McLerran–Venugopalan Model

• The McLerran–Venugopalan model is a semi-classical framework that describes soft gluon fields generated by static, high‑momentum (large‑x) charges using a Gaussian ensemble.
• It employs classical Yang–Mills equations and Wilson-line techniques to compute key observables like the dipole amplitude and the saturation scale in high-energy QCD.
• The model underpins various applications including Glasma initial conditions, the study of transverse-momentum dependent distributions, and non‑Gaussian extensions for improved color neutrality.

Searching arXiv for fresh relevant papers on the McLerran–Venugopalan model and closely related applications. I’m checking relevant arXiv records for the McLerran–Venugopalan model, including small-$x$ TMDs, Glasma initial conditions, finite-thickness extensions, and non-Gaussian generalizations. The McLerran–Venugopalan model is a semi-classical description of the soft gluon fields generated by a large, highly boosted hadron or nucleus at small Bjorken-$x$. In this framework, the large-$x$ valence partons are treated as static, random color sources, while the small-$x$ gluons are classical Yang–Mills fields sourced by those charges. Its defining approximation is a Gaussian ensemble for the color charge density, which makes the model a tractable baseline for saturation physics, Wilson-line observables, dipole amplitudes, gluon transverse-momentum-dependent distributions, and Glasma initial conditions (Dumitru et al., 2014).

1. Physical content and basic assumptions

In the Color Glass Condensate formulation, the target appears in the infinite-momentum frame as a thin sheet of color charge localized in light-cone time, with transverse structure governed by the fluctuating color charge density $\rho^a$. The model assumes that the large-$x$ partons are static and recoilless on the time scale relevant for the soft fields, so that the small-$x$ sector can be computed from classical Yang–Mills equations with random sources. In its standard form, the source ensemble is Gaussian and local in the longitudinal coordinate and in the transverse plane (Dumitru et al., 2014).

A standard representation of the effective action is

$$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$

with corresponding probability functional

$$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$

The associated two-point correlator is

$$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$

Here $x$0 encodes the color charge squared per unit transverse area; in a large nucleus it is proportional to the thickness of the target along the beam direction (Dumitru et al., 2014).

The saturation scale $x$1 is the characteristic momentum scale at which the small-$x$2 fields become strong. In the MV model one finds, up to a Coulomb logarithm,

$x$3

and in lattice implementations $x$4 is often defined operationally from the dipole $x$5-matrix through

$x$6

This makes $x$7 both a physical measure of nonlinear gluon dynamics and a convenient phenomenological parameter (Dumitru et al., 2014).

2. Classical fields, Wilson lines, and dipole scattering

For a given source configuration, the classical field is obtained from a transverse Poisson equation. In covariant gauge,

$x$8

so that

$x$9

The nonvanishing field-strength components are the Weizsäcker–Williams fields

$x$0

High-energy scattering is then encoded by an eikonal Wilson line through the target field,

$x$1

or equivalently by its adjoint analogue for gluonic probes (Dumitru et al., 2014).

The central gauge-invariant observable is the dipole $x$2-matrix,

$x$3

Its real part is $x$4-even and its imaginary part is $x$5-odd,

$x$6

with the symmetry properties $x$7 and $x$8. This decomposition underlies the interpretation of the MV model in terms of pomeron-like and odderon-like exchanges (Dumitru et al., 2014).

Color neutrality is not automatic in every implementation. In practice it is enforced either by removing the zero mode of $x$9 or by introducing a small mass in the Coulomb kernel, so that $x$0 and $x$1 are well defined. This point becomes numerically important when the Poisson equation is solved on a finite torus, because long-range Coulomb tails generate substantial finite-volume artifacts unless an infrared regularization or a correlator-based construction is used (Korcyl, 2021).

3. Operator correlators, gluon distributions, and azimuthal structure

A major strength of the model is that it expresses small-$x$2 observables directly in terms of Wilson-line or field-strength correlators. For gluon transverse-momentum-dependent distributions, the gauge-invariant correlator is written with process-dependent gauge links,

$x$3

and the unpolarized sector is parametrized by the gluon TMDs $x$4 and $x$5 (Pisano et al., 2016).

The gauge-link structure determines which small-$x$6 gluon distribution is being probed. Two future-pointing links map to the Weizsäcker–Williams distributions $x$7, while a future and a past pointing link map to the dipole distributions $x$8. This process dependence is intrinsic to gluon TMD factorization at small $x$9, not a model artifact (Pisano et al., 2016).

Within the MV model, the Weizsäcker–Williams ratio $\rho^a$0 is given by a Fourier–Bessel expression,

$\rho^a$1

with $\rho^a$2 and $\rho^a$3 Bessel functions. In the perturbative tail $\rho^a$4, this ratio approaches the positivity bound; in the saturation region $\rho^a$5, multiple scattering suppresses $\rho^a$6 for the Weizsäcker–Williams case (Pisano et al., 2016).

The same Wilson-line machinery also generates azimuthal anisotropies. For a fundamental dipole at fixed impact parameter, one may write

$\rho^a$7

or, configuration by configuration,

$\rho^a$8

$\rho^a$9

In lattice MV calculations the elliptic amplitude $x$0 is the largest harmonic, reaching $x$1–$x$2 around $x$3, while odd harmonics arise from $x$4-odd fluctuations and impact-parameter dependence (Dumitru et al., 2014).

A closely related construction appears for the dipole gluon GTMD. In the correlation limit, its elliptic component $x$5 is proportional to $x$6, and in the MV model it is driven by the second derivative of the impact-parameter-dependent saturation scale,

$x$7

This makes the quadrupole anisotropy of the gluon Wigner/GTMD sector directly sensitive to gradients of nuclear geometry (Zhou, 2016).

4. Glasma initial conditions and early-time dynamics

In heavy-ion collisions, the MV model supplies the pre-collision fields for the Glasma. In Fock–Schwinger gauge, the post-collision fields are solved in a small-$x$8 expansion, with boost-invariant boundary conditions

$x$9

and longitudinal chromo-electric and chromo-magnetic fields

$x$0

These are the standard Glasma flux-tube fields immediately after the collision (Chen et al., 2012).

The Yang–Mills energy–momentum tensor is

$x$1

At $x$2, the energy density is

$x$3

while the pressures are strongly anisotropic: the transverse pressures are $x$4 at leading order, and the longitudinal pressure is initially negative. In the all-orders leading-$x$5 resummation, one finds

$x$6

with $x$7 and $x$8 decreasing monotonically and $x$9 increasing toward zero while remaining negative in the classical framework (Li et al., 2016).

A generalized MV model with $$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$0 slowly varying across the transverse plane makes early-time flow computable analytically. The averaged Poynting vector contains an even-in-rapidity term

$$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$1

which is aligned with $$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$2, and an odd-in-rapidity term

$$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$3

which produces directed flow. This extends the original boost-invariant Glasma picture to initial conditions with radial, elliptic, and directed flow already at the classical level (Chen et al., 2012).

Finite longitudinal thickness breaks boost invariance explicitly. Replacing the $$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$4 shock-wave source by a Gaussian longitudinal profile,

$$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$5

and evolving the full $$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$6D Yang–Mills equations in the laboratory frame yields non-flat, approximately Gaussian rapidity profiles for the Glasma energy density. Their width depends both on the Lorentz-contracted thickness and on the infrared regulator used in the initial conditions (Ipp et al., 2017).

5. Generalizations of the baseline model

Several extensions modify the source ensemble while retaining the basic CGC logic. One class of modifications imposes color neutrality beyond the white-noise approximation. In the Lam–Mahlon three-dimensional, color-neutral variant, the correlator is

$$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$7

so that

$$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$8

This implements global color neutrality on a finite scale $$S_{\rm eff}[\rho] = \int dx^-\, d^2 x_T \; \frac{\rho^a(x^-,x_T)\,\rho^a(x^-,x_T)}{2\,\mu^2},$$9, removes infrared divergences, and makes the resulting unintegrated gluon distribution explicitly $$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$0-dependent through the finite longitudinal extent of the nucleus (Ozonder et al., 2013).

A related modification introduces colored noise or an effective Coulomb kernel directly in the source action. In momentum space the screened correlator becomes

$$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$1

which vanishes at $$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$2 and hence enforces global color neutrality. In coordinate space the screening cloud is proportional to $$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$3, and the screening scale is of order $$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$4 (McLerran et al., 2016).

Another class of extensions introduces non-Gaussian operators in the effective action. The first even-$$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$5-parity correction is quartic in $$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$6,

$$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$7

while the first nontrivial odd-$$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$8-parity term is cubic and generates a classical odderon in SU(3). For dipole scattering, the quartic term suppresses the classical bremsstrahlung tail at dipole sizes a few times smaller than the inverse saturation scale, and the correction decreases with nuclear size as $$W[\rho]\propto \exp\!\left[ -\int d^2x_T \; \frac{\rho^a(x_T)\,\rho^a(x_T)}{2\,\mu^2} \right].$$9 (Dumitru et al., 2011). A fully non-perturbative analysis shows that the quadratic coupling $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$0 and quartic coupling $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$1 must be renormalized to preserve the two-point function, and that the continuum behavior depends sensitively on the renormalization prescription (2002.03547).

The model has also been extended to helicity-dependent observables by supplementing the eikonal field $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$2 with a sub-eikonal field strength $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$3 and quark fields $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$4. The resulting Gaussian functional $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$5 reduces to the standard MV model after integrating out the sub-eikonal sector and provides the initial condition for the helicity generalization of the JIMWLK equation (Cougoulic et al., 2020).

Two technical issues are repeatedly emphasized in the literature. First, the “infinitesimally thin” nucleus is not equivalent to a single longitudinal sheet without path ordering: the longitudinal randomness survives the $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$6 limit, and neglecting path ordering can underestimate gauge-field correlators and initial Glasma energy densities by large factors (0711.2364). Second, numerical solutions of the Poisson equation on a torus develop large finite-volume effects because the MV kernel is long-ranged; a correlator-driven Wilson-line construction can remove these artifacts while preserving the intended continuum two-point function (Korcyl, 2021).

6. Phenomenology, process dependence, and theoretical boundaries

The MV model is widely used as a quasi-classical baseline for observables at an Electron–Ion Collider, RHIC, and the LHC. In electron–proton scattering it predicts sizable $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$7 azimuthal asymmetries in heavy-quark pair and dijet production driven by the Weizsäcker–Williams linearly polarized gluon distribution $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$8. It also supplies positivity-based upper bounds for single-spin asymmetries tied to the gluon Sivers function and related $$\langle \rho^a(x_T)\,\rho^b(y_T)\rangle = \delta^{ab}\,\mu^2\,\delta^{(2)}(x_T-y_T).$$9-odd distributions, and it implies sign-change relations between $x$00 channels with two future-pointing gauge links and $x$01 color-singlet channels with two past-pointing gauge links (Pisano et al., 2016).

The same formalism underlies a polarized odderon mechanism for transverse single-spin asymmetries at small $x$02. In a transversely polarized proton, an axially asymmetric valence-quark distribution in impact-parameter space generates a spin-dependent odderon, with the expectation value

$x$03

This mechanism contributes to backward jet production in $x$04 collisions and to open-charm production in semi-inclusive DIS (Zhou, 2013).

In ultra-peripheral heavy-ion collisions, the dense target nucleus can be approximated as a highly saturated MV ensemble with a regulated adjoint dipole correlator

$x$05

This simplification, together with the factorized dipole approximation, enables efficient calculations of two-gluon production and associated azimuthal harmonics. In that setting the MV model can also be used for the hadronic component of the quasi-real photon by assigning it a Gaussian transverse color-charge profile (Duan et al., 2022).

The principal limitations of the baseline model are explicit in these applications. It is a non-evolving quasi-classical input: without BK or JIMWLK evolution, the $x$06-dependence of $x$07 is not generated dynamically. This is particularly important for observables sensitive to the transition between the saturation region and the perturbative tail, and for anisotropies whose evolution can be slow or even enhanced under JIMWLK (Dumitru et al., 2014). The model is also sensitive to infrared regularization, lattice volume, and the precise implementation of color neutrality, especially in three-dimensional or finite-thickness variants (Ozonder, 2012).

A further boundary concerns which distributions the model can estimate reliably. For $x$08-even Weizsäcker–Williams gluon TMDs, the MV model gives concrete transverse-momentum dependence and sizable EIC asymmetries. For $x$09-odd Weizsäcker–Williams gluon TMDs, however, perturbative arguments indicate suppression by an additional factor of $x$10 relative to their dipole counterparts, and the quasi-classical saturation framework does not provide a reliable asymptotic small-$x$11 estimate (Pisano et al., 2016).

In its standard and extended forms, the McLerran–Venugopalan model therefore occupies a specific role: it is the canonical quasi-classical initial condition for dense small-$x$12 QCD, formulated in terms of Gaussian or weakly non-Gaussian color sources, classical fields, and Wilson lines. Its durability derives from the fact that this structure can be specialized to dipoles, TMDs, GTMDs, polarized operators, Glasma initial conditions, finite-thickness collisions, and numerical JIMWLK initial data without abandoning the original separation between static large-$x$13 sources and classical small-$x$14 gauge fields.