betaFeedback

WorksPhysics and Astronomy

Three-dimensional parton structure in QCD with CP-violation

Abstract

The Energy-Momentum tensor encodes the mechanical properties of a system such as the internal distributions energy, momentum, forces, and particle fluxes. The QCD energy momentum tensor respects CP symmetry even when QCD is modified with a CP-violating θθ-term. In this theory, there are 5 unique generalized form factors in the Lorentz decomposition of the total Energy Momentum tensor and 3 more which only vanish when summing over all flavors. We introduce two new lightcone GPDs whose Mellin moments give the electromagnetic and gravitational form factors. These generalize the observables for QCD-based CP violation such as the neutron electric dipole moment to three dimensional light cone structure.

Introduction

The energy-momentum tensor (EMT) of a system, such as a proton, encodes how the internal energy, pressure, and other mechanical properties are distributed. These features provide insight into how the constituents, called partons, conspire to form the bulk properties. For hadronic systems, the EMT provides distributions of the mass, spin, internal forces and flux.

The off-forward matrix elements of the EMT are parameterized by the generalized form factors (GFFs). In an analogous way to how the form factors of the electromagnetic current detail properties such as the charge and magnetic moment distribution, the form factors of the EMT detail the mechanical properties and distribution of forces. These form factors are of great interest both theoretically and phenomenologically as they provide a window into the mechanical and internal properties of hadrons.

In the standard QCD Lagrangian, the theory is CPCP-invariant and the nucleon matrix elements of the total, conserved EMT TμνT^{\mu\nu} can be decomposed into three, non-zero, independent gravitational form factors. Specifically, for the Belinfante improved EMT [1, 2], the matrix elements decompose into the GFFs A(t)A(t), B(t)B(t) and C(t)C(t). Some conventions introduce an additional trace term Cˉ(t)\bar{C}(t); such that in the sum of quark and gluon contributions Cˉ(t)=0\bar{C}(t) = 0. These nucleon GFFs have been targets of recent Lattice QCD calculations. [3, 4, 5, 6, 7]

When a CPCP-violating θ\theta-term is added to the QCD action, the EMT operator is unchanged, but the matrix elements are affected because of the effects on the hadronic states. Therefore nothing forbids CPCP-odd Dirac structures in its matrix elements. As a result, additional pseudoscalar or axial structures can appear, leading to new gravitational form factors. Our goal is to perform this decomposition explicitly.

There has been significant interest in the study of CPCP-violating effects in QCD. The absence of observed CPCP violation in strong interactions, despite the theoretical allowance for a θ\theta-term, is the famous strong CPCP problem. See Ref. [8] for a review on this subject. Most sensitive of experimental probes has been the neutron electric dipole moment (nEDM), which constrains the θ\theta parameter to be extremely small, ∣θ∣≲10−10|\theta| \lesssim10^{-10} [9]. There have been several theoretical proposals to explain this smallness, including the Peccei-Quinn mechanism [10, 11, 12] and related axion models [8]. However, a lack of experimental signal for axions [13] has motivated continued interest in the direct study of CPCP-violating effects in QCD, including the effects of a non-zero θ\theta on hadronic properties. This is the motivation and context for exploring the EMT decomposition in the presence of a θ\theta-term.

Certain GFFs are related to the Mellin moments of Generalized Parton Distributions (GPDs). GPDs encode the internal structure of the hadron in terms of the constituent’s longitudinal momentum fraction xx, the longitudinal momentum transfer encoded in skewness ξ\xi and the total momentum transfer tt. Specific integrals of the GPDs give the Impact Parameter Distributions (IPDs) which provide a 3-dimensional image of the hadron in terms of the parton’s transverse separation or impact parameter and either the parton’s momentum fraction xx [14] or the parton’s interaction time or Ioffe time ν\nu [15].

GPDs can be accessed experimentally through processes such as deeply virtual Compton scattering (DVCS) [16, 17, 18], deeply virtual meson production (DVMP) [19, 20] and constrained by single diffractive hard exclusive processes (SDHEP) [21], which are measured, or planned to be measured, at CE-BAF [22, 23], J-PARC [24], SPS [25, 26] and the EIC [27]. GPDs are also the target of lattice QCD calculations [28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38]. Phenomenological analysis of experiments and other constraints, some even including lattice QCD data, have begun to extract GPDs [39, 40, 41, 42, 43]. Similar questions about the influence of a non-zero topological term are relevant in this sector as well.

In this work we discuss prospects for using hadronic processes and structure studies to probe for a non-zero θ\theta term. We do this by considering the Lorentz decomposition of matrix elements of the energy-momentum tensor. When enumerating the GFFs in the presence of a non-zero θ\theta-term, we find five structures that survive after imposing the on-shell Dirac and Gordon identities, hermiticity, PTPT invariance, and energy-momentum conservation. This is one fewer than the decompositions of [44] and [45] the former of which considered gravitational interactions with fermions; the latter extended their work to neutrinos. Note, we neglect gauge-fixing and ghost contributions since we are ultimately interested in the Lorentz decomposition of the matrix elements of the EMT operator between physical hadronic states.

Other studies of parton structure of QCD with a θ\theta-term have introduced parity-violating fragmentation functions [46] and parton distribution functions (PDF) [47, 48] and investigated their contributions to inclusive deep-inelastic scattering [49]. These are different from what will be studied below in that they are studying matrix elements of operators which are usually dropped due to parity, while the Lorentz decompositions of the GFFs and GPDs presented below are the matrix elements of standard CPCP-even operators and we are asking what CPCP-odd contributions can be generated when CPCP is broken. We extend these investigations to the non-local lightcone correlators introducing two new unpolarized and two new helicity GPDs which appear in the presence of CPCP-violation. We identify the relationships between the new GFFs and GPDs. There will also be GPDs generalizations of the parity-violating PDF as well to be investigated in future work.

The outline of this work is as follows. In Section II we present the Lorentz decomposition of the EMT for nonzero θ\theta. Section III extends these considerations to the matrix elements defining GPDs and derive their relationship to the GFFs. Finally, Sect. IV summarizes our findings and suggests future directions.

EMT in θ\theta-QCD

The QCD action permits an additional term built from the gauge field-strength tensor and its dual,

\mathcal{L}_{\theta\mathrm{QCD}}=\bar{\psi}\left(i\slashed{D}-m\right)\psi-\frac{1}{4}G_a^{\mu\nu}G^a_{\mu\nu}-\frac{\theta g^2}{32\pi^2}G_a^{\mu\nu}\widetilde{G}^a_{\mu\nu} \tag*{(1)}

Here, Gμνa=∂μAνa−∂νAμa+gfabcAμbAνcG^a_{\mu\nu}=\partial_\mu A^a_\nu-\partial_\nu A^a_\mu+g f^{abc}A^b_\mu A^c_\nu and G~aμν=12ϵαβμνGαβa\widetilde{G}^{a\mu\nu}=\frac{1}{2}\epsilon^{\alpha\beta\mu\nu}G^a_{\alpha\beta} are the field strength tensor and its dual. Including a CPCP-violating θ\theta term in the QCD Lagrangian does not modify the local operator form of the EMT. It is the vacuum and hadronic states that are altered. Thus, the CPCP-odd structures result from these altered states, not from a change to the EMT operator itself.

For this work, we use the Belinfante-improved EMT, which is symmetric and gauge-invariant

Tμν=i4ψˉ(γμD↔ν+γνD↔μ)ψ−GμαG αν+14gμνGαβGαβ(2)T^{\mu\nu}=\frac{i}{4}\bar{\psi}\left(\gamma^\mu\overleftrightarrow{D}^{\nu}+\gamma^\nu\overleftrightarrow{D}^{\mu}\right)\psi-G^{\mu\alpha}G^\nu_{\ \alpha}+\frac{1}{4}g^{\mu\nu}G^{\alpha\beta}G_{\alpha\beta} \tag*{(2)}

Notice that the field strength tensor remains unchanged for nonzero θ\theta. The hadronic GFFs are obtained by decomposing the off-forward matrix elements of the EMT into the allowed Lorentz covariant structures,

⟨p′,λ′∣Tμν∣p,λ⟩=uˉ(p′,λ′)Γμν(P,Δ)u(p,λ)=∑iAi(t)Tiμν(P,Δ,λ,λ′).(3)\langle p',\lambda'|T^{\mu\nu}|p,\lambda\rangle=\bar{u}(p',\lambda')\Gamma^{\mu\nu}(P,\Delta)u(p,\lambda) =\sum_i A_i(t)T_i^{\mu\nu}(P,\Delta,\lambda,\lambda'). \tag*{(3)}

where P=p′+p2P=\frac{p'+p}{2}, and Δ=p′−p\Delta=p'-p, Δ2=t\Delta^2=t are the total momentum and momentum transfer variables and λ\lambda and λ′\lambda' are the nucleon helicities. The Dirac spinors u(p,λ)u(p,\lambda) and uˉ(p′,λ′)\bar{u}(p',\lambda') correspond to the initial and final spin-1/2 hadron states, respectively. The possible Lorentz structures, TiμνT_i^{\mu\nu}, are constrained by Lorentz invariance, hermiticity, and discrete symmetries such as parity and time reversal. These constraints are used to reduce the number of independent terms in the decomposition.

The set of symmetric rank-two tensors, TiμνT_i^{\mu\nu}, are constructed from a basis of Lorentz and Dirac structures multiplying the kinematic variables PP and Δ\Delta. The only available Lorentz and Dirac structures are,

Lorentz structures : gμνg^{\mu\nu}, ϵαβμν\epsilon^{\alpha\beta\mu\nu}, PμP^\mu, Δμ\Delta^\mu,

Dirac structures : 1\mathbb{1}, γμ\gamma^\mu, σμν\sigma^{\mu\nu}, γ5\gamma^5, γ5γμ\gamma^5\gamma^\mu, γ5σμν\gamma^5\sigma^{\mu\nu}.

Notice the doubling of available Dirac structures from CPCP-symmetry breaking. As detailed in the Appendix A, combinations and entire structures, like the Levi-Civita tensor, can be immediately removed from this overcomplete list. After removing such terms, an explicit decomposition is given in Eq. (20). This decomposition is then reduced further using the Dirac equation, Gordon identities, hermiticity, and discrete symmetry constraints. The total number of independent factors can be reduced to eight terms

Γaμν=Aa(t)γ{μPν}+Ba(t)P{μiσν}ρΔρ2M+Ca(t)ΔμΔν−gμνtM+MgμνCˉa(t)+B5,a(t)iγ5MPμPν+C5,a(t)iγ5M(ΔμΔν−gμνt)+gμνiMγ5Cˉ5,a(t)+iFˉ5,a(t)γ5γ{μΔν}.(4)\Gamma_a^{\mu\nu}=A_a(t)\gamma^{\{\mu}P^{\nu\}}+B_a(t)\frac{P^{\{\mu}i\sigma^{\nu\}\rho}\Delta_\rho}{2M}+C_a(t)\frac{\Delta^\mu\Delta^\nu-g^{\mu\nu}t}{M}+Mg^{\mu\nu}\bar{C}_a(t)+B_{5,a}(t)\frac{i\gamma_5}{M}P^\mu P^\nu+C_{5,a}(t)\frac{i\gamma_5}{M}\left(\Delta^\mu\Delta^\nu-g^{\mu\nu}t\right)+g^{\mu\nu}iM\gamma_5\bar{C}_{5,a}(t)+i\bar{F}_{5,a}(t)\gamma_5\gamma^{\{\mu}\Delta^{\nu\}} . \tag*{(4)}

The conservation of the EMT, ΔμTμν(t)=0\Delta_\mu T^{\mu\nu}(t)=0, reduces this set further. The conservation equation manifests as three distinct constraints on the set of allowed structures for the total EMT. The additional constraints come from the linear independence of the tensor structures that remain after contracting with Δμ\Delta_\mu. Ultimately, in contrast to [44, 45], we show that the EMT in θ\theta-QCD has only 5 linearly independent terms in its decomposition. In the appendix we provide two different counts of the independent structures, including a complete derivation of the remaining GFFs. The subscript aa specifies quark flavor or gluon choice. Conservation of the EMT requires ∑aCˉa=∑aCˉ5,a=∑aFˉ5,a=0\sum_a\bar{C}_a=\sum_a\bar{C}_{5,a}=\sum_a\bar{F}_{5,a}=0, leaving 5 structures. Three of these are the standard parity-even form factors, A(t)A(t), B(t)B(t) and C(t)C(t), while the remaining two are CPCP-odd form factors, B5(t)B_5(t) and C5(t)C_5(t).

Although the θ\theta-term breaks CPCP, CPTCPT invariance remains intact. Therefore, the full matrix element of the EMT must satisfy CPTCPT invariance, which constrains the real and imaginary parts of the form factors. In particular, hermiticity implies that all GFFs are real functions for spacelike momentum transfer.

GPDs in θ\theta-QCD

Generalized parton distributions appear in the factorization of DVCS, DVMP, and other SDHEPs. They encode the structure of the hadron in terms of its longitudinal momentum fraction xx, the skewness, which is the longitudinal momentum fraction transferred ξ\xi, and the total momentum transfer tt. They are given by the Fourier transform of the generalized Ioffe time distributions (GITD). Those in turn are the Lorentz invariant functions in the decomposition of the matrix elements

Iq,g(z,p′,p,λ′,λ)=⟨p′,λ′∣Oq,g(z)∣p,λ⟩(5)I^{q,g}(z,p',p,\lambda',\lambda)=\langle p',\lambda'|\mathcal{O}^{q,g}(z)|p,\lambda\rangle \tag*{(5)}

where for quarks

Oq(z)=qˉ(−z2)̸ ⁣zW(−z2,z2)q(z2)(6)\mathcal{O}^{q}(z)=\bar{q}\left(-\frac{z}{2}\right)\not\!zW\left(-\frac{z}{2},\frac{z}{2}\right)q\left(\frac{z}{2}\right) \tag*{(6)}

and, for gluons,

Og(z)=zμzνGμα(−z2)W(−z2,z2)Gνα(z2).(7)\mathcal{O}^{g}(z)=z_\mu z_\nu G^{\mu\alpha}\left(-\frac{z}{2}\right)W\left(-\frac{z}{2},\frac{z}{2}\right)G^\nu{}_{\alpha}\left(\frac{z}{2}\right). \tag*{(7)}

where ii are summed transver indices. Here zz is in the direction of the light cone defined by the spatial direction of the average momentum PP, conventionally, such that z+=0z^+=0.

Both of these matrix elements have a Lorentz decomposition into 4 GITDs

Iq,g(z,p′,p,λ′,λ)=uˉ(p′,λ′)̸ ⁣zu(p,λ)Hq,g(ν,νˉ,t)+uˉ(p′,λ′)iσzαΔα2mu(p,λ)Eq,g(ν,νˉ,t)+uˉ(p′,λ′)γ5̸ ⁣zu(p,λ)H5q,g(ν,νˉ,t)+uˉ(p′,λ′)qαγ5Δα2mu(p,λ)E5q,g(ν,νˉ,t).(8)I^{q,g}(z,p',p,\lambda',\lambda)=\bar{u}(p',\lambda')\not\!z u(p,\lambda)\mathcal{H}^{q,g}(\nu,\bar{\nu},t)+\bar{u}(p',\lambda')\frac{i\sigma^{z\alpha}\Delta_\alpha}{2m}u(p,\lambda)\mathcal{E}^{q,g}(\nu,\bar{\nu},t)+\bar{u}(p',\lambda')\gamma_5\not\!z u(p,\lambda)\mathcal{H}_5^{q,g}(\nu,\bar{\nu},t)+\bar{u}(p',\lambda')\frac{q^\alpha\gamma_5\Delta_\alpha}{2m}u(p,\lambda)\mathcal{E}_5^{q,g}(\nu,\bar{\nu},t). \tag*{(8)}

Here we introduce two new GITDs that appear in the presence of a θ\theta term, H5\mathcal{H}_5 and E5\mathcal{E}_5, from now on dropping the parton’s flavor label.

All GITDs are defined in terms of Lorentz invariants ν=P⋅z\nu=P\cdot z, νˉ=−Δ⋅z/2\bar{\nu}=-\Delta\cdot z/2, and t=Δ2t=\Delta^2. The Lorentz invariant definition of skewness is given by ξ=νˉ/ν\xi=\bar{\nu}/\nu which gives the traditional ratio for light cone zz. At fixed zz, p′p', and pp, all possible combinations of λ,λ′\lambda,\lambda' create a 2×22\times2 complex matrix. Such a matrix can be completely parameterized by 4 independent complex GITDs, so no more terms are needed. Their values at ν=0=νˉ\nu=0=\bar{\nu} are given by local z−=0z_-=0 matrix elements. H(0,0,t)\mathcal{H}(0,0,t) and E(0,0,t)\mathcal{E}(0,0,t) give the Dirac EM form factors F1F_1 and F2F_2. H5(0,0,t)\mathcal{H}_5(0,0,t) gives the parity-violating anapole form factor FAF_A and E5(0,0,t)\mathcal{E}_5(0,0,t) is the F3(t)F_3(t) form factor related to neutron electric dipole moment.

The GPDs themselves are given by the Fourier transform of ν\nu at fixed skewness νˉ=ξν\bar{\nu}=\xi\nu

H5(x,ξ,t)=∫−∞∞dν2πeiνxH5(ν,ξν,t).(9)H_5(x,\xi,t)=\int_{-\infty}^{\infty}\frac{d\nu}{2\pi}e^{i\nu x}\mathcal{H}_5(\nu,\xi\nu,t). \tag*{(9)}

and similarly for HH, EE, and E5E_5 from their respective GITDs. Following [18, 50], Double Distributions

h5(β,α,t)=∫−∞∞dν2πeiνβ∫−∞∞dνˉ2πeiνˉαH5(ν,νˉ,t),(10)h_5(\beta,\alpha,t)=\int_{-\infty}^{\infty}\frac{d\nu}{2\pi}e^{i\nu\beta}\int_{-\infty}^{\infty}\frac{d\bar{\nu}}{2\pi}e^{i\bar{\nu}\alpha}\mathcal{H}_5(\nu,\bar{\nu},t), \tag*{(10)}
GPDEM FF (n=1n=1)GFF (n=2n=2)
HHF1F_1A+4ξ2CA+4\xi^2 C
EEF2F_2B−4ξ2CB-4\xi^2 C
H5H_5FAF_A2ξFˉ52\xi\bar{F}_5
E5E_5F3F_3B5+4ξ2C5B_5+4\xi^2 C_5

Table 1. Table of GPDs and related Mellin moments.

are defined and their relationship to the GPDs explain the standard polynomiality property of GPD Mellin moments. Finally, there are impact parameter distributions in momentum fraction [14] or Ioffe time [15] spaces calculated from the Fourier transform of transverse ΔT\Delta_T at ξ=0\xi=0

h5(ν,bT)=∫d2ΔT(2π)2e−iΔT⋅bTH5(ν,0,−ΔT2),(11)h_5(\nu,b_T)=\int\frac{d^2\Delta_T}{(2\pi)^2}e^{-i\Delta_T\cdot b_T}H_5(\nu,0,-\Delta_T^2), \tag*{(11)}

for example.

Finally, the first and second Mellin moments in xx are related to the EM form factors and the Gravitational form factors. Table I shows the relationship between the specific GPDs, Dirac EM FFs and the GFFs in Sec. II, which can be found by contracting the vector current or EMT with zμz_\mu.

The symmetry properties of the spinor bilinears and of the matrix element will dictate the (ν,x,ξ)(\nu,x,\xi) parities of the GITD and GPD. Consider instead matrix elements of the hermitian eigen-operators

O±(z)=12(O(z)∓O(−z))O_{\pm}(z)=\frac{1}{2}\left(O(z)\mp O(-z)\right)

which have the properties O±†(z)=±O±(z)O_{\pm}^{\dagger}(z)=\pm O_{\pm}(z) and O±(−z)=∓O±(z)O_{\pm}(-z)=\mp O_{\pm}(z). Their matrix elements expose even/odd behavior in the GITDs’ invariants. The factors of ii prescribed in the Lorentz decomposition Eq. (8) are such that the real components of the GITDs are even and the imaginary components components are odd under (ν,νˉ)→(−ν,−νˉ)(\nu,\bar{\nu})\to(-\nu,-\bar{\nu}). From Eq. (9) one can also see the real components are the cosine transforms of H5(x,ξ,t)+H5(−x,ξ,t)H_5(x,\xi,t)+H_5(-x,\xi,t) and the imaginary components are the sine transforms of H5(x,ξ,t)−H5(−x,ξ,t)H_5(x,\xi,t)-H_5(-x,\xi,t).

Consider the impact of hermitian conjugation, parity and time reversal as done in Ref [5]. These combine to the transformation of ξ→−ξ\xi\to-\xi and λi,f→−λi,f\lambda_{i,f}\to-\lambda_{i,f}. Using spinor identities to invert the λi,f\lambda_{i,f} sign again, one finds HH, EE, and E5E_5 to be even under ξ→−ξ\xi\to-\xi while H5H_5 is odd. Polynomiality forces the Mellin moments to be polynomials of ξ\xi whose order depends on the moment’s power. The Mellin moments are exposed in the Taylor expansion of any generic GITD

G(ν,ξν,t)=∑n=1∞(−iν)n−1(n−1)!∫−11dx xn−1G(x,ξ,t)=∑n=1∞(−iν)n−1(n−1)!∑k=0nξkGnk(t)(12)\begin{aligned} \mathcal{G}(\nu,\xi\nu,t)=\sum_{n=1}^{\infty}\frac{(-i\nu)^{n-1}}{(n-1)!}\int_{-1}^{1}dx\,x^{n-1}G(x,\xi,t) \\ &=\sum_{n=1}^{\infty}\frac{(-i\nu)^{n-1}}{(n-1)!}\sum_{k=0}^{n}\xi^kG_{nk}(t) \tag*{(12)} \end{aligned}

and the parity of ν\nu and ξ\xi govern which Gnk(t)G_{nk}(t) vanish where ξn\xi^n term is from the DD term explicitly. For H5H_5 which is odd in ξ\xi, the first moment, the anapole form factor FAF_A, vanishes

FA(t)=∫H5(x,ξ,t)dx=H5(0,0,t)=0.(13)F_A(t)=\int H_5(x,\xi,t)dx=H_5(0,0,t)=0. \tag*{(13)}

Since Fˉ5\bar{F}_5 vanishes when summed over all flavors, the sum of the second Mellin moments of H5H_5 also vanishes. Similar to the unpolarized GITDs are the helicity distributions from the matrix elements I~q,g\widetilde{I}^{q,g} of the operators

O~q(z)=qˉ(−z2)γ5γzW(−z2,z2)q(z2)(14)\widetilde{O}^{q}(z)=\bar{q}\left(-\frac{z}{2}\right)\gamma_5\gamma^{z}W\left(-\frac{z}{2},\frac{z}{2}\right)q\left(\frac{z}{2}\right) \tag*{(14)}

and

O~g(z)=zμzνGaμi(−z2)W(−z2,z2)G~aνi(z2).(15)\widetilde{O}^{g}(z)=z_\mu z_\nu G_a^{\mu i}\left(-\frac{z}{2}\right)W\left(-\frac{z}{2},\frac{z}{2}\right)\widetilde{G}_a^{\nu i}\left(\frac{z}{2}\right). \tag*{(15)}

These will have the Lorentz decomposition

I~q,g(z,p′,p,λ′,λ)=uˉ(p′,λ′)γzγ5u(p,λ)H~q,g(ν,νˉ,t)+uˉ(p′,λ′)iσzαqα2mu(p,λ)E~q,g(ν,νˉ,t)+iu(p′,λ′)γ5u(p,λ)H~5q,g(ν,νˉ,t)+uˉ(p′,λ′)σzαqα2mu(p,λ)E~5q,g(ν,νˉ,t).(16)\widetilde{I}^{q,g}(z,p',p,\lambda',\lambda)=\bar{u}(p',\lambda')\gamma^z\gamma_5u(p,\lambda)\widetilde{\mathcal{H}}^{q,g}(\nu,\bar{\nu},t) +\bar{u}(p',\lambda')\frac{i\sigma^{z\alpha}q_\alpha}{2m}u(p,\lambda)\widetilde{\mathcal{E}}^{q,g}(\nu,\bar{\nu},t) +iu(p',\lambda')\gamma_5u(p,\lambda)\widetilde{\mathcal{H}}_5^{q,g}(\nu,\bar{\nu},t) +\bar{u}(p',\lambda')\frac{\sigma^{z\alpha}q_\alpha}{2m}u(p,\lambda)\widetilde{\mathcal{E}}_5^{q,g}(\nu,\bar{\nu},t). \tag*{(16)}

into the 4 helicity GITDs. Similar definitions occur for the related GPDs, DDs, and IPDs. The Mellin moments of the helicity GPDs will be related to the Lorentz decomposition of the axial vector current and other generalized form factors for higher moments. Quark transversity GITDs/GPDs as well as new GITDs/GPDs generalizing the CPCP-odd PDFs of Ref. [47, 48] also can be defined by their own Lorentz decompositions.

Conclusions and outlook

This work considers the energy-momentum tensor and generalized parton distributions of the nucleon in QCD with a θ\theta term. The CPCP violations of the Lagrangian do not translate to CPCP violation of the EMT operator itself. Still, changes in the vacuum and hadronic state create matrix elements that violate CPCP. An overcomplete list of Lorentz structures is given and 8 remain after applying equations of motion, and symmetry constraints. Conservation of the EMT reduces to 5 GFFs for the sum of flavors. Similarly, two new unpolarized GPDs and two new helicity GPDs appear in the decomposition of the standard light cone matrix elements. The derivation may be extended to higher-spin hadrons, transition form factors, GPD extensions of the CPCP-odd PDFs of Ref. [47, 48], and the uncontracted GPD operators used for lattice [31, 51], where the number of independent CPCP-odd structures may differ. Systematic methods such as in Ref. [52] applied in App. B will be necessary for finding the minimal basis. Ref. [53] demonstrates an approach for determining GFFs applied to bosons and fermions of many different spins.

Quantitative estimates of new GFFs and GPDs, as well as the θ\theta-QCD corrections to traditional GFFs and GPDs, could be obtained using chiral effective theory, holographic QCD, QCD sum rules or lattice simulations that incorporate a small θ\theta term or imaginary θ\theta methods. Perturbatively approximating the θ\theta term for lattice GFF and GPD calculations is possible through recycling the majority of numerical work already performed, though statistical quality may be lacking without additional computation. The potential interplay with axion dynamics and implications for nucleon structure are also worthy of exploration.

In principle, the GPDs introduced from CPCP violation can be measure via forbidden modes within the scatterings angles of DVCS or other SDHEPs. Consider the SDHEP frame [54] which describe azimuthal angles based upon the plane of the diffracting hadrons’ momenta and the momenta plane of the particles leaving the hard scattering. In the case of DVCS with CPCP conservation, scattering unpolarized nucleon and electron leads to cos⁡ϕ\cos\phi harmonics while longitudinally polarized electrons off unpolarized nucleons leads to sin⁡ϕ\sin\phi harmonics. Parity violation could be measured in the forbidden sin⁡ϕ\sin\phi and cos⁡ϕ\cos\phi modes respectively. The weak sector PP-violation could be a substantial background over these signals which would need to be handled due to θ\theta’s expected small size. A cleverly chosen and precisely measured SDHEPs, or asymmetries of SDHEPs such as comparing electron scattering to positron scattering at Ce++BAF [55], could provide an alternative, though potentially difficult, discovery mechanism for θ\theta-QCD. Future analysis is needed to find ideal observables which have sufficiently suppressed weak sector CPCP violation, so that θ\theta-QCD can be meaningfully constrained to the same extent of neutron EDM measurements.

Acknowledgments

The authors gratefully thank Adam Freese and Jianwei Qiu for productive discussions. This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Physics under Contract No. 8924326CSSC000213. A.M.S. is supported in part by U.S. DOE Grant #DE-SC0025908. This work has benefited from the collaboration enabled by the Quark-Gluon Tomography (QGT) Topical Collaboration, U.S. DOE award #DE-SC0023646.

Lorentz and Dirac structure reductions

From the available Lorentz and Dirac structures, Γμν\Gamma^{\mu\nu} will contain terms of the form

Tn,A,cμν=[P⊗nP⊗Δ⊗nΔ⊗g⊗ng⊗ϵ⊗nϵ⊗DA]c{μν}(17)T_{n,A,c}^{\mu\nu}=\left[P^{\otimes n_P}\otimes\Delta^{\otimes n_\Delta}\otimes g^{\otimes n_g}\otimes\epsilon^{\otimes n_\epsilon}\otimes D_A\right]_{c}^{\{\mu\nu\}} \tag*{(17)}

Where the Dirac factor, DAD_A, is kept general since arbitrary products of gamma matrices can be reduced to a linear combination of {1,γ5,γμ,γ5γμ,σμν}\{1,\gamma_5,\gamma^\mu,\gamma_5\gamma^\mu,\sigma^{\mu\nu}\}. The exponents, {nP,nΔ,ng,nϵ}\{n_P,n_\Delta,n_g,n_\epsilon\}, count the number of occurrence of each structure. The cc subscript labels a particular contraction of all the indices, that necessarily leaves μ\mu and ν\nu free.

The Levi-Civita tensor can be removed from consideration completely. To see this, first note that any product containing two Levi-Civita tensors can be reduced to products of metric tensors through the generalized Kronecker delta, ϵμναβϵρσλκ=−δρσλκμναβ\epsilon^{\mu\nu\alpha\beta}\epsilon_{\rho\sigma\lambda\kappa}=-\delta^{\mu\nu\alpha\beta}_{\rho\sigma\lambda\kappa}. Therefore, terms containing more than one Levi-Civita tensor generate no new Lorentz structures, and we need only argue away terms with a single Levi-Civita tensor. Additionally, such terms have at most 1 free index. Both free indices cannot come from the anti-symmetric tensor because we restrict to a symmetric EMT.

The remaining terms can be classified according to the Dirac factor DAD_A. For terms containing the Dirac scalars, 11 and γ5\gamma_5, the Levi-Civita tensor must contract with at least three kinematic vectors. Since there are only two kinematic vectors, any choice will disappear by antisymmetry. The Levi-Civita tensor can be removed from terms containing tensor Dirac structures by use of the duality relation

σρσ=i2ϵρσαβσαβ,(18)\sigma^{\rho\sigma}=\frac{i}{2}\epsilon^{\rho\sigma\alpha\beta}\sigma_{\alpha\beta}, \tag*{(18)}

along with the generalized Kronecker expression introduced earlier. For a vector or axial-vector Dirac factor, nonzero contributions must contract the vector’s index with the Levi-Civita tensor otherwise duplicated kinematic vectors are contracted and thereby 0 by antisymmetry. The other two indices must be contracted with PP and Δ\Delta. Then, by use of the identity

ϵρκαβγβ=−iγκγαγργ5+igκαγργ5−igκργαγ5+igαργκγ5(19)\epsilon^{\rho\kappa\alpha\beta}\gamma_{\beta}=-i\gamma^{\kappa}\gamma^{\alpha}\gamma^{\rho}\gamma_{5}+ig^{\kappa\alpha}\gamma^{\rho}\gamma_{5}-ig^{\kappa\rho}\gamma^{\alpha}\gamma_{5}+ig^{\alpha\rho}\gamma^{\kappa}\gamma_{5} \tag*{(19)}

such terms can be reduced to products of metric tensors and γ\gamma-matrices. For the terms that contain three γ\gamma-matrices, this results in terms like \slashed{P}\slashed{\Delta}. Such terms can be preemptively removed since when they are placed between spinors they can be reduced to single γ\gamma-terms by means of the Dirac equation after some reordering. Therefore, any term containing the Levi-Civita tensor can be recast into terms not containing the tensor or whose matrix elements evaluate to 0.

The Dirac equation for the total and momentum transfer are \bar{u}(p')\slashed{P}u(p)=m\bar{u}(p')u(p) and \bar{u}(p')\slashed{\Delta}u(p)=0, where for convenience here and in the following we drop the helicity argument. Therefore, terms proportional to \slashed{P} and \slashed{\Delta} can be absorbed into the coefficient functions aia_i or disappear altogether. The tensor structures that remain after removing combinations involving ϵαβγδ\epsilon^{\alpha\beta\gamma\delta} are

Γμν=a1(t)gμν+a2(t)PμPν+a3(t)ΔμΔν+a4(t)P{μΔν}+a5(t)γ{μPν}+a6(t)γ{μΔν}+a7(t)P{μσν}αΔα+a8(t)Δ{μσν}αPα+a9(t)P{μσν}αPα+a10(t)Δ{μσν}αΔα+a11(t)γ5gμν+a12(t)γ5PμPν+a13(t)γ5ΔμΔν+a14(t)γ5P{μΔν}+a15(t)γ5γ{μPν}+a16(t)γ5γ{μΔν}+a17(t)γ5P{μσν}αΔα+a18(t)γ5Δ{μσν}αPα+a19(t)γ5P{μσν}αPα+a20(t)γ5Δ{μσν}αΔα.(20)\begin{aligned} \Gamma^{\mu\nu}={}&a_1(t)g^{\mu\nu}+a_2(t)P^\mu P^\nu+a_3(t)\Delta^\mu\Delta^\nu+a_4(t)P^{\{\mu}\Delta^{\nu\}}+a_5(t)\gamma^{\{\mu}P^{\nu\}}+a_6(t)\gamma^{\{\mu}\Delta^{\nu\}}\\ &+a_7(t)P^{\{\mu}\sigma^{\nu\}\alpha}\Delta_\alpha+a_8(t)\Delta^{\{\mu}\sigma^{\nu\}\alpha}P_\alpha+a_9(t)P^{\{\mu}\sigma^{\nu\}\alpha}P_\alpha+a_{10}(t)\Delta^{\{\mu}\sigma^{\nu\}\alpha}\Delta_\alpha+a_{11}(t)\gamma_5g^{\mu\nu}\\ &+a_{12}(t)\gamma_5P^\mu P^\nu+a_{13}(t)\gamma_5\Delta^\mu\Delta^\nu+a_{14}(t)\gamma_5P^{\{\mu}\Delta^{\nu\}}+a_{15}(t)\gamma_5\gamma^{\{\mu}P^{\nu\}}+a_{16}(t)\gamma_5\gamma^{\{\mu}\Delta^{\nu\}}\\ &+a_{17}(t)\gamma_5P^{\{\mu}\sigma^{\nu\}\alpha}\Delta_\alpha+a_{18}(t)\gamma_5\Delta^{\{\mu}\sigma^{\nu\}\alpha}P_\alpha+a_{19}(t)\gamma_5P^{\{\mu}\sigma^{\nu\}\alpha}P_\alpha+a_{20}(t)\gamma_5\Delta^{\{\mu}\sigma^{\nu\}\alpha}\Delta_\alpha. \tag*{(20)} \end{aligned}

Notice the terms that appear alongside aia_i for i≥11i\ge11 are the P-violating terms.

Using the Gordon identity, uˉ(p′)γμu(p)=uˉ(p′)[Pμm+i2mσμΔ]u(p)\bar{u}(p')\gamma^\mu u(p)=\bar{u}(p')\left[\frac{P^\mu}{m}+\frac{i}{2m}\sigma^{\mu\Delta}\right]u(p) several of these terms can be further reduced. First, consider

uˉ(p′)P{μσν}Δu(p)=−2miuˉ(p′)P{μγν}u(p)+2iPμPνuˉ(p′)u(p).(21)\bar{u}(p')P^{\{\mu}\sigma^{\nu\}\Delta}u(p)=-2mi\bar{u}(p')P^{\{\mu}\gamma^{\nu\}}u(p)+2iP^\mu P^\nu\bar{u}(p')u(p). \tag*{(21)}

This shows that the tensor structures multiplying a2a_2 are related to those multiplying a5a_5 and a7a_7. A similar argument can be made to relate a4a_4 to the terms a6a_6 and a10a_{10}.

A slightly more involved argument is needed to show that the tensor structures multiplying a3a_3 and a8a_8 are related. The term multiplying a8a_8 can be simplified after expanding out σνΔP\sigma^{\nu\Delta}P and using the Dirac equation

\Delta^{\{\mu}\bar{u}(p')\sigma^{\nu\}P}u(p)=\frac{i\Delta^{\{\mu}}{4}\bar{u}(p')\left[\gamma^{\nu\}} \slashed{p}'-\slashed{p}'\gamma^{\nu\}}\right]u(p) =\frac{i}{2}\Delta^\mu\Delta^\nu\bar{u}(p')u(p). \tag*{(22)}

The second equality comes after rewriting the remaining terms using \gamma^{\nu}\slashed{p}'=\{\gamma^{\nu},\slashed{p}'\}-\slashed{p}'\gamma^{\nu} then simplifying with the Clifford algebra identity {γμ,γν}=2gμν\{\gamma^\mu,\gamma^\nu\}=2g^{\mu\nu} and Dirac equation. Therefore, a8a_8 is not independent of a3a_3 and can be removed from the remaining set of terms. A similar relationship holds between the tensor structures of a4a_4 and a9a_9 after replacing Δμ\Delta^\mu with PμP^\mu.

A symmetric set of arguments holds for the corresponding axial structures. By rewriting σμν=i(γμγν−gμν)\sigma^{\mu\nu}=i(\gamma^\mu\gamma^\nu-g^{\mu\nu}) and σμν=i(gμν−γνγμ)\sigma^{\mu\nu}=i(g^{\mu\nu}-\gamma^\nu\gamma^\mu) the axial Gordon identity

uˉ(p′)γ5[Pμm+i2mσμαΔα]u(p)=0(23)\bar{u}(p')\gamma^5\left[\frac{P^\mu}{m}+\frac{i}{2m}\sigma^{\mu\alpha}\Delta_\alpha\right]u(p)=0 \tag*{(23)}

is easily derived, and notably simpler than the standard case.

Proceeding in a similar way as above by focusing on the tensor structures multiplying a12a_{12}, it is related to the structure multiplying a17a_{17} by Eq. (23).

uˉ(p′)γ5P{μσν}αΔαu(p)=2iPμPνuˉ(p′)γ5u(p).(24)\bar{u}(p')\gamma^5P^{\{\mu}\sigma^{\nu\}\alpha}\Delta_\alpha u(p)=2iP^\mu P^\nu\bar{u}(p')\gamma^5u(p). \tag*{(24)}

Identical reasoning applies in relating the structures corresponding to a14a_{14} and a20a_{20}. In echo of a3a_3 the relation between a13a_{13} to other terms is less immediate, but by employing the same strategy one finds

uˉ(p′)γ5Δ{μσν}Pu(p)=iΔμ2uˉ(p′)γ5(2mγν+Δν)u(p).(25)\bar{u}(p')\gamma^5\Delta^{\{\mu}\sigma^{\nu\}P}u(p)=\frac{i\Delta^\mu}{2}\bar{u}(p')\gamma^5(2m\gamma^\nu+\Delta^\nu)u(p). \tag*{(25)}

Therefore, a13a_{13}, a16a_{16} and a18a_{18} are linearly dependent. The a14a_{14} structure is related to a15a_{15} and a19a_{19} by the same line of reasoning. Therefore, eight of the terms in Eq. (20) can be absorbed into the other ones. What remains from the original twenty is a set of twelve independent tensor structures, six of which correspond to factors independent of γ5\gamma_5 and are present in the standard θ=0\theta= 0 case. The other six are multiplied by γ5\gamma_5 and therefore parity-odd. They are only present in the decomposition because the hadronic states are not eigenstates of parity. To specify, we choose to retain the terms a1a_1, a3a_3, a5a_5, a6a_6, a7a_7, a10a_{10} to align with familiar decompositions [56]. The new corresponding terms kept are a11a_{11}, a13a_{13}, a15a_{15}, a16a_{16}, a17a_{17}, and a20a_{20}.

Hermiticity of the EMT matrix element,

⟨p′,s′∣Tμν∣p,s⟩†=⟨p,s∣Tμν∣p′,s′⟩(26)\langle p',s'|T^{\mu\nu}|p,s\rangle^\dagger= \langle p,s|T^{\mu\nu}|p',s'\rangle \tag*{(26)}

provides constraints on the real and imaginary components of the form factors. Of the remaining terms that are parity-conserving a1a_1, a3a_3, a5a_5 and a10a_{10} are clearly real-valued while a6a_6 and a7a_7 are purely imaginary. When taking the conjugate of the terms containing γ5\gamma_5, several factors of (−1)(-1) arise from anticommuting with γ5\gamma_5. Therefore, in contrast to the previous case, the terms corresponding to a11a_{11}, a13a_{13}, a16a_{16} and a20a_{20} are purely imaginary, while the terms corresponding to a15a_{15} and a17a_{17} are real. We introduce factors of ii into the tensor structures multiplying imaginary terms in order to keep all coefficient functions real-valued.

Consider now the consequences of the discrete symmetries. First note that while the θ\theta term is separately parity- and time-reversal odd, it is C-invariant and therefore PT remains conserved. Although parity and time reversal separately reverse the spatial momenta, their combined action leaves the four-momenta unchanged: pμ→pμp^\mu\to p^\mu, p′μ→p′μp'^\mu\to p'^\mu, Pμ→PμP^\mu\to P^\mu, Δμ→Δμ\Delta^\mu\to\Delta^\mu. Spin, however, is flipped under PTPT. So while factors of PP and Δ\Delta are unchanged, the external spinor labels are reversed. In the following, PT[ ]PT\lbrack\ \rbrack denotes the transformation after applying the spinor transformation laws and expressing the result in the original spinor basis. Because PTPT is antiunitary, invariance requires

PT[⟨f∣Tμν∣i⟩]=⟨f∣Tμν∣i⟩∗(27)PT\left[\langle f|T^{\mu\nu}|i\rangle\right] = \langle f|T^{\mu\nu}|i\rangle^* \tag*{(27)}

The coefficient functions depend only on tt, which remains unchanged under PTPT, and have been chosen real by hermiticity. Consequently, any remaining spinor bilinear must transform into its complex conjugate. Any component in the decomposition that instead transforms into minus its complex conjugate must have a vanishing coefficient. Covariants that then transform into minus their complex conjugate must vanish. Such minus signs will come from the transformation of the Dirac structures or the effect of commuting antiunitary operators through the factors of ii introduced to keep the coefficient functions real.

As an explicit example, consider the a6a_6 term,

PT[u‾(p′)a6(t)iγ{μΔν}u(p)]=−iu‾(p′)a6∗(t)γ{μΔν}u(p)(28)PT\left[\overline{u}(p')a_6(t)i\gamma^{\{\mu}\Delta^{\nu\}}u(p)\right] = -i\overline{u}(p')a_6^*(t)\gamma^{\{\mu}\Delta^{\nu\}}u(p) \tag*{(28)}

Since we know a6a_6 to be real, this combination is PTPT-odd requiring a6a_6 vanish. A similar argument holds for a10a_{10}, therefore

a6(t)=a10(t)=0.(29)a_6(t) = a_{10}(t) = 0. \tag*{(29)}

Now consider the terms containing a γ5\gamma_5 factor. This time the tensor structures multiplying a11a_{11}, a13a_{13}, a15a_{15}, and a16a_{16} are all odd under PTPT due to the additional γ5\gamma_5 factor. Whereas a17a_{17} and a20a_{20} are PT even. However, only the terms a11a_{11}, a13a_{13}, a16a_{16}, and a20a_{20} come with a compensating factor of ii. Therefore,

a15(t)=a20(t)=0.(30)a_{15}(t) = a_{20}(t) = 0. \tag*{(30)}

Altogether, the constraints from hermiticity and PT invariance reduce the number of independent form factors to the eight shown in Eq. (4).

The final reductions come from the conservation of the EMT, ∂μTμν=0\partial_\mu T^{\mu\nu} = 0. Applied to the matrix elements, this gives

Δμ⟨p′,s′∣Tμν∣p,s⟩=0.(31)\Delta_\mu\langle p',s'|T^{\mu\nu}|p,s\rangle= 0. \tag*{(31)}

Contracting with the remaining structures gives the following relations between the form factors

0=u‾(p′)ΔμΓμνu(p)=u‾(p′)[(a1+ta3)Δν+iγ5(a11+ta13)Δν+ia162γ5(−2mΔν+tγν)]u(p)(32)\begin{aligned} 0 &= \overline{u}(p')\Delta_\mu\Gamma^{\mu\nu}u(p) \\ &= \overline{u}(p')\left[(a_1+ta_3)\Delta^\nu+i\gamma_5(a_{11}+ta_{13})\Delta^\nu\right.\\ &\qquad\left.{}+\frac{ia_{16}}{2}\gamma_5(-2m\Delta^\nu+t\gamma^\nu)\right]u(p) \tag*{(32)} \end{aligned}

where terms containing P⋅ΔP\cdot\Delta or Δ\Delta but not γ5\gamma_5 have been removed since they vanish when contracted with spinors. Gordon identities have also been used to re-express terms in a more familiar form. This conservation equation must hold for every choice of external spins.

For t≠0t\ne0, the structures multiplying a16a_{16} do not point along the same directions as the other terms in spin-space. Then, for the conservation equation to hold, a16a_{16} must be 00. The remaining scalar and pseudoscalar bilinears are also independent in the external spin space, as can be easily confirmed by evaluating the spinor bilinears in the Breit frame. This means that their coefficients must vanish separately as well. In total, there are three constraints

a1(t)+ta3(t)=0,a11(t)+ta13(t)=0,a16(t)=0.(33)a_1(t)+ta_3(t)=0, \qquad a_{11}(t)+ta_{13}(t)=0, \qquad a_{16}(t)=0. \tag*{(33)}

reducing the number of independent form factors to 5.

Crossed-Channel JPCJ^{PC} Counting

As an independent check of the decomposition above, we can count the number of allowed form factors using the crossed-channel argument of Ji and Lebed [52], as reviewed for the EMT in regular QCD in Ref. [57]. Rather than explicitly constructing the tensor basis from the available Lorentz and Dirac structures, this method decomposes the rank-two operator into states of definite angular momentum and counts the number of corresponding hadron-antihadron partial waves. The two approaches therefore enumerate the same space of tensor structures, but in different bases.

Beginning with the off-forward matrix elements of the EMT, crossing symmetry relates these to the matrix elements of the EMT between the vacuum and a nucleon-antinucleon pair,

⟨p′,λ′∣Tμν∣p,λ⟩=⟨N(p′,λ′)Nˉ(−p,λ)∣Tμν∣0⟩.(34)\langle p',\lambda'|T^{\mu\nu}|p,\lambda\rangle = \langle N(p',\lambda')\bar{N}(-p,\lambda)|T^{\mu\nu}|0\rangle. \tag*{(34)}

In this channel, the momentum transfer of the elastic matrix element, Δ\Delta, becomes the total momentum of the NNˉN\bar{N} pair, while PP is related to their relative momentum. The counting can then be performed in the center-of-mass frame of the crossed channel, where Δμ=(t,0⃗)\Delta^\mu=(\sqrt{t},\vec{0}). Although this corresponds to a timelike value of tt, the number of independent form factors is unchanged by analytic continuation.

The general rank-two tensor transforms in the Lorentz representation as

(12,12)⊗(12,12)≃(0,0)⊕(1,0)⊕(0,1)⊕(1,1).(35)\left(\frac{1}{2},\frac{1}{2}\right)\otimes\left(\frac{1}{2},\frac{1}{2}\right) \simeq(0,0)\oplus(1,0)\oplus(0,1)\oplus(1,1). \tag*{(35)}

The (1,0)(1,0) and (0,1)(0,1) representations are antisymmetric and therefore do not contribute to the symmetric EMT considered here. Therefore, the symmetric EMT transforms as (0,0)⊕(1,1)(0,0)\oplus(1,1). Clearly, the (0,0)(0,0) representation contains a single scalar, J=0J=0, component. While the (1,1)(1,1) representation contains a single J=0J=0 component, a three-component J=1J=1 vector, and a five-component J=2J=2 symmetric-traceless tensor,

(1,1)≃0⊕1⊕2.(36)(1,1)\simeq0\oplus1\oplus2. \tag*{(36)}

This decomposition is visible by inspecting the components of the symmetric EMT. The component T00T^{00} is a rotational scalar and therefore corresponds to J=0J=0. The three components T0iT^{0i} transform as a vector and correspond to J=1J=1. Finally, the spatial tensor TijT^{ij} can be separated into its trace and symmetric-traceless pieces,

Tij=13δijTkk+(Tij−13δijTkk),(37)T^{ij}=\frac{1}{3}\delta^{ij}T^{kk}+\left(T^{ij}-\frac{1}{3}\delta^{ij}T^{kk}\right), \tag*{(37)}

which correspond to J=0J=0 and J=2J=2, respectively. Therefore, before imposing conservation, the symmetric EMT contains two J=0J=0 components, one J=1J=1 component, and one J=2J=2 component. The dimensions provide a simple check,

2(1)+3+5=10,(38)2(1)+3+5=10, \tag*{(38)}

which is the number of independent components of a symmetric rank-two tensor in four dimensions.

For the usual parity-conserving case, these components may be labelled by

(0,0):0++,(1,1):0++, 1−+, 2++.(39)\begin{aligned} (0,0)&:\quad0^{++},\\ (1,1)&:\quad0^{++},\ 1^{-+},\ 2^{++}. \tag*{(39)} \end{aligned}

The second superscript is fixed because the EMT is even under charge conjugation. For the case considered here, however, parity is not a symmetry of the theory. The JPCJ^{PC} labels remain useful for identifying the transformation properties of the possible NNˉN\bar{N} partial waves, but parity can no longer be imposed as a selection rule. Charge conjugation remains a symmetry of QCD with a θ\theta term, so the crossed-channel state must still have C=+C=+. Thus, for each angular momentum JJ appearing in Eq. (38), we must count all NNˉN\bar{N} partial waves with the same JJ and C=+C=+, irrespective of their parity.

For a fermion-antifermion pair, the parity and charge-conjugation quantum numbers are

P=(−1)L+1,(40)P=(-1)^{L+1}, \tag*{(40)}
C=(−1)L+S,(41)C=(-1)^{L+S}, \tag*{(41)}

where LL and SS are the orbital and total spin angular momenta, respectively. The allowed CC-even partial waves for J≤2J\leq2 are given in Table II. There are two independent J=0J=0 components of the symmetric tensor, and each can couple to either of the two J=0J=0, CC-even partial waves. These therefore give four independent amplitudes. Since parity is no longer a symmetry of the theory, the single J=1J=1 component can couple to one CC-even partial wave, while the single J=2J=2 component can couple to three. The total number of independent form factors before imposing conservation is therefore

JJNNˉN\bar{N} partial waves (LL)
00−+(0)0^{-+}(0), 0++(1)0^{++}(1)
11++(1)1^{++}(1)
22−+(2)2^{-+}(2), 2++(1)2^{++}(1), 2++(3)2^{++}(3)

Table 2. The CC-even spin-1/2 hadron-antihadron partial waves which can couple to the angular-momentum components of a symmetric rank-two tensor. The orbital quantum number is displayed for classification.

NGFF=(2×2)+(1×1)+(1×3)=8.(42)N_{\mathrm{GFF}}=(2\times2)+(1\times1)+(1\times3)=8. \tag*{(42)}

This agrees with the eight independent structures obtained from the explicit Lorentz and Dirac decomposition above after imposing hermiticity and the surviving discrete symmetries.

As a check, consider restoring parity invariance. The two J=0J=0 components of the EMT both have positive parity and can therefore couple only to the 0++0^{++} channel. The J=1J=1 component has quantum numbers 1−+1^{-+}, for which there is no corresponding spin-1/2 NNˉN\bar{N} partial wave. Finally, the 2++2^{++} component can couple to the 3P2{}^{3}P_{2} and 3F2{}^{3}F_{2} channels. The count then becomes

NGFFP-even=2+0+2=4,(43)N_{\mathrm{GFF}}^{P\text{-even}}=2+0+2=4, \tag*{(43)}

which is the familiar number of form factors for the symmetric EMT, without conservation. The additional four amplitudes in Eq. (42) are therefore precisely those made available by relaxing parity invariance.

We can now impose conservation directly within the crossed-channel counting. For the total EMT,

Δμ⟨NNˉ∣Tμν∣0⟩=0.(44)\Delta_{\mu}\langle N\bar{N}|T^{\mu\nu}|0\rangle=0. \tag*{(44)}

In the center-of-mass frame, Δμ=(t,0⃗)\Delta^\mu=(\sqrt{t},\vec{0}), so for generic non-zero tt this gives

⟨NNˉ∣T0ν∣0⟩=0.(45)\langle N\bar{N}|T^{0\nu}|0\rangle=0. \tag*{(45)}

Consequently, the T00T^{00} component and all of the T0iT^{0i} components are removed. In terms of angular momentum, conservation therefore removes one of the two J=0J=0 components and the complete J=1J=1 component. The remaining J=0J=0 component can couple to the two channels 0−+(L=0)0^{-+}(L=0) and 0++(1)0^{++}(1), while the J=2J=2 component can couple to 2−+(2)2^{-+}(2), 2++(1)2^{++}(1), and 2++(3)2^{++}(3). Therefore, the conserved symmetric EMT contains NGFFcons.=2+3=5N_{\mathrm{GFF}}^{\mathrm{cons.}}=2+3=5 independent form factors. Of these five amplitudes, the positive-parity channels 0++(1)0^{++}(1), 2++(1)2^{++}(1), and 2++(3)2^{++}(3) account for the three standard GFFs, and the negative-parity channels 0−+(0)0^{-+}(0) and 2−+(2)2^{-+}(2) account for the two additional structures allowed when parity is violated. Thus, the crossed-channel counting independently reproduces the result of the explicit covariant decomposition.

This also provides a useful interpretation of the three constraints obtained from EMT conservation in the previous section. Conservation removes one parity-even J=0J=0 amplitude, one parity-odd J=0J=0 amplitude, and the single CC-even J=1J=1 amplitude. This agrees with the three relations

a1(t)+ta3(t)=0,a11(t)+ta13(t)=0,a16(t)=0,(46)\begin{aligned} a_1(t)+t a_3(t)&=0,\\ a_{11}(t)+t a_{13}(t)&=0,\\ a_{16}(t)&=0, \tag*{(46)} \end{aligned}

obtained directly from the Lorentz decomposition. The JPCJ^{PC} counting provides a representation-theoretic enumeration of the same independent matrix elements and a useful check that the explicit tensor basis is both complete and minimal.

† e-mail: [email protected]

References

[1] F. J. Belinfante, On the current and the density of the electric charge, the energy, the linear momentum and the angular momentum of arbitrary fields, Physica 7, 449 (1940).

[2] L. Rosenfeld, On the energy-momentum tensor, Mem. Acad. Roy. Belg 18, N6 (1940).

[3] D. C. Hackett, D. A. Pefkou, and P. E. Shanahan, Gravitational Form Factors of the Proton from Lattice QCD, Phys. Rev. Lett. 132, 251904 (2024), arXiv:2310.08484 [hep-lat].

[4] S. Bhattacharya, K. Cichy, M. Constantinou, X. Gao, A. Metz, J. Miller, S. Mukherjee, P. Pete­rczky, F. Steffens, and Y. Zhao, Moments of proton GPDs from the OPE of nonlocal quark bilinears up to NNLO, Phys. Rev. D 108, 014507 (2023), arXiv:2305.11117 [hep-lat].

[5] H. Dutrieux, R. G. Edwards, C. Egerer, J. Karpie, C. Monahan, K. Orginos, A. Radyushkin, D. Richards, E. Romero, and S. Zafeiropoulos (Had-Struc), Towards unpolarized GPDs from pseudo-distributions, JHEP 08, 162, arXiv:2405.10304 [hep-lat].

[6] S. Bhattacharya, K. Cichy, M. Constantinou, X. Gao, A. Metz, J. Miller, S. Mukherjee, P. Pe-

References

  1. [1]F. J. Belinfante, On the current and the density of the electric charge, the energy, the linear momentum and the angular momentum of arbitrary fields, Physica 7, 449 (1940).DOI
  2. [2]L. Rosenfeld, On the energy-momentum tensor, Mem. Acad. Roy. Belg 18, N6 (1940).
  3. [3]D. C. Hackett, D. A. Pefkou, and P. E. Shanahan, Gravitational Form Factors of the Proton from Lattice QCD, Phys. Rev. Lett. 132, 251904 (2024), arXiv:2310.08484 [hep-lat].DOI
  4. [4]S. Bhattacharya, K. Cichy, M. Constantinou, X. Gao, A. Metz, J. Miller, S. Mukherjee, P. Petreczky, F. Steffens, and Y. Zhao, Moments of proton GPDs from the OPE of nonlocal quark bilinears up to NNLO, Phys. Rev. D 108, 014507 (2023), arXiv:2305.11117 [hep-lat].DOI
  5. [5]H. Dutreix, R. G. Edwards, C. Egerer, J. Karpie, C. Monahan, K. Orginos, A. Radyushkin, D. Richards, E. Romero, and S. Zafeiropoulos (Had-Struc), Towards unpolarized GPDs from pseudo-distributions, JHEP 08, 162, arXiv:2405.10304 [hep-lat].
  6. [6]S. Bhattacharya, K. Cichy, M. Constantinou, X. Gao, A. Metz, J. Miller, S. Mukherjee, P. Pe- treczky, F. Steffens, and Y. Zhao, Moments of axial-vector GPD from lattice QCD: quark helicity, orbital angular momentum, and spin-orbit correlation, JHEP 01, 146, arXiv:2410.03539 [hep-lat].
  7. [7]C. Alexandrou, S. Bacchio, M. Constantinou, C. Kummer, Y. Li, and G. Spanoudes, The nucleon unpolarized generalized form factors and Mellin moments up to fourth order, (2026), arXiv:2607.03458 [hep-lat].arxiv.org/abs/2607.03458
  8. [8]L. Di Luzio, M. Giannotti, E. Nardi, and L. Visinelli, The landscape of QCD axion models, Phys. Rept. 870, 1 (2020), arXiv:2003.01100 [hep-ph].DOI
  9. [9]C. Abel et al., Measurement of the Permanent Electric Dipole Moment of the Neutron, Phys. Rev. Lett. 124, 081803 (2020), arXiv:2001.11966 [hep-ex].DOI
  10. [10]R. D. Peccei and H. R. Quinn, CP Conservation in the Presence of Instantons, Phys. Rev. Lett. 38, 1440 (1977).
  11. [11]R. D. Peccei and H. R. Quinn, Constraints Imposed by CP Conservation in the Presence of Instantons, Phys. Rev. D 16, 1791 (1977).
  12. [12]R. D. Peccei, The Strong CP problem and axions, Lect. Notes Phys. 741, 3 (2008), arXiv:hep-ph/0607268.DOI
  13. [13]S. Navas et al. (Particle Data Group), Review of particle physics, Phys. Rev. D 110, 030001 (2024).
  14. [14]M. Burkardt, Impact parameter dependent parton distributions and transverse single spin asymmetries, Phys. Rev. D 66, 114005 (2002), arXiv:hep-ph/0209179.DOI
  15. [15]R. G. Edwards, J. Karpie, C. Monahan, K. Orginos, A. Radyushkin, D. Richards, E. Romero, and S. Zafeir opoulos, Femtoscale imaging of the proton with Ioffe-time distributions, (2026), arXiv:2608.23737 [hep-ph].arxiv.org/abs/2608.23737
  16. [16]D. Müller, D. Robaschik, B. Geyer, F. M. Dittes, and J. Hořejši, Wave functions, evolution equations and evolution kernels from light ray operators of QCD, Fortsch. Phys. 42, 101 (1994), arXiv:hep-ph/9812448.
  17. [17]X.-D. Ji, Deeply virtual Compton scattering, Phys. Rev. D 55, 7114 (1997), arXiv:hep-ph/9609381.DOI
  18. [18]A. V. Radyushkin, Scaling limit of deeply virtual Compton scattering, Phys. Lett. B 380, 417 (1996), arXiv:hep-ph/9604317.DOI
  19. [19]A. V. Radyushkin, Asymmetric gluon distributions and hard diffractive electroproduction, Phys. Lett. B 385, 333 (1996), arXiv:hep-ph/9605431.DOI
  20. [20]J. C. Collins, L. Frankfurt, and M. Strikman, Factorization for hard exclusive electroproduction of mesons in QCD, Phys. Rev. D 56, 2982 (1997), arXiv:hep-ph/9611433.DOI
  21. [21]J.-W. Qiu and Z. Yu, Single diffractive hard exclusive processes for the study of generalized parton distributions, Phys. Rev. D 107, 014007 (2023), arXiv:2210.07995 [hep-ph].DOI
  22. [22]J. Dudek et al., Physics Opportunities with the 12 GeV Upgrade at Jefferson Lab, Eur. Phys. J. A 48, 187 (2012), arXiv:1208.1244 [hep-ex].DOI
  23. [23]A. Accardi et al., Strong interaction physics at the luminosity frontier with 22 GeV electrons at Jefferson Lab, Eur. Phys. J. A 60, 173 (2024), arXiv:2306.09300 [nucl-ex].
  24. [24]T. Sawada, W.-C. Chang, S. Kumano, J.-C. Peng, S. Sawada, and K. Tanaka, Accessing proton generalized parton distributions and pion distribution amplitudes with the exclusive pion-induced Drell-yan process at j-parc, Physical Review D 93, 10.1103/physrevd.93.114034 (2016).DOI
  25. [25]F. Gautheron et al. (COMPASS), COMPASS-II Proposal, Tech. Rep. (CERN, Geneva, 2010).
  26. [26]B. Adams et al., COMPASS++/AMBER: Proposal for Measurements at the M2 beam line of the CERN SPS Phase-1: 2022-2024, Tech. Rep. (CERN, Geneva, 2019) the collaboration has not yet constituted itself, thus instead of a Spokesperson currently the nominated Contact Person is acting in place.
  27. [27]A. Accardi et al., Electron Ion Collider: The Next QCD Frontier: Understanding the glue that binds us all, Eur. Phys. J. A 52, 268 (2016), arXiv:1212.1701 [nucl-ex].
  28. [28]J.-W. Chen, H.-W. Lin, and J.-H. Zhang, Pion generalized parton distribution from lattice QCD, Nucl. Phys. B 952, 114940 (2020), arXiv:1904.12376 [hep-lat].DOI
  29. [29]C. Alexandrou, K. Cichy, M. Constantinou, K. Hadjiyiannakou, K. Jans en, A. Scapellato, and F. Steffens, Unpolarized and helicity generalized parton distributions of the proton within lattice QCD, Phys. Rev. Lett. 125, 262001 (2020), arXiv:2008.10573 [hep-lat].DOI
  30. [30]H.-W. Lin, Nucleon Tomography and Generalized Parton Distribution at Physical Pion Mass from Lattice QCD, Phys. Rev. Lett. 127, 182001 (2021), arXiv:2008.12474 [hep-ph].DOI
  31. [31]S. Bhattacharya, K. Cichy, M. Constantinou, J. Dodson, X. Gao, A. Metz, S. Mukherjee, A. Scapellato, F. Steffens, and Y. Zhao, Generalized parton distributions from lattice QCD with asymmetric momentum transfer: Unpolarized quarks, Phys. Rev. D 106, 114512 (2022), arXiv:2209.05373 [hep-lat].DOI
  32. [32]S. Bhattacharya, K. Cichy, M. Constantinou, J. Dodson, A. Metz, A. Scapellato, and F. Steffens, Chiral-even axial twist-3 GPDs of the proton from lattice QCD, Phys. Rev. D 108, 054501 (2023), arXiv:2306.05533 [hep-lat].arxiv.org/abs/2306.05533
  33. [33]H.-W. Lin, Pion valence-quark generalized parton distribution at physical pion mass, Phys. Lett. B 846, 138181 (2023), arXiv:2310.10579 [hep-lat].DOI
  34. [34]S. Bhattacharya, K. Cichy, M. Constantinou, A. Metz, N. Nurminen, and F. Steffens, Generalized parton distributions from the pseudodistribution approach on the lattice, Phys. Rev. D 110, 054502 (2024), arXiv:2405.04414 [hep-lat].DOI
  35. [35]H.-T. Ding, X. Gao, S. Mukherjee, P. Petreczky, Q. Shi, S. Syritsyn, and Y. Zhao, Three-dimensional imaging of pion using lattice QCD: generalized parton distributions, JHEP 02, 056, arXiv:2407.03516 [hep-lat].DOI
  36. [36]S. Bhattacharya, K. Cichy, M. Constantinou, A. Metz, J. Miller, P. Petreczky, and F. Steffens, Generalized parton distributions from lattice QCD with asymmetric momentum transfer: Tensor case, Phys. Rev. D 112, 114504 (2025), arXiv:2505.11288 [hep-lat].arxiv.org/abs/2505.11288
  37. [37]M.-H. Chu, M. Colaço, S. Bhattacharyya, K. Cichy, M. Constantinou, A. Metz, and F. Steffens, Generalized parton distributions from lattice QCD with asymmetric momentum transfer: Unpolarized quarks at nonzero skewness, Phys. Rev. D 112, 094510 (2025), arXiv:2508.17998 [hep-lat].DOI
  38. [38]H. Dutrieux, R. G. Edwards, K. Jarpie, C. Mezrag, C. Monahan, K. Orginos, A. Radyushkin, D. Richards, E. Romero, and S. Zafeiropoulos, Reconstructing the full kinematic dependence of GPDs from pseudo-distributions, (2026), arXiv:2604.21476 [hep-lat].arxiv.org/abs/2604.21476
  39. [39]K. Kumerički, S. Liuti, and H. Moutarde, GPD phenomenology and DVCS fitting: Entering the high-precision era, Eur. Phys. J. A 52, 157 (2016), arXiv:1602.02763 [hep-ph].DOI
  40. [40]K. Kumerički and D. Müller, Extraction of generalized parton distributions from deeply virtual Compton scattering data, AIP Conf. Proc. 1819, 050004 (2017).DOI
  41. [41]Y. Guo, F. P. Aslan, X. Ji, and M. G. Santiago, First Global Extraction of Generalized Parton Distributions from Experiment and Lattice Data with Next-to-Leading-Order Accuracy, Phys. Rev. Lett. 135, 261903 (2025), arXiv:2509.08037 [hep-ph].
  42. [42]J. Xu et al., Neural Network Representation of Generalized Parton Distributions (NNGPD), (2026), arXiv:2605.06994 [hep-ph].arxiv.org/abs/2605.06994
  43. [43]C. Mezrag, P. Sznajder, and J. Wagner, Extraction of DVCS amplitudes off the nucleon, (2026), arXiv:2608.20244 [hep-ph].arxiv.org/abs/2608.20244
  44. [44]I. Y. Kobzarev and L. B. Okun’, Gravitational Interaction of Fermions, Zh. Eksp. Teor. Fiz. 43, 1904 (1962).
  45. [45]K. L. Ng, Gravitational form-factors of the neutrino, Phys. Rev. D 47, 5187 (1993), arXiv:vir-qc/9305002.DOI
  46. [46]Z.-B. Kang and D. E. Kharzeev, Quark fragmentation in the Θ-vacuum, Phys. Rev. Lett. 106, 042001 (2011), arXiv:1006.2132 [hep-ph].DOI
  47. [47]C.-Y. Seng, Relating hadronic CP-violation to higher-twist distributions, Phys. Rev. Lett. 122, 072001 (2019), arXiv:1809.00307 [hep-ph].DOI
  48. [48]W. Yang, Parity-odd parton distribution functions from θ-vacuum, Int. J. Mod. Phys. A 34, 1950145 (2019), arXiv:1907.00129 [hep-ph].DOI
  49. [49]A. Bacchetta, M. Cerutti, L. Manna, M. Radici, and X. Zheng, Signals of strong parity violation in deep inelastic scattering, Phys. Lett. B 849, 138455 (2024), arXiv:2306.04704 [hep-ph].arxiv.org/abs/2306.04704
  50. [50]A. V. Radyushkin, Double distributions and pseudo-distributions, Phys. Rev. D 109, 014514 (2024), arXiv:2311.06007 [hep-ph].arxiv.org/abs/2311.06007
  51. [51]J. Schoenleber, R. S. Sufian, T. Izubuchi, and Y.-B. Yang, Gluon unpolarized, polarized, and transversity GPDs from lattice QCD: Lorentz-covariant parametrization, Phys. Rev. D 111, 094510 (2025), arXiv:2412.14110 [hep-lat].arxiv.org/abs/2412.14110
  52. [52]X.-D. Ji and R. F. Lebed, Counting form-factors of twist-two operators, Phys. Rev. D 63, 076005 (2001), arXiv:hep-ph/0012160.DOI
  53. [53]H. Sun, T. Tan, and J.-H. Yu, Covariant Construction of Generalized Form Factors, (2026), arXiv:2604.25771 [hep-ph].arxiv.org/abs/2604.25771
  54. [54]J.-W. Qiu, N. Sato, and Z. Yu, Unified and optimal frame choice for generalized parton distributions, Phys. Rev. D 111, 094014 (2025), arXiv:2409.06882 [hep-ph].DOI
  55. [55]J. Grames et al., Positron Beams At Ce⁺BAF, in 14th International Particle Accelerator Conference (2023) arXiv:2309.15581 [physics.acc-ph].arxiv.org/abs/2309.15581
  56. [56]X.-D. Ji, Decomposition of Nucleon Spin, Phys. Rev. Lett. 78, 610 (1997), arXiv:hep-ph/9603249.
  57. [57]W. Cosyn, S. Cotogno, A. Freese, and C. Lorcé, The energy-momentum tensor of spin-1 hadrons: formalism, Eur. Phys. J. C 79, 476 (2019), arXiv:1903.00408 [hep-ph].DOI

Paper details

vlnc:W_9jh6kc56
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