdot n}S_{LT}^\sigma\left[\frac{n-1}{n}a_n-\frac{(n-1)(n-2)}{2n}d_n\right], \tag*{(23)}

where the twist-4 term O(M2/(P⋅n)2)O(M^2/(P\cdot n)^2) is neglected.

The antiquark distribution is given by the distribution at negative xx as fˉ(x)=−f(−x)\bar{f}(x)=-f(-x) for the chiral-even functions f1LL(x)f_{1LL}(x) and fLT(x)f_{LT}(x) [13] and the distributions f±(x)f^\pm(x) are defined by

f±(x)≡f(x)±fˉ(x).(24)f^\pm(x)\equiv f(x)\pm\bar{f}(x). \tag*{(24)}

Identifying each term of the left-hand sides of Eqs. (17) and (23), we obtain the moments of the tensor-polarized PDFs as

∫−11dx xn−1f1LL(x)=∫01dx xn−1f1LL±(x)=2(n−1)3an,\int_{-1}^{1} dx\,x^{n-1}f_{1LL}(x)=\int_{0}^{1}dx\,x^{n-1}f^\pm_{1LL}(x)=\frac{2(n-1)}{3}a_n,
∫−11dx xn−1fLT(x)=∫01dx xn−1fLT±(x)=n−1nan−(n−1)(n−2)2ndn,(25)\int_{-1}^{1} dx\,x^{n-1}f_{LT}(x)=\int_{0}^{1}dx\,x^{n-1}f^\pm_{LT}(x) =\frac{n-1}{n}a_n-\frac{(n-1)(n-2)}{2n}d_n, \tag*{(25)}

where f+f^+ is for even integer n=2,4,⋯n=2,4,\cdots and f−f^- is for odd integer n=1,3,⋯n=1,3,\cdots.

For the distribution f(x)f(x) defined in the range 0≤x≤10\leq x\leq1, we have the following relation

∫01dx xn−1∫x1dyyf(y)=1n∫01dx xn−1f(x).(26)\int_{0}^{1}dx\,x^{n-1}\int_{x}^{1}\frac{dy}{y}f(y)=\frac{1}{n}\int_{0}^{1}dx\,x^{n-1}f(x). \tag*{(26)}

From the moments of f1LL±f^\pm_{1LL} and fLT±f^\pm_{LT} in Eq. (25), one obtains by using Eq. (26)

∫01dx xn−1fLT±(x)=∫01dx xn−1[32∫x1dyyf1LL±(y)]−(n−1)(n−2)2ndn.(27)\int_{0}^{1}dx\,x^{n-1}f^\pm_{LT}(x)=\int_{0}^{1}dx\,x^{n-1}\left[\frac{3}{2}\int_{x}^{1}\frac{dy}{y}f^\pm_{1LL}(y)\right]-\frac{(n-1)(n-2)}{2n}d_n. \tag*{(27)}

On the right-hand side, the first term comes from the twist-2 term and the second one is from the dynamical twist-3 term as given by the coefficient dnd_n. This equation indicates that the PDF fLT(x)f_{LT}(x) is written in terms of the WW-like term fLT±twist-2(x)f^\pm_{LT}{}^{\mathrm{twist\text{-}2}}(x) and the remaining dynamical twist-3 one fLT±(HT)(x)f^\pm_{LT}{}^{\mathrm{(HT)}}(x)

fLT±(x)=fLT±twist-2(x)+fLT±(HT)(x),(28)f^\pm_{LT}(x)=f^\pm_{LT}{}^{\mathrm{twist\text{-}2}}(x)+f^\pm_{LT}{}^{\mathrm{(HT)}}(x), \tag*{(28)}
fLT±twist-2(x)=32∫x1dyyf1LL±(y),(29)f^\pm_{LT}{}^{\mathrm{twist\text{-}2}}(x)=\frac{3}{2}\int_x^1\frac{dy}{y}f^\pm_{1LL}(y), \tag*{(29)}
∫01dx xn−1fLT±(HT)(x)=−(n−1)(n−2)2ndn.(30)\int_0^1 dx\,x^{n-1}f^\pm_{LT}{}^{\mathrm{(HT)}}(x)=-\frac{(n-1)(n-2)}{2n}d_n. \tag*{(30)}

Here, fLT±twist-2f^\pm_{LT}{}^{\mathrm{twist\text{-}2}} is the twist-3 function part determined by the twist-2 function f1LLf_{1LL}, and it contains kinematical and intrinsic terms [21, 22]. Equation (29) is the WW-like relation obtained in Ref. [13]. If we define the distribution f2LT±(x)f^\pm_{2LT}(x) by

f2LT±twist-2(x)=23fLT±(x)−f1LL±(x),(31)f^\pm_{2LT}{}^{\mathrm{twist\text{-}2}}(x)=\frac{2}{3}f^\pm_{LT}(x)-f^\pm_{1LL}(x), \tag*{(31)}

Eq. (29) becomes

f2LT±twist-2(x)=−f1LL±(x)+∫x1dyyf1LL±(y),(32)f^\pm_{2LT}{}^{\mathrm{twist\text{-}2}}(x)=-f^\pm_{1LL}(x)+\int_x^1\frac{dy}{y}f^\pm_{1LL}(y), \tag*{(32)}

which is the same form as the original WW relation in Eq. (8). Integrating this equation over xx, we obtain

∫01dx f2LT±twist-2(x)=0,(33)\int_0^1 dx\,f^\pm_{2LT}{}^{\mathrm{twist\text{-}2}}(x)=0, \tag*{(33)}

which is similar to the BC sum rule for g2g_2 in Eq. (9). We note that the local operators in Eq. (16) are defined for n≥2n\geq2, so that the WW relation and BC sum rule could not be rigorously proven in the operator-product-expansion formalism [23, 24, 25, 26]; however, the WW relation and the BC sum rule are consistent with other derivations. Because the WW- and BC-like relations exist for both f+f^+ and f−f^-, the same relations are valid for ff and fˉ\bar{f} separately.

In this way, we confirmed the WW- and BC-like relations derived in Ref. [13] by the independent formal method. Therefore, these relations can be used as reliable first estimate of the twist-3 function fLT(x)f_{LT}(x) in future theoretical and experimental studies. The JLab experimental results will be reported in the near future on the tensor-polarized deuteron. The JLab data could contain significant higher-twist contributions in a few GeV2^2 region, so that this work should be valuable. There are recent theoretical estimates of fLTf_{LT} for the deuteron [16] and the ρ\rho meson [27]. The fLTf_{LT} of the deuteron could be measured in semi-inclusive DIS with the tensor-polarized deuteron target [28] and proton-deuteron Drell-Yan process [29] by looking at the dependence of the azimuthal-angle (ϕLT\phi_{LT}) of the transverse tensor-polarization vector SLTμS^\mu_{LT}.

The present work is limited to the quark distributions. Similar relations have not been investigated for gluon distributions because higher-twist tensor-polarized gluon distributions are not formulated yet for spin-1 hadrons, although analogous relations were derived for spin-1/2 hadrons [30, 31].

The tensor-polarized PDFs contain different aspects of high-energy spin physics from the ones of the spin-1/2 nucleon. Especially, they are sensitive to dynamical aspects beyond the simple addition of proton and neutron structure functions. A new spin-physics field could be created by investigating spin-1 PDFs and structure functions along the experimental projects at the aforementioned lepton and hadron accelerator facilities starting from the JLab experiment.

Summary

Using the OPE method with the local operators, we derived the WW-like relation and the BC-like sum rule for the twist-3 tensor-polarized PDF fLTf_{LT}. The results confirm the relations obtained in Ref. [13], where the matrix element of the non-local operator was expanded in terms of possible Lorentz vectors, by an independent formal method. Now, these relations can be used as reliable ones in future theoretical and experimental studies. In particular, the JLab data will be reported for the tensor-polarized deuteron and they could contain higher-twist effects, so that our studies should be valuable in analyzing the experimental data and in making theoretical models.

Acknowledgments

The authors were supported by the Chinese Academy of Sciences. KK was also supported by the Gansu-province postdoctoral foundation. They thank Qin-Tao Song for suggestions.

References

  1. [1]S. Kumano, Euro. Phys. J. A 60 (2024) 205 and references therein.
  2. [2]D.-Y. Fu, Y.-B. Dong, S. Kumano, and J.-J. Xie, Phys.Rev.D 113 (2026) L111901 and references therein.
  3. [3]K. Allada et al., Proposal to Jefferson Lab PAC-40, PR12-13-011 (Update to PR12-11-110) (2023).
  4. [4]J. Poudel, A. Bacchetta, J.-P. Chen, and N. Santiesteban, Euro. Phys. J. A 61 (2025) 81.
  5. [5]M. Brooks et al., Fermilab proposal (2017).
  6. [6]D. Keller, D. Crabb, and D. Day, Nucl. Instrum. Meth. A 981 (2020) 164504.
  7. [7]A. Arbuzov et al., Prog. Nucl. Part. Phys. 119 (2021) 103858.
  8. [8]C. A. Aidala et al., arXiv:1901.08002.
  9. [9]R. Abdul Khalek et al., Nucl. Phys. A 1026 (2022) 122447.
  10. [10]D. P. Anderle et al., Front. Phys. 16 (2021) 64701.
  11. [11]A. Bacchetta and P. Mulders Phys. Rev. D 62 (2000) 114004.
  12. [12]S. Kumano and Q.-T. Song, Phys. Rev. D 103 (2021) 014025.
  13. [13]S. Kumano and Q.-T. Song, J. High Energy Phys. 2021 (2021) 141.
  14. [14]S. Wandzura and F. Wilczek, Phys. Lett. B 72 (1977) 195.
  15. [15]H. Burkhardt and W. N. Cottingham, Ann. Phys. 56 (1970) 453.
  16. [16]S. Kumano and K. Kuroki, Phys. Lett. B 875 (2026) 140348.
  17. [17]R. L. Jaffe, Comm. Nucl. Part. Phys. 19 (1990) 239.
  18. [18]R. L. Jaffe, arXiv:hep-ph/9602236.
  19. [19]I. V. Anikin and O. V. Teryaev, Phys. Lett. B 509 (2001) 95.
  20. [20]J. Kodaira and K. Tanaka, Prog. Theor. Phys. 101 (1999) 191.
  21. [21]K. Kanazawa, Y. Koike, A. Metzl, D. Pitonyak, and M. Schlegel, Phys. Rev. D 93 (2016) 054024.
  22. [22]Q.-T. Song, Phys. Rev. D 108 (2023) 094041.
  23. [23]V. Barone and R. G. Ratcliffe, Transverse Spin Physics (World Scientific, Singapore, 2003).
  24. [24]J. Blümlein and N. Kochelev, Nucl. Phys. B 498 (1997) 285.
  25. [25]M. Anselmino, A. Efremov, and E. Leader, Phys. Rept. 261 (1995) 1.
  26. [26]A. Deur, S. J. Brodsky, and G. F de Teramond, Rep. Prog. Phys. 82 (2019) 076201.
  27. [27]S. Puhan, S. Sharma, N. Kumar, and H. Dahiya, Phys. Rev. D 113 (2026) 036030.
  28. [28]J. Zhao, A. Bacchetta, S. Kumano, T. Liu, and Y.-J. Zhou, JHEP 12 (2025) 067.
  29. [29]S.-Y. Qiao and Q.-T. Song, Phys. Rev. D 111 (2025) 054026.
  30. [30]Y. Hatta, K. Tanaka, and S. Yoshida, JHEP 02 (2013) 003.
  31. [31]Y. Koike, K. Yabe, and S. Yoshida, Phys. Rev. D 101 (2020) 054017.

Paper details

vlnc:W_p8wv3spu
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents