Rapidity-Modified Ji Identities

• Rapidity-Modified Ji Identities are generalized angular-momentum relations that extend the standard Ji sum rule by incorporating off-forward kinematics and rapidity gaps.
• They encompass formulations such as moving-frame transverse extensions, finite-skewness GPD moments, and staple-dependent GTMD/Wigner deformations to interpolate between different OAM definitions.
• The framework clarifies distinctions between Belinfante and canonical decompositions and emphasizes the roles of rapidity, skewness, and gauge-link geometry in modifying spin observables.

Searching arXiv for recent and foundational papers on rapidity-dependent extensions of Ji relations, transverse Ji analogues, and GTMD/Wigner OAM interpolations. “Rapidity-Modified Ji Identities” denotes a family of extensions, reinterpretations, and operator deformations of the Ji angular-momentum relation in which the standard forward, zero-skewness, straight-link setting is replaced by either boosted transverse kinematics, finite-skewness off-forward kinematics, or gauge-link geometries carrying TMD-like rapidity structure. The common theme is that the usual Ji decomposition,

$J_a=\frac12\int dx\,x\,[H_a(x,0,0)+E_a(x,0,0)],$

is no longer read as the unique statement about partonic angular momentum once one departs from the diagonal proton matrix element at $(\eta,t)=(0,0)$, introduces transverse polarization in a moving state, or replaces the straight Wilson line by staple-shaped links. In the literature, these generalizations do not form a single formalism: one line of work derives an energy-dependent transverse analogue of Ji’s relation, another formulates exact finite-skewness GPD moment identities with an explicit rapidity gap, and a third constructs GTMD/Wigner OAM observables that interpolate continuously between Ji and Jaffe-Manohar definitions through link geometry and Collins–Soper-type parameters (Leader, 2011, Hechenberger et al., 24 Jul 2025, Engelhardt, 2017).

1. Standard Ji relation and the meaning of its modification

The conventional Ji identity is the relation for the longitudinal component of Belinfante quark angular momentum in a longitudinally polarized nucleon,

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$

In this form, “quark” means quark plus antiquark of a given flavor, the relation is obtained from matrix elements of the Belinfante energy-momentum tensor, and the separate quark and gluon pieces are renormalization-scale dependent. A recurrent clarification in the literature is that this formula concerns a specified component of angular momentum and a specified operator definition; it is not merely a frame-free statement about an undifferentiated “quark angular momentum” (Leader, 2011).

A modification of the Ji identity arises whenever one preserves the same basic GPD moment structure but changes the kinematics or operator content. Three distinct modifications recur. The first is a moving-frame transverse extension, in which the transverse component of angular momentum acquires explicit dependence on the nucleon energy $P_0$. The second is a finite-skewness generalization, in which the relevant moments are evaluated at $\eta\neq0$ and $t=-c_\eta$, with

$$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$

so that the object being decomposed is an off-forward parton–nucleon correlation rather than the spin of a single proton state. The third is an operator deformation by Wilson-line geometry, in which a straight link gives Ji OAM and a staple link gives Jaffe-Manohar OAM, with finite staple extent and finite Collins–Soper parameter defining an interpolating family rather than a closed sum rule (Hechenberger et al., 24 Jul 2025, Engelhardt, 2017, Engelhardt et al., 2020).

This multiplicity of meanings is essential. In this literature, “rapidity-modified” may refer to rapidity as a skewness-induced longitudinal gap, to TMD-style off-light-cone staple structure regulated by a Collins–Soper variable, or, more loosely, to boost dependence that can be re-expressed through rapidity. The phrase therefore names a research direction rather than a single canonical identity.

2. Transverse moving-frame extension of Ji’s relation

A rigorous transverse analogue of Ji’s familiar longitudinal relation was derived for the expectation value of the transverse component of Belinfante quark angular momentum in a transversely polarized moving nucleon,

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$

A similar relation holds for gluons,

$$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$

The defining novelty is the explicit factor of the nucleon energy $P_0$, which makes the separate transverse quark and gluon contributions frame dependent. The derivation uses the Belinfante angular-momentum operator, the EMT parameterization in terms of $(\eta,t)=(0,0)$0 and $(\eta,t)=(0,0)$1, and the identifications

$(\eta,t)=(0,0)$2

In the rest frame, where $(\eta,t)=(0,0)$3, the transverse relation reduces to the Ji form, consistent with the absence of an essential distinction between transverse and longitudinal directions there (Leader, 2011).

The same work emphasizes that this is not a rapidity-modified identity in the modern TMD sense. It does not invoke Wilson lines with rapidity regulators, Collins–Soper evolution, or the Pauli–Lubanski construction. Nevertheless, because a nucleon moving along $(\eta,t)=(0,0)$4 obeys

$(\eta,t)=(0,0)$5

the explicit energy factor can be re-expressed in terms of rapidity $(\eta,t)=(0,0)$6. This suggests a boost-dependent or rapidity-parametrized extension of Ji’s relation, even though rapidity language is not used in the paper itself. The total transverse spin remains protected: summing over flavors and gluons, the $(\eta,t)=(0,0)$7-dependent term cancels through

$(\eta,t)=(0,0)$8

so that the total transverse angular momentum is still

$(\eta,t)=(0,0)$9

The literature therefore treats this result as a transverse, moving-frame extension of Ji’s identity rather than a replacement for it (Leader, 2011).

3. Finite-skewness rapidity-dependent spin decomposition

A more literal use of the expression “rapidity-modified Ji identities” appears in a finite-skewness GPD framework in which the transverse Fourier transform of a GPD is interpreted differently at $$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$0 and $$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$1. At zero skewness it gives the familiar impact-parameter density, whereas at finite skewness it is taken to encode a genuine parton–nucleon correlation whose norm decreases with the rapidity gap

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$2

The resulting decomposition keeps the Ji moment structure but relocates it from the forward proton spin budget to the angular-momentum content of an off-forward correlation (Hechenberger et al., 24 Jul 2025).

For each parton species $$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$3, the relevant finite-skewness quantities are defined by

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$4

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$5

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$6

with

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$7

Summed over quarks and gluons, the rapidity-modified Ji identity becomes

$$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$8

At $$\langle J_L^{\mathrm{bel}(\text{quark})}\rangle = \frac12\left[ \int_{-1}^{1}dx\,x\,E_q(x,0,0) + \int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$9, one recovers the ordinary Ji decomposition and $P_0$0. At finite $P_0$1, the total is no longer fixed to $P_0$2; instead it decreases as the rapidity gap grows, reflecting the depletion of the off-forward correlation rather than a change in the spin of a single proton state (Hechenberger et al., 24 Jul 2025).

The same framework expresses the integrated norm of the transverse-space correlation as

$P_0$3

This norm is the quantity said to decrease with $P_0$4. The formal derivation uses exact Mellin or conformal moments of $P_0$5, $P_0$6, and $P_0$7, specialized to $P_0$8, so $P_0$9. The paper further states that D-term contributions proportional to $\eta\neq0$0 cancel between the unpolarized and helicity-flip form factors in the combination entering $\eta\neq0$1, and that the $\eta\neq0$2 form factors vanish for the singlet, so that

$\eta\neq0$3

In this formulation, the rapidity modification is neither a regulator artifact nor a simple frame effect; it is an intrinsic feature of off-forward kinematics interpreted through a rapidity gap between initial and final nucleon states (Hechenberger et al., 24 Jul 2025).

4. GTMD/Wigner formulations and staple-dependent deformations of Ji OAM

A different but closely related development defines quark OAM directly from a Wigner distribution,

$\eta\neq0$4

where the gauge link $\eta\neq0$5 is part of the operator definition. In this language, a straight Wilson line yields Ji OAM, while a staple-shaped Wilson line yields Jaffe-Manohar OAM. The phase-space correlator is built from GTMD matrix elements with a soft factor $\eta\neq0$6, and the gauge-link geometry introduces the same rapidity and self-energy divergences familiar from TMD factorization. The staple direction is therefore taken off the light cone, with associated Collins–Soper-type parameter

$\eta\neq0$7

The gauge link is

$\eta\neq0$8

so that $\eta\neq0$9 gives the straight-link Ji limit and $t=-c_\eta$0 gives the long-staple Jaffe-Manohar limit. The resulting family $t=-c_\eta$1 is therefore staple- and rapidity-dependent, but it is not a modified Ji sum rule in GPD moments; it is a continuous operator interpolation in GTMD/Wigner phase space (Engelhardt, 2017).

The exploratory lattice calculation of this operator family was performed at the pion mass $t=-c_\eta$2 MeV and was reported to quasi-continuously interpolate between the Ji and Jaffe-Manohar definitions. The difference was clearly resolved, and Jaffe-Manohar OAM was found to be enhanced in magnitude compared to its Ji counterpart. For the isovector $t=-c_\eta$3 combination, the integrated torque observable $t=-c_\eta$4 was found to be of order $t=-c_\eta$5 in the large-$t=-c_\eta$6 extrapolation, with the effect tending to grow with $t=-c_\eta$7. In this interpretation, the Ji–Jaffe-Manohar difference is the torque accumulated by the outgoing quark through final-state interactions encoded by the staple path, and is associated with a genuine twist-three Qiu–Sterman-type correlator (Engelhardt, 2017).

A later lattice study sharpened the same program by replacing the earlier finite-difference estimate of the $t=-c_\eta$8-derivative with a direct derivative method. Its working formula for longitudinal quark OAM was

$t=-c_\eta$9

with the nonlocal bilocal matrix element

$$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$0

The direct derivative removed the significant numerical bias seen previously, and the value obtained for Ji quark OAM was reconciled with an independent lattice extraction from Ji’s sum rule. On a $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$1 clover ensemble with $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$2, $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$3, $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$4 gauge configurations, $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$5 data samples, and $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$6, the isovector Jaffe-Manohar OAM was found to be about $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$7 larger in magnitude than Ji OAM for the two nonzero values of $$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$8 (Engelhardt et al., 2020).

These GTMD studies are central to the broader topic because they show how a straight-link Ji observable can be embedded in a larger gauge-invariant operator family carrying explicit staple length and Collins–Soper dependence. Their “rapidity modification” is thus operator- and path-based rather than a closed-form finite-$$c_\eta=\frac{4\eta^2 m_N^2}{1-\eta^2}, \qquad \Delta y=2\,\operatorname{artanh}\eta,$$9 GPD identity.

5. Small-$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$0 gluon OAM and analytic interpolation between Ji and Jaffe-Manohar

For gluons at small $$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$1, an analytic interpolation between Ji-type and Jaffe-Manohar-type OAM can be built with a generalized light-cone Wilson line gauge link. The framework uses the Chen et al. decomposition,

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$2

and a finite-extent staple link

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$3

whose parameter limits reproduce both infinite staple and near-straight geometries. In this setting, the distinction between Ji and Jaffe-Manohar again comes from the gauge-link geometry: the straight or near-straight limit approaches Ji, while future- or past-pointing infinite staples give the Jaffe-Manohar operator structure (Abir et al., 2021).

The associated gluon Wigner distribution is

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$4

and gluon OAM is defined by the phase-space moment

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$5

A crucial operator identity is the derivative of the generalized link, which in Chen variables becomes

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$6

This formula is the mechanism by which link-extent dependence enters the OAM operator and makes the Ji/Jaffe-Manohar interpolation explicit (Abir et al., 2021).

At small $$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$7, the OAM distribution is expanded as

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$8

The eikonal term vanishes,

$$\langle J_T^{\mathrm{bel}(\text{quark})}\rangle = \frac{1}{2M} \left[ P_0\int_{-1}^{1}dx\,x\,E_q(x,0,0) + M\int_{-1}^{1}dx\,x\,H_q(x,0,0) \right].$$9

so the first nonzero gluon OAM appears only at sub-eikonal order. This feature makes the path-dependent operator content especially consequential: the finite-link deformation is not merely formal, but enters at the first nontrivial order in the small-$$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$0 regime. In this literature, the interpolation is best understood as a generalized Wilson-line-regulated extension of Ji-type gluon OAM rather than as a standalone rapidity-evolution equation (Abir et al., 2021).

6. Conceptual distinctions, caveats, and adjacent terminology

Several distinctions are necessary to prevent conflation of formally different constructions. First, Belinfante and canonical decompositions are not interchangeable. The transverse relation derived from GPD moments concerns Belinfante angular momentum, whereas the transversity-plus-OAM decomposition associated with Jaffe-Manohar relies on canonical operators. For the total angular momentum the two coincide, but not for separate quark or gluon pieces. This matters whenever Ji-type formulas are compared to staple-link OAM observables (Leader, 2011).

Second, the symbol $$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$1 is overloaded across the literature. In finite-skewness GPD decompositions, $$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$2 is the skewness parameter and determines the rapidity gap $$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$3. In GTMD/Wigner and lattice staple-link calculations, $$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$4 denotes the staple extent. These are conceptually distinct deformations: one changes off-forward kinematics, the other changes Wilson-line geometry. The two are related only at the level of broad analogy, not by a direct identification (Hechenberger et al., 24 Jul 2025, Engelhardt, 2017).

Third, not every rapidity-sensitive operator identity in QCD is a Ji identity. A separate line of work on subleading-power resummation introduces “rapidity identity operators” required by rapidity-renormalization-group consistency, with operators such as

$$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$5

and evolution constrained by

$$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$6

That framework does not discuss Ji’s sum rule, EMT decompositions, or GPD moment identities directly; its relevance is structural rather than explicit. It shows that rapidity-sensitive problems can require new operator-level identities and nontrivial mixing, but it does not itself produce a rapidity-modified Ji relation (Moult et al., 2019).

Finally, experimental and interpretive status differs across frameworks. The finite-skewness GPD identities are presented as exact analytic relations within the GPD formalism but require modeled or reconstructed GPDs for numerical realization. The GTMD/Wigner observables are gauge invariant and lattice accessible, but finite-$$\langle J_T^{\mathrm{bel}(\text{gluon})}\rangle = \frac{1}{2M} \left[ P_0\int_{0}^{1}dx\,x\,E_G(x,0,0) + M\int_{0}^{1}dx\,x\,H_G(x,0,0) \right].$$7, finite staple length, and matching limitations mean that they are not closed-form generalizations of Ji’s sum rule. The transverse moving-frame identity is rigorous but concerns expectation values of components of Belinfante angular momentum that are not directly measurable. Across all these cases, the most precise encyclopedic characterization is that rapidity-modified Ji identities are generalized Ji-type angular-momentum relations whose modification comes from off-forward rapidity gap, boosted transverse kinematics, or rapidity-sensitive gauge-link structure, with the standard Ji identity recovered in the appropriate forward, rest-frame, or straight-link limit (Hechenberger et al., 24 Jul 2025, Engelhardt et al., 2020).