---
title: Diquark Spectator Model Overview
url: https://www.emergentmind.com/topics/diquark-spectator-model
type: topic
---

# Diquark Spectator Model Overview

Searching arXiv for recent and foundational papers on the diquark spectator model and closely related applications.
The diquark spectator model is an effective quark–diquark description in which a hadron is approximated as an active quark plus a spectator diquark remnant. In the literature, this framework appears both in a minimal scalar diquark realization and in broader spectator constructions that include axial-vector or vector diquarks, light-cone wave functions, and phenomenological vertex regulators. It has been used to study standard and quasi parton distributions, transverse-momentum-dependent distributions, generalized parton distributions, generalized and gravitational form factors, fragmentation functions, and angular-momentum observables for the nucleon, hyperons, and spin-\(\tfrac32\) baryons [1808.01437, 1412.3401, 1408.4873].

## 1. Core construction and spectator approximation

In its simplest field-theoretic form, the scalar diquark spectator model is a relativistic Yukawa-type theory containing a spin-\(\tfrac12\) hadron field \(\Psi\), a quark field \(\psi\), and a scalar diquark field \(\varphi\),
\[
{\cal L}_{\rm SDM}
= \bar{\Psi} \big(i \slashed{\partial}-M\big)\Psi
+ \bar{\psi} \big(i \slashed{\partial}-m_q\big)\psi
+ \frac{1}{2}\big(\partial_\mu\varphi\,\partial^\mu\varphi-m_s^2\varphi^2\big)
+ g\big(\bar{\Psi}\psi\varphi+\bar{\psi}\Psi\varphi\big),
\]
with the stability condition \(M<m_s+m_q\) [1808.01437]. In this form, the model deliberately retains only the lowest nontrivial \({\cal O}(g^2)\) contribution, neglecting virtual diagrams that contribute only at \(x=1\) [1808.01437].

The defining spectator approximation consists of inserting a complete set of intermediate states into the quark correlator and truncating it to a single on-shell diquark state. In nucleon applications this state is taken to be either a scalar diquark \(X=s\) or an axial-vector diquark \(X=a\), so that the hadron effectively fluctuates as \(N\to q+X\) [1412.3401, 1511.05242]. In spectator language, the active quark is the line coupled to the probe, while the diquark carries the remainder of the hadron momentum and quantum numbers.

A closely related Abelian implementation is the QED-like scalar diquark model used for angular-momentum decomposition. There the proton is modeled as a quark plus scalar diquark with a point-like quark–nucleon–diquark vertex of coupling \(g\), and photon couplings to the quark and spectator denoted \(e_q\) and \(e_s\). A key assumption is that the target is electromagnetically neutral, emulating color neutrality, even though \(e_q\) and \(e_s\) are kept as independent parameters in the displayed formulas [1907.05268].

## 2. Model variants, vertices, and spin structure

The diquark spectator model is not a single unique ansatz but a family of related constructions that differ mainly in the spectator spin assignment, vertex structure, and ultraviolet regulator. A common covariant choice uses a scalar vertex \(i\,g_s(k^2)\) and an axial-vector vertex
\[
i\,\frac{g_a(k^2)}{\sqrt{2}}\,\gamma^\mu\gamma_5,
\]
with a dipole form factor
\[
g_X(k^2)=g_X\,\frac{k^2-m^2}{(k^2-\Lambda_X^2)^2}
= g_X\,\frac{(k^2-m^2)(1-x)^2}{\big(\bm k_T^2+L_X^2(\Lambda_X^2)\big)^2}
\]
for \(X=s,a\) [2206.06937]. This regulator suppresses large virtualities and renders transverse-momentum integrals finite.

Other applications use different nonperturbative dressings. In \(\Lambda\)-hyperon fragmentation, the hyperon–quark–diquark vertex is taken as
\[
\Upsilon_s=g_s\,\mathbf 1,\qquad
\Upsilon_v^\mu=\frac{g_v}{\sqrt3}\,\gamma_5\left(\gamma^\mu+\frac{P_\Lambda^\mu}{M_\Lambda}\right),
\]
with \(g_s=g_v=g_D\), and the ultraviolet region is controlled by a Gaussian form factor
\[
g_D(k^2)=\frac{g_D'}{z}\exp\!\left[-\frac{k^2}{\lambda^2 z^\alpha(1-z)^\beta}\right]
\]
[1706.03355, 1802.01843]. In proton twist-3 GPD calculations, an exponential form factor is used at the nucleon–quark–diquark vertex specifically to regularize the transverse-momentum integration and to drive all distributions to zero as \(|x|\to0\) and \(|x|\to1\) [2407.17448]. In a light-cone spectator-diquark model for nucleon generalized form factors, the momentum dependence is encoded through the Brodsky–Huang–Lepage ansatz
\[
\phi_D(x,\mathbf k_\perp)
= A_D\exp\!\left\{-\frac{1}{8\beta_D^2}
\left[\frac{m_q^2+\mathbf k_\perp^2}{x}
+\frac{m_D^2+\mathbf k_\perp^2}{1-x}\right]\right\}
\]
for scalar and axial-vector spectators [1408.4873].

The treatment of the spin-1 spectator also varies. Some calculations include both transverse and longitudinal polarization states of the axial-vector diquark [2601.21430], whereas the proton twist-3 GPD study keeps only the physical light-cone transverse polarization states through a specific polarization tensor [2407.17448]. In still other quasi-TMD applications the simplification \(d^{\mu\nu}=-g^{\mu\nu}\) is adopted for the axial-vector propagator, with the explicit caveat that unphysical polarizations are then included [2206.06937]. In light-cone spectator-diquark models, relativistic spin effects are incorporated by Melosh–Wigner rotation for both the quark and the spin-1 diquark, so that instant-form spin states are converted into light-cone helicity amplitudes before overlap calculations are performed [1408.4873].

## 3. TMDs, rescattering, and fragmentation observables

A major use of spectator-diquark models is the analytic calculation of transverse-momentum-dependent distributions. In a scalar-plus-axial spectator framework with dipolar vertices, the model yields explicit expressions for the T-even leading-twist TMDs \(f_1\), \(g_{1L}\), \(g_{1T}\), \(h_1\), \(h_{1L}^\perp\), and \(h_{1T}^\perp\), and generates the Sivers and Boer–Mulders functions by expanding the gauge link to first order, i.e. through one-gluon exchange between the struck quark and the spectator [1003.1328]. This same framework was then used to construct weighted SIDIS and Drell–Yan asymmetries, with the transverse-momentum weights collapsing TMD convolutions into products of moments [1003.1328].

In a spin-1 spectator realization of SIDIS, the proton is treated as a two-body light-front state consisting of a spin-\(\tfrac12\) quark and a spin-1 diquark. There the single-spin asymmetry arises from the interference of a real tree amplitude with a one-loop amplitude containing a final-state gluon exchange; the required imaginary phase follows from an on-shell pole,
\[
\operatorname{Im}\left(\frac{1}{P^-_{\rm init}-P^-_{\rm interm}+i\epsilon}\right)
= -\pi\,\delta(P^-_{\rm init}-P^-_{\rm interm}),
\]
and the resulting model gives explicit nonzero Sivers and Boer–Mulders functions through helicity-dependent light-front wave-function overlaps [1504.03784]. In that calculation the two functions are proportional, differing by a factor of \(2\) in the displayed formulas, and both exhibit a node controlled by the factor \(M-\frac{m}{x}\), with sign reversal near \(x\sim0.3\) for the chosen parameters [1504.03784].

The same spectator logic has been extended to twist-3. In a scalar diquark model for the quark–gluon–quark correlator, eight time-reversal-odd interaction-dependent twist-3 TMDs are extracted from the Dirac decomposition of \(\tilde\Phi_A^\alpha(x,p_T)\). A dipole form factor,
\[
\lambda(p^2)=\frac{N_s(p^2-m^2)}{(p^2-\Lambda^2)^2},
\]
is essential: with a point-like coupling, only \(\tilde g^\perp\) and \(\tilde f_L^\perp\) remain finite, while the other six distributions diverge [1202.0700]. The same model reproduces the ETQS–Sivers identity
\[
e_q T_{q,F}(x,x)=-2M\,f_{1T}^{\perp(1)}(x),
\]
but does not satisfy the exact QCD equations-of-motion relations between interaction-dependent and interaction-independent twist-3 TMDs, which is explicitly attributed to the lowest-order truncation [1202.0700].

Fragmentation observables are treated analogously by replacing the initial baryon with a final hadron and a spectator remnant. For \(\Lambda\) production, the spectator model with scalar and axial-vector diquarks fits \(D_1^\Lambda\) to the DSV parametrization at \(\mu_{LO}^2=0.23~\mathrm{GeV}^2\), using
\[
g_D=1.983,\qquad m_D=0.745~\mathrm{GeV},\qquad \lambda=5.967~\mathrm{GeV},\qquad \alpha=0.5,\qquad \beta=0,
\]
and finds \(G_1^\Lambda\) to be strange-quark dominated, while the naive T-odd \(D_{1T}^\perp\) is driven mainly by \(u,d\) quarks and yields a negative transverse polarization \(P_T^\Lambda\) in both SIDIS and single-inclusive \(e^+e^-\) annihilation [1706.03355]. The \(\Lambda\) Collins function \(H_1^\perp\) has also been computed with scalar and vector spectators and the same Gaussian vertex structure; in that case one-gluon-loop rescattering and eikonal-line contributions generate the T-odd phase, QCD evolution is implemented for the first transverse moment, and the weighted \(\cos2\phi_0\) asymmetry in \(e^+e^-\to\Lambda\bar\Lambda X\) is found to be positive, sizable, and sensitive to evolution at \(Q=10.52~\mathrm{GeV}\) [1802.01843].

## 4. GPDs, generalized form factors, and baryon structure

Spectator-diquark models provide a compact framework for off-forward observables. In a light-cone spectator-diquark model with scalar and axial-vector spectators, plus Melosh–Wigner rotation for both the quark and the axial-vector diquark, the twist-two generalized form factors at \(n=2\) are computed from overlaps of light-cone wave functions. The resulting \(u\)- and \(d\)-quark contributions to \(A(Q^2)\), \(B(Q^2)\), \(\tilde A(Q^2)\), \(A_T(Q^2)\), \(\tilde A_T(Q^2)\), and \(B_T(Q^2)+2\tilde B_T(Q^2)\) are reported to be comparable with lattice QCD results [1408.4873].

At twist \(3\), a proton spectator model with scalar and axial-vector diquarks and an exponential vertex form factor has been used to calculate the chiral-even GPDs at nonzero skewness. The correlator is evaluated by contour integration in \(k^-\), with the contributing pole depending on whether \(x\) lies in the ERBL or DGLAP region. A central result is that almost all twist-3 GPDs are discontinuous at \(x=\pm\xi\), with \(G_1\) and \(\tilde G_1\) as the explicit exceptions; the forward limit yields the twist-3 PDF \(g_T\), and the same calculation compares kinetic orbital angular momentum and spin-orbit correlations as extracted from twist-2 and twist-3 GPDs [2407.17448].

The framework has also been pushed beyond the nucleon. For strange and non-strange baryons, a quark–diquark spectator model with scalar and axial-vector channels computes gravitational form factors from the second Mellin moments of chiral-odd GPDs,
\[
A_{T20}(t)=\int dx\,x\,H_T(x,\xi,t),\qquad
B_{T20}(t)=\int dx\,x\,E_T(x,\xi,t),
\]
and related relations for \(\tilde H_T\) and \(\tilde E_T\) [2601.21430]. In the reported \(u\)-quark sector, \(A_{T20}(0)=0.445\) for the proton and \(A_{T20}(0)=-0.106\) for \(\Xi^0\), while \(\bar B_{T20}(0)=1.057\) for the proton; this is presented as a complementary sign pattern tied to flavor dependence in transversity-related GPD moments [2601.21430].

For the spin-\(\tfrac32\) \(\Delta^+\), the spectator approach generates the full leading-twist set of eight unpolarized and eight polarized GPDs and verifies the required time-reversal symmetry under \(\xi\to-\xi\) [2305.02680]. In the forward limit, the structure functions are obtained from \(H_1\), \(H_4\), \(\tilde H_5\), and \(\tilde H_8\), and after evolution from \(Q_0=0.528~\text{GeV}\) to \(Q=4~\text{GeV}\) the model predicts, for example, a decrease of the struck \(d\)-quark longitudinal momentum fraction in \(F_1^d\) from about \(37\%\) to about \(25\%\) [2305.02680]. A later extension to transversity GPDs introduces \(16\) independent spin-\(\tfrac32\) tensor GPDs; with the simplifying choice \(c_2=c_3=0\), only \(H_1^{qT}\), \(H_3^{qT}\), \(H_5^{qT}\), and \(H_7^{qT}\) survive in practice, and the model gives the \(d\)-quark tensor charge in \(\Delta^+\) as
\[
\delta d=2\,[G_1^{qT}(0)-G_2^{qT}(0)]=0.876,
\]
with \(g_T^{\Delta^+}=0.876\) quoted for the isobar [2508.15245].

## 5. Quasi-distributions and finite-momentum extrapolation

A distinct branch of spectator-model work concerns quasi-distributions, i.e. equal-time spatial correlators intended as lattice-QCD proxies for light-cone observables. For standard collinear PDFs one replaces the light-cone bilocal operators by spatial correlators along \(z\),
\[
\tilde f_1(x,P_z)=\int \frac{d\xi^z}{4\pi}\,e^{-ixP_z\xi^z}
\langle P|\bar\psi(\xi^z)\gamma^z U_{n_z}[\xi^z,0]\psi(0)|P\rangle,
\]
with analogous definitions for \(\tilde g_1\) and \(\tilde h_1\) [1511.05242]. In the spectator model these quasi-PDFs have support on the entire real axis, reduce analytically to the standard distributions as \(P_z\to\infty\), and provide a controlled estimate of the finite-boost systematics relevant for lattice QCD [1412.3401, 1511.05242].

The main numerical conclusion is consistent across the two quasi-PDF studies: at intermediate momentum fractions, roughly \(0.1\lesssim x\lesssim0.4\text{--}0.5\), quasi-PDFs are within about \(20\text{--}30\%\) of the standard PDFs once \(P_z\gtrsim1.5\text{--}2~\mathrm{GeV}\), whereas near \(x\sim1\) much larger boosts, \(P_z\gtrsim4~\mathrm{GeV}\), are required for satisfactory agreement [1412.3401, 1511.05242]. The breakdown at large \(x\) is traced to the spectator on-shell kinematics, specifically to factors of \((1-x)^2P_z^2\), and the same studies emphasize that the Soffer positivity bound should not be assumed for quasi-PDFs at finite \(P_z\) [1511.05242].

Generalized quasi-distributions behave similarly. In the scalar diquark model, quasi-GPDs are defined from spatial correlators with either \(\gamma^0\) or \(\gamma^3\), remain continuous functions of \(x\), and reduce analytically to the standard GPDs as \(P^3\to\infty\) [1808.01437]. Finite-momentum mismatch is largest near \(x\to0\), \(x\to1\), and in narrow ERBL regions, and the moment analysis of the extended twist-2 study shows that the \(\Gamma^3\) definitions are preferred for second moments related to quark total angular momentum, because the \(\Gamma^0\) projection receives an additional higher-twist contamination from the \(D^q\) form factor [1903.05721].

The quasi-program has also been extended to T-odd TMDs. In a spectator model with scalar and axial-vector diquarks, the quasi Sivers and quasi Boer–Mulders functions are defined by replacing the light-cone projectors \(\gamma^+\) and \(i\sigma^{i+}\gamma_5\) with \(\gamma_z\) and \(i\sigma^{iz}\gamma_5\). The resulting quasi-distributions reduce analytically to the standard TMDs when \(P_z\to\infty\), and their first transverse moments are reported to be fair approximations to the standard ones, again at the \(20\text{--}30\%\) level, in the region \(0.1<x<0.5\) when \(P_z\ge2.5\text{--}3~\mathrm{GeV}\) [2206.06937].

## 6. Angular momentum, spectator torque, and known limitations

One of the most detailed conceptual uses of the scalar diquark spectator model concerns the distinction between Ji and Jaffe–Manohar angular-momentum decompositions. The Ji relation for total angular momentum,
\[
J^a=\frac12\int dx\,x\big[H^a(x,0,0)+E^a(x,0,0)\big],
\]
is contrasted with the canonical Jaffe–Manohar definition through the “potential” term
\[
(L)^{\text{Ji}}-(L)^{\text{JM}}
= -e_q\langle \bar\psi(r)\,\mathbf r\times \mathbf A_{\text{phys}}(r)\,\psi(r)\rangle
\]
[1907.05268]. In the scalar diquark model this difference vanishes at one loop,
\[
(L)^{\text{Ji}}=(L)^{\text{JM}}
=\langle\bar\psi(r)\,\mathbf r\times(-i\nabla)\,\psi(r)\rangle,
\]
but appears at two loops at order \({\cal O}(g^2 e_q e_s)\), supporting the interpretation that the struck quark acquires a spectator-induced torque through the gauge-link structure and the same rescattering mechanism that produces the Sivers effect [1907.05268].

The spectator framework has also been used to test factorization itself. In a scalar diquark model for inclusive DIS at sub-asymptotic \(Q^2\), the hadronic tensor can be computed exactly and compared with a collinear-factorized approximation built with an inclusive jet function. This analysis argues that a mass-corrected scaling variable,
\[
\xi_q=\xi\left(1+\frac{m_q^2}{Q^2}\right),
\]
extends the region of validity of leading-twist collinear factorization beyond what is obtained with Bjorken \(x_B\), but also shows an intrinsic breakdown near threshold due to exact four-momentum conservation, with failure beginning roughly when \(W^2\lesssim4~\text{GeV}^2\) [2010.07339]. The same study emphasizes that the decisive limitation is kinematic: in the exact model, \(k_T\) is bounded by phase space, whereas in the factorized approximation the transverse integration extends to infinity [2010.07339].

Across applications, several limitations recur. These models are low-scale effective descriptions, not full nonperturbative QCD calculations; they commonly require dipole, Gaussian, exponential, or BHL-type regulators to suppress ultraviolet sensitivity; and lowest-order truncations can violate exact field-theoretic identities. Concrete examples include the failure of exact QCD equations-of-motion relations at twist \(3\) [1202.0700], approximate or explicit violations of positivity bounds at large \(z\) for T-odd \(\Lambda\)-fragmentation functions [1706.03355, 1802.01843], and the non-guaranteed status of Soffer positivity for quasi-PDFs at finite \(P_z\) [1511.05242]. These features do not nullify the spectator approach; rather, they delimit its role as a controlled phenomenological laboratory in which quark–spectator dynamics, rescattering phases, and finite-momentum effects can be computed explicitly.

Source: https://www.emergentmind.com/topics/diquark-spectator-model