---
title: Leading-Twist Quark Wigner Distributions
url: https://www.emergentmind.com/topics/leading-twist-quark-wigner-distributions
type: topic
---

# Leading-Twist Quark Wigner Distributions

Leading-twist quark Wigner distributions are light-front phase-space distributions of the nucleon that depend on the longitudinal momentum fraction $x$, the quark transverse momentum $\boldsymbol{k}_\perp$, and the transverse impact parameter $\boldsymbol{b}_\perp$. They are obtained as transverse Fourier transforms of generalized transverse-momentum dependent distributions (GTMDs), and thereby combine in a single picture the information separately contained in transverse-momentum dependent distributions (TMDs) and generalized parton distributions (GPDs). At leading twist, they form a complete set of real-valued but non-positive-definite quasi-distributions, with 16 independent polarization structures for a spin-$1/2$ target. Their main theoretical significance is that they expose spin-orbit and spin-spin correlations inaccessible in pure GPD or TMD projections, and they provide a direct phase-space route to quark orbital angular momentum (OAM) [1106.0139] [1208.3065].

## 1. Operator definition and light-front kinematics

The standard construction starts from the off-forward quark correlator at fixed light-front time $z^+=0$,
$$
W^{[\Gamma]q}_{\Lambda'\Lambda}(P,x,\boldsymbol{k}_\perp,\Delta,n)
=\frac{1}{2}\int\frac{dz^-\,d^2\boldsymbol{z}_\perp}{(2\pi)^3}\,
e^{ixP^+z^- - i\boldsymbol{k}_\perp\cdot\boldsymbol{z}_\perp}
\,\langle p',\Lambda'|\,\bar\psi\!\left(-\tfrac{z}{2}\right)\Gamma\,\mathcal W\,\psi\!\left(\tfrac{z}{2}\right)\,|p,\Lambda\rangle\Big|_{z^+=0},
$$
with $P=(p'+p)/2$, $\Delta=p'-p$, and $x=k^+/P^+$. The Wilson line $\mathcal W$ ensures color gauge invariance and is specified by a path choice, typically straight or staple-like [1208.3065].

The five-dimensional Wigner distribution is then defined by Fourier transforming the transverse momentum transfer to impact-parameter space,
$$
\mathcal{W}^{[\Gamma]}_{\Lambda'\Lambda}(x,\boldsymbol{k}_\perp,\boldsymbol{b}_\perp;n)
=\int\frac{d^2\boldsymbol{\Delta}_\perp}{(2\pi)^2}\,
e^{-i\boldsymbol{\Delta}_\perp\cdot\boldsymbol{b}_\perp}\,
W^{[\Gamma]}_{\Lambda'\Lambda}(P,x,\boldsymbol{k}_\perp,\Delta;n).
$$
In the infinite-momentum frame, or equivalently the Drell–Yan–West frame with $\Delta^+=0$, these objects admit a particularly clean interpretation because transverse boosts form a Galilean subgroup, allowing transversely localized target states without additional relativistic ambiguities [1106.0139].

The leading-twist quark bilinears are
$$
\Gamma=\gamma^+,\qquad \gamma^+\gamma_5,\qquad i\sigma^{i+}\gamma_5\quad (i=1,2),
$$
corresponding respectively to unpolarized, longitudinally polarized, and transversely polarized quark probes. Although GTMDs are generally complex, their two-dimensional Fourier transforms are real, consistent with a phase-space interpretation. The resulting Wigner distributions are nevertheless quasi-probabilistic rather than genuine probabilities, because the underlying position and momentum operators do not commute [1208.3065].

## 2. Leading-twist classification and polarization content

For a spin-$1/2$ target, leading twist contains 16 independent Wigner distributions covering all target/quark polarization combinations. In the longitudinal sector, a convenient decomposition is
$$
\rho_{\Lambda\lambda}(\boldsymbol{b}_\perp,\boldsymbol{k}_\perp,x)
=\tfrac{1}{2}\big[\rho_{UU}+\Lambda\,\rho_{LU}+\lambda\,\rho_{UL}+\Lambda\lambda\,\rho_{LL}\big],
$$
with $\Lambda,\lambda=\pm1$ denoting target and quark longitudinal polarizations [1106.0139].

In terms of GTMDs, the longitudinal channels are
$$
\begin{aligned}
\rho_{UU}(\boldsymbol{b}_\perp,\boldsymbol{k}_\perp,x)
&= \mathcal{F}_{1,1}(x,0,\boldsymbol{k}_\perp^2,\boldsymbol{k}_\perp\!\cdot\!\boldsymbol{b}_\perp,\boldsymbol{b}_\perp^2),\\
\rho_{LU}(\boldsymbol{b}_\perp,\boldsymbol{k}_\perp,x)
&= -\frac{1}{M^2}\,\epsilon_\perp^{ij} k^i_\perp \frac{\partial}{\partial b^j_\perp}\,
\mathcal{F}_{1,4}(x,0,\boldsymbol{k}_\perp^2,\boldsymbol{k}_\perp\!\cdot\!\boldsymbol{b}_\perp,\boldsymbol{b}_\perp^2),\\
\rho_{UL}(\boldsymbol{b}_\perp,\boldsymbol{k}_\perp,x)
&= \frac{1}{M^2}\,\epsilon_\perp^{ij} k^i_\perp \frac{\partial}{\partial b^j_\perp}\,
\mathcal{G}_{1,1}(x,0,\boldsymbol{k}_\perp^2,\boldsymbol{k}_\perp\!\cdot\!\boldsymbol{b}_\perp,\boldsymbol{b}_\perp^2),\\
\rho_{LL}(\boldsymbol{b}_\perp,\boldsymbol{k}_\perp,x)
&= \mathcal{G}_{1,4}(x,0,\boldsymbol{k}_\perp^2,\boldsymbol{k}_\perp\!\cdot\!\boldsymbol{b}_\perp,\boldsymbol{b}_\perp^2).
\end{aligned}
$$
This makes explicit that $\rho_{LU}$ is controlled by $F_{1,4}$ and $\rho_{UL}$ by $G_{1,1}$ [1106.0139].

A recurring feature of the literature is that notation is not fully uniform. Some papers use $\rho_{LU}$ for unpolarized quarks in a longitudinally polarized target, while others note that some literature uses $\rho_{UL}$ for the same physical configuration; the operator definition removes the ambiguity [1107.3984].

Beyond the longitudinal sector, the full leading-twist set can be organized by basic multipoles in transverse momentum and transverse position space. The complete T-even and T-odd multipole decomposition shows that each polarization channel carries a definite spin-spin or spin-orbit content: monopoles encode unpolarized or helicity densities, dipoles encode net transverse or azimuthal flows, and quadrupoles encode transversity or pretzelous-type structures [1512.06744]. This multipole language is especially useful because it simultaneously clarifies which structures survive in TMD or GPD projections and which require the full phase-space description.

## 3. Relations to GTMDs, TMDs, and GPDs

Wigner distributions are the impact-parameter representation of GTMDs, and GTMDs are their purely momentum-space counterparts. The standard reduction relations are
$$
\int d^2\boldsymbol{b}_\perp\,\mathcal{W}^{[\Gamma]}(x,\boldsymbol{k}_\perp,\boldsymbol{b}_\perp;S)
=\text{TMD}^{[\Gamma]}(x,\boldsymbol{k}_\perp;S),
$$
and
$$
\int d^2\boldsymbol{k}_\perp\,\mathcal{W}^{[\Gamma]}(x,\boldsymbol{k}_\perp,\boldsymbol{b}_\perp;S)
=\int\frac{d^2\boldsymbol{\Delta}_\perp}{(2\pi)^2}
e^{-i\boldsymbol{\Delta}_\perp\cdot \boldsymbol{b}_\perp}\,
\text{GPD}^{[\Gamma]}(x,0,-\boldsymbol{\Delta}_\perp^2;S).
$$
Thus, integrating over $\boldsymbol{b}_\perp$ yields the forward limit of GTMDs, i.e. TMDs, while integrating over $\boldsymbol{k}_\perp$ yields impact-parameter GPDs [1208.3065].

Some GTMDs are genuine “mother distributions” in the sense that both their TMD and GPD limits survive. In particular,
$$
f_1(x,\boldsymbol{k}_\perp^2)=F_{1,1}(x,0,\boldsymbol{k}_\perp^2,0,0),
\qquad
g_{1L}(x,\boldsymbol{k}_\perp^2)=G_{1,4}(x,0,\boldsymbol{k}_\perp^2,0,0),
$$
while their $\boldsymbol{k}_\perp$ integrals generate $H$ and $\tilde H$. By contrast, integrating over either $\boldsymbol{b}_\perp$ or $\boldsymbol{k}_\perp$ kills $\rho_{LU}$ and $\rho_{UL}$, showing that $F_{1,4}$ and $G_{1,1}$ encode correlations absent from any leading-twist TMD or GPD alone [1106.0139].

The symmetry constraints are correspondingly strong. Hermiticity enforces the reality of the Wigner distributions, parity fixes the allowed tensor structures, and time reversal interacts with the gauge-link geometry. With a straight lightlike Wilson line, naively T-odd components vanish; with a staple-like Wilson line, naively time-reversal odd structures are allowed and encode effects such as the Sivers and Boer–Mulders functions [1208.3065].

A central conceptual limitation concerns observability. The leading-twist GTMD combination associated with the unpolarized-quark/longitudinally polarized-proton configuration is the $F_{14}$ term,
$$
\bar{u}(p',\Lambda') \frac{i \sigma^{ij} \bar{k}_T^i \Delta_T^j}{M^2} u(p,\Lambda)\,F_{14},
$$
which is proportional to $\langle {\bf S}_L \cdot \bar{\bf k}_T \times {\bf \Delta}_T \rangle$. Parity constraints imply that this structure is parity-odd and nonzero only for imaginary values of the helicity amplitudes. In coplanar tree-level kinematics it therefore cannot contribute at leading twist to deeply virtual Compton scattering (DVCS) without final-state interactions; moreover, integrating $F_{14}$ over $\bar{\bf k}_T$ gives zero, so it drops out of GPDs [1310.5157]. This is the main reason why OAM-sensitive observables in DVCS appear at twist three rather than directly through leading-twist $F_{14}$.

## 4. Orbital angular momentum and competing definitions

The most prominent application of leading-twist quark Wigner distributions is the extraction of the longitudinal quark OAM from the phase-space average of $\boldsymbol{b}_\perp\times\boldsymbol{k}_\perp$,
$$
L_z^q=\int dx\,d^2\boldsymbol{k}_\perp\,d^2\boldsymbol{b}_\perp\,
\big(\boldsymbol{b}_\perp\times \boldsymbol{k}_\perp\big)_z\,
\rho^{[\gamma^+]q}_{\Lambda\Lambda}(x,\boldsymbol{k}_\perp,\boldsymbol{b}_\perp;n).
$$
In impact-parameter language, this projects precisely the $F_{1,4}$ structure [1208.3065].

For a straight gauge link one obtains the compact relation
$$
L_z^q=-\int dx\,d^2\boldsymbol{k}_\perp\,
\frac{\boldsymbol{k}_\perp^2}{M^2}\,
F_{1,4}^q\big(x,\xi=0,\boldsymbol{k}_\perp^2,\boldsymbol{k}_\perp\!\cdot\!\boldsymbol{\Delta}_\perp=0,\boldsymbol{\Delta}_\perp^2=0\big).
$$
The path choice of the Wilson line determines the OAM notion. A straight lightlike Wilson line yields kinetic OAM in the sense of Ji’s decomposition, whereas a staple-like Wilson line yields canonical OAM in the sense of Jaffe–Manohar [1208.3065]. Earlier model analyses also emphasized that in the absence of explicit gauge-field degrees of freedom these distinctions can collapse for the flavor-summed total even while flavor-separated values differ [1106.0139].

The Ji sum rule gives the kinetic OAM through twist-2 GPDs,
$$
L_z^q=\frac{1}{2}\int_{-1}^{1}dx\,\Big\{
x\big[H^q(x,0,0)+E^q(x,0,0)\big]-\tilde H^q(x,0,0)\Big\}.
$$
The same quantity also appears in twist-three form. Using the twist-three parametrization of Meissner–Metz–Schlegel, Lorcé and Pasquini derived
$$
-\int dx\,x\,\tilde E_{2T}(x,0,0)=L_z^q+2S_z^q,
$$
which is equivalent to
$$
\int dx\,x\,G_2(x,0,0)=-L_z^q.
$$
These relations connect longitudinal OAM to the Penttinen–Polyakov–Shuvaev–Strikman sum rule and clarify why twist-three GPDs are central to observability [1208.3065].

A frequently discussed proxy is the pretzelosity-based expression
$$
\mathcal{L}_z^q=-\int dx\,d^2\boldsymbol{k}_\perp\,
\frac{\boldsymbol{k}_\perp^2}{2M^2}\,
h_{1T}^{\perp q}(x,\boldsymbol{k}_\perp^2),
$$
but the literature explicitly stresses that no rigorous expression of OAM in terms of leading-twist TMDs is known [1208.3065].

The observability analysis sharpens this distinction further. In DVCS, the OAM-relevant helicity structure is accessed through twist-three GTMDs and GPDs, especially $\widetilde E_{2T}$ or $G_2$, rather than through leading-twist $F_{14}$. In the Wandzura–Wilczek approximation,
$$
L_q^{WW}(x,0,0)
=
x \int_x^1 \frac{dy}{y} \big(H_q(y,0,0)+E_q(y,0,0)\big)
-
x \int_x^1 \frac{dy}{y^2} \widetilde{H}_q(y,0,0),
$$
while beyond Wandzura–Wilczek the difference between kinetic and canonical OAM is attributed to final-state interactions and gauge-link structure [1310.5157].

## 5. Model calculations and tomographic patterns

The earliest detailed pictures were obtained in valence-quark light-front models. In the light-front constituent quark model (LFCQM), the $x$-integrated distribution of unpolarized quarks in a longitudinally polarized nucleon exhibits a clear dipole pattern in impact-parameter space at fixed transverse momentum, for example at $\boldsymbol{k}_\perp=k_\perp\,\boldsymbol{e}_y$ with $k_\perp=0.3\,\mathrm{GeV}$. For $u$ quarks the dipole indicates OAM aligned with the proton helicity, whereas for $d$ quarks the distortion tends to favor OAM anti-aligned with the proton helicity. The average transverse momentum field,
$$
\langle\boldsymbol{k}_\perp\rangle^q(\boldsymbol{b}_\perp)
=
\int dx\,d^2\boldsymbol{k}_\perp\,
\boldsymbol{k}_\perp\,
\rho^{[\gamma^+]q}_{\Lambda\Lambda}(x,\boldsymbol{k}_\perp,\boldsymbol{b}_\perp;n),
$$
shows a vortex-like flow, and for $d$ quarks the sign of the OAM density changes at about $0.25$ fm from the transverse center. A similar qualitative picture is obtained in the light-front chiral quark-soliton model (LF$\chi$QSM) [1208.3065].

The same papers compared three OAM extractions in the valence sector: kinetic OAM from Ji’s sum rule, canonical OAM from the Wigner integral, and the pretzelosity proxy. The reported values are:

| Model | Definition | Values |
|---|---|---|
| LFCQM | Kinetic (Ji) | $L_u=0.071,\ L_d=0.055,\ L_{u+d}=0.126$ |
| LFCQM | Canonical (Jaffe–Manohar) | $\ell_u=0.131,\ \ell_d=-0.005,\ \ell_{u+d}=0.126$ |
| LFCQM | TMD-based proxy | $\mathcal{L}_u=0.169,\ \mathcal{L}_d=-0.042,\ \mathcal{L}_{u+d}=0.126$ |
| LF$\chi$QSM | Kinetic (Ji) | $L_u=-0.008,\ L_d=0.077,\ L_{u+d}=0.069$ |
| LF$\chi$QSM | Canonical (Jaffe–Manohar) | $\ell_u=0.073,\ \ell_d=-0.004,\ \ell_{u+d}=0.069$ |
| LF$\chi$QSM | TMD-based proxy | $\mathcal{L}_u=0.093,\ \mathcal{L}_d=-0.023,\ \mathcal{L}_{u+d}=0.069$ |

In both models, the total quark OAM coincides across definitions, whereas flavor-separated values do not. In the LFCQM both $u$ and $d$ kinetic OAM are positive, while canonical and TMD-based extractions give a negative $d$-quark OAM. In the LF$\chi$QSM, kinetic OAM gives a small negative $u$ and positive $d$, closer in sign to lattice indications for the isovector combination [1208.3065].

Models with explicit gluonic degrees of freedom alter this pattern. In the dressed-quark calculation of Mukherjee, Nair, and Ojha, performed in light-front gauge with the gauge link set to unity, $\rho_{UU}$ peaks at the origin and $\rho_{LU}$ exhibits a clear dipole structure. In that model both kinetic and canonical OAM are negative, and their magnitudes differ once gluons are included, in contrast to quark-only models [1409.7238]. Related perturbative dressed-quark analyses found a negative longitudinal spin-orbit correlation, with $F_{14}=-G_{11}$ and consequently $C_z^q=-l_z^q$ [1511.00922].

The model landscape is broader. Light-cone spectator models compute all 16 leading-twist distributions for $u$ and $d$ quarks and emphasize strong directional dependence in the transverse-spin sectors, while soft-wall AdS/QCD-inspired quark-diquark models report circular $\rho_{UU}$ and $\rho_{LL}$ distributions, dipolar $\rho_{UL}$ and $\rho_{LU}$, quadrupolar pretzelous structures, and a Soffer-bound-type inequality for Wigner distributions [1501.07690] [1701.08551].

## 6. Multipoles, skewness, and longitudinal extensions

A major later development was the complete multipole decomposition of the transverse phase space. This framework classifies every leading-twist Wigner distribution into basic multipoles in $\boldsymbol{k}_T$ and $\boldsymbol{b}_T$, separately in the T-even and T-odd sectors. It provides a transparent interpretation of the spin-spin and spin-orbit correlations encoded in each channel, and it shows how impact-parameter GPD multipoles and TMD multipoles arise as projections of the more general phase-space patterns [1512.06744].

Nonzero skewness introduces an additional layer of structure. In a light-front dressed-quark model, increasing $\xi$ produces characteristic distortions in the spatial and momentum correlations, including dipole and quadrupole patterns, asymmetries, and localization effects. These features were attributed to spin-orbit correlations and quantum interference between light-front wave-function components with differing orbital angular momentum. The forward-limit symmetry progressively deforms as $\xi$ increases, and mixed-space ridges sharpen, indicating stronger localization in phase space [2507.18168].

A complementary line of work introduces a boost-invariant longitudinal coordinate, either $\sigma=\tfrac12 b^-P^+$ or $\tilde z=b^-P^+$, as the Fourier conjugate of the skewness. In this formulation, the Fourier transform of $\xi$-dependent GTMDs defines Wigner distributions in longitudinal position space, which exhibit diffraction patterns analogous to optical single-slit interference. This extends the usual five-dimensional picture to a six-dimensional light-front phase space $(\tilde z,x,\boldsymbol{b}_\perp,\boldsymbol{k}_\perp)$ and retains the same 16 independent leading-twist polarization structures [2202.08635] [2505.06615].

These generalized constructions sharpen two points. First, the standard five-dimensional distributions are not the only possible light-front phase-space representation; they are the $\xi=0$ sector of a broader off-forward geometry. Second, while Wigner distributions themselves are not directly measurable, their $\xi$ dependence and twist-three projections create experimentally relevant observables. In DVCS, the twist-three effective axial Compton form factor enters the $\sin 2\phi$ modulation of the longitudinal target-spin asymmetry,
$$
A_{UL}=\frac{a\sin\phi+b\sin 2\phi}{c_0+c_1\cos\phi+c_2\cos 2\phi},
$$
with the coefficient $b$ controlled by $\widetilde{\mathcal H}^{\mathrm{eff}}$ and therefore sensitive to $\widetilde E_{2T}$ and OAM-associated GTMD/Wigner content. HERMES reported a sizable $\sin 2\phi$ modulation, enabling an extraction of $G_2\equiv \widetilde E_{2T}$ within the Wandzura–Wilczek approximation [1310.5157].

Taken together, these developments establish leading-twist quark Wigner distributions as the central phase-space language for nucleon tomography. They unify GPD and TMD information, isolate the GTMD structures that carry spin-orbit content, and provide the most direct leading-twist description of how transverse position and transverse momentum correlations encode quark orbital motion. At the same time, the literature makes clear that gauge-link geometry, parity constraints, and twist-three dynamics are not technical details but defining ingredients of the subject.

Source: https://www.emergentmind.com/topics/leading-twist-quark-wigner-distributions