---
title: Continuous Coulomb–Hodge Gauge
url: https://www.emergentmind.com/topics/continuous-coulomb-hodge-gauge
type: topic
---

# Continuous Coulomb–Hodge Gauge

Searching arXiv for the cited papers to ground the response in current records.
arxiv_search(query="2110.11682 OR 2303.00664 OR 2112.14273 OR 2606.31409 OR 2006.11594 OR 2103.01683", max_results=10, sort_by="relevance")
Continuous Coulomb–Hodge gauge denotes a family of gauge-fixing constructions organized around a Coulomb or co-closed condition together with a Hodge-type decomposition, but the phrase is used in several technically distinct settings. In Hamiltonian electromagnetism it refers to the co-closed representative selected by Helmholtz/Hodge projections built from the inverse Laplacian, thereby isolating radiative transverse modes from the instantaneous Coulomb sector [2110.11682]. In non-associative gauge theory it is the divergence-free torsion condition $(d^\omega)^* T^{(s,\omega)} = 0$ together with a local Coulomb–Hodge slice obtained by an Implicit Function Theorem [2303.00664]. In the Einstein–Yang–Mills literature it appears as part of the constant mean curvature, spatial harmonic, generalized Coulomb gauge, where elliptic gauge equations are coupled to hyperbolic evolution and the resulting solutions are continuous in time [2112.14273]. In Berry-geometry transport it denotes a globally smooth Hodge potential $\mathcal{A}$ on the Brillouin torus satisfying $\delta \mathcal{A} = 0$ and vanishing cycle holonomies, thereby replacing singular Berry connections in topological bands [2606.31409]. In perturbative Yang–Mills theory, closely related “continuous” Coulomb gauges arise from $\theta$-interpolating gauges that define strict Coulomb gauge as a controlled limit of BRST-renormalizable families [2006.11594, 2103.01683].

## 1. Defining features across contexts

A common structural element is the replacement of a gauge-dependent field by a decomposition into exact and co-closed sectors, followed by a gauge choice that selects the co-closed representative. In Euclidean $\mathbb{R}^3$, this is the familiar Coulomb condition $\nabla \cdot A = 0$, while on a Riemannian manifold it is the Hodge-theoretic condition $\delta A = 0$ [2110.11682]. In non-associative gauge theory the same role is played not by a connection 1-form itself, but by the torsion $T^{(s,\omega)}$, with Coulomb–Hodge gauge defined as $(d^\omega)^* T^{(s,\omega)} = 0$ [2303.00664]. In Berry geometry on the Brillouin torus $T^d$, the gauge condition is $\delta \mathcal{A} = 0$ together with vanishing holonomies $\oint_{\gamma_i} \mathcal{A} = 0$, which fixes the harmonic 1-form freedom [2606.31409].

The qualifier “continuous” is also context-dependent. In Hamiltonian electromagnetism, it refers to the inverse Laplacian $\nabla^{-2}$ as a continuous Green operator used to define the projection operators
$$
P_T = I - \nabla \nabla^{-2} \nabla \cdot,\qquad
P_L = \nabla \nabla^{-2} \nabla \cdot
$$
on function space [2110.11682]. In non-associative gauge theory, it refers to continuous and indeed smooth dependence of the Coulomb–Hodge slice on the torsion input in Sobolev topology [2303.00664]. In the Einstein–Yang–Mills setting, it refers to solutions lying in spaces such as
$$
(g,K,A,E) \in C([0,T]; H^s \times H^{s-1} \times H^s \times H^{s-1})
$$
with elliptically determined gauge variables also continuous in time [2112.14273]. In perturbative Yang–Mills theory, continuity is realized through a parameter $\theta$ interpolating between covariant gauges and Coulomb gauge [2006.11594, 2103.01683]. In semiclassical transport, it denotes a globally smooth proxy potential on the torus, free of Dirac strings and robust under discrete $\mathbf{k}$-grid noise [2606.31409].

This suggests that “Continuous Coulomb–Hodge gauge” is best understood not as a single universal formalism but as a recurring pattern: a co-closed gauge condition, an associated Hodge decomposition, and a mechanism that makes the gauge selection global, stable, or continuously controllable.

## 2. Hamiltonian electromagnetism and the Coulomb–Hodge split

In Hamiltonian electromagnetism, the construction begins from the $3+1$ Lagrangian
$$
L = \int d^3x \left[ \frac12 (E^2 - B^2) - \rho A_0 + J \cdot A \right],
$$
with
$$
E = -\partial_t A - \nabla A_0,\qquad B = \nabla \times A,
$$
canonical momenta
$$
\pi^i = -E^i,\qquad \pi^0 = 0,
$$
and Hamiltonian density
$$
H = \frac12 (E^2 + B^2) + A_0 (\nabla \cdot E - \rho) - J \cdot A.
$$
Preservation of the primary constraint $\pi^0 \approx 0$ yields Gauss’s law
$$
G(x) \equiv \nabla \cdot E(x) - \rho(x) \approx 0,
$$
and the Castellani generator
$$
G[\lambda] = \int d^3x \left[ - \dot{\lambda}\,\pi^0 + \lambda (\nabla \cdot E - \rho) \right]
$$
produces
$$
\delta_\lambda A_i = \partial_i \lambda,\qquad
\delta_\lambda E^i = 0,\qquad
\delta_\lambda A_0 = -\partial_t \lambda
$$
[2110.11682].

The Helmholtz decomposition then separates both $E$ and $A$ into transverse and longitudinal sectors:
$$
E = E_T + E_L,\qquad E_L = -\nabla \Phi_C,\qquad \nabla^2 \Phi_C = -\rho,
$$
and
$$
A = A_T + \nabla \chi,\qquad \nabla \cdot A_T = 0.
$$
Choosing $\lambda = -\chi$ reaches Coulomb gauge, so that $\nabla \cdot A = 0$ and $A = A_T$ [2110.11682]. The continuous character is explicit in the Green operator
$$
(\nabla^{-2} f)(x) = \int d^3y\, G(x,y) f(y),\qquad
G(x,y) = -\frac{1}{4\pi |x-y|},
$$
which yields the orthogonal projectors
$$
P_T = I - \nabla \nabla^{-2} \nabla \cdot,\qquad
P_L = \nabla \nabla^{-2} \nabla \cdot,
$$
with component formula
$$
P_T^{ij} = \delta^{ij} - \partial^i \partial^j \nabla^{-2}.
$$

After using Gauss’s law and integrating out the Coulomb potential, the Hamiltonian splits into a radiative and an instantaneous term:
$$
H = \frac12 \int d^3x \left[ E_T^2 + B(A_T)^2 \right]
+ \frac12 \int d^3x\, d^3x' \frac{\rho(x)\rho(x')}{4\pi |x-x'|}
- \int d^3x\, J \cdot A_T.
$$
Only the transverse sector propagates:
$$
(\partial_t^2 - \nabla^2) A_T = P_T J,\qquad
E_T = -\partial_t A_T,\qquad
E_L = -\nabla \Phi_C.
$$
The physical content is correspondingly sharp: $A_T$ and its conjugate $E_T$ carry the two transverse radiative degrees of freedom, whereas $E_L$ and $\Phi_C$ encode the instantaneous Coulomb interaction fixed by $\rho$ [2110.11682].

The reduced phase-space structure is encoded by the second-class pair
$$
\chi_1(x) = \nabla \cdot A(x) \approx 0,\qquad
\chi_2(x) = \nabla \cdot E(x) - \rho(x) \approx 0,
$$
which leads to the transverse Dirac bracket
$$
\{ A_T^i(x), E_T^j(y) \}_D
=
\left[ \delta^{ij} - \partial^i \partial^j \nabla^{-2} \right]\delta^3(x-y).
$$
In Fourier space each $\mathbf{k}$-mode has two physical transverse polarizations [2110.11682].

## 3. Symplectic and Hodge-theoretic formulation

The Hamiltonian derivation also admits a symplectic interpretation. The symplectic form on phase space is
$$
\Omega = \int d^3x\, \delta E^i \wedge \delta A_i.
$$
Representing the Coulombic coordinate by $E_c = \nabla \phi$ with $\nabla^2 \phi = -\rho$, one defines $A_r$ to be symplectically orthogonal to $E_c$:
$$
0 = \Omega(A_r, E_c) = \int d^3x\, A_r \cdot \nabla \phi
= -\int d^3x\, \phi\, \nabla \cdot A_r,
$$
for all $\phi$, assuming decay so that boundary terms vanish. Hence $\nabla \cdot A_r = 0$, so $A_r$ is precisely the Coulomb-gauge representative, equivalently
$$
A_r = P_T A = A - \nabla \nabla^{-2} \nabla \cdot A.
$$
Conversely, if $A_c = \nabla \lambda$ is pure gauge, then
$$
0 = \Omega(A_c, E_r)
= \int d^3x\, E_r \cdot \nabla \lambda
= -\int d^3x\, \lambda\, \nabla \cdot E_r
$$
for all $\lambda$, so $\nabla \cdot E_r = 0$, i.e. $E_r = P_T E$ and $E_c = P_L E = -\nabla \Phi_C$ [2110.11682].

The resulting symplectic splitting is
$$
\Omega = \int d^3x \left[ \delta E_T^i \wedge \delta A_T^i + \delta E_L^i \wedge \delta A_L^i \right],
$$
with the longitudinal pair nondynamical and the transverse pair dynamical. This makes the emergence of Coulomb gauge an orthogonality statement rather than merely a convenient coordinate condition [2110.11682].

The same structure extends to differential forms on a Riemannian 3-manifold $(M,g)$. If $A$ and $E$ are treated as 1-forms, with Hodge star $*$ and codifferential $\delta = d^* = (-1)^{np+n+1} * d *$, then the Hodge Laplacian is
$$
\Delta = d\delta + \delta d,
$$
and under suitable conditions one has
$$
\Omega^1(M) = d\Omega^0(M) \oplus \delta \Omega^2(M) \oplus \mathcal{H}^1(M).
$$
Coulomb gauge becomes the co-closed condition
$$
\delta A = 0,
$$
which on Euclidean $\mathbb{R}^3$ reduces to $\nabla \cdot A = 0$ [2110.11682]. On $\mathbb{R}^3$ with suitable decay there are no nontrivial harmonic 1-forms, but on manifolds with nontrivial topology or boundaries, harmonic modes and boundary conditions affect uniqueness and the explicit form of the decomposition.

The contrast with Lorenz gauge is also explicit in this framework. The Lorenz condition $\partial_\mu A^\mu = 0$ preserves manifest Lorentz covariance and yields wave equations for all components of $A_\mu$, whereas Coulomb–Hodge gauge solves the longitudinal sector instantaneously through Gauss’s law and leaves the transverse sector as the true propagating subsystem [2110.11682].

## 4. Non-associative gauge theory and the Coulomb–Hodge slice

In non-associative gauge theory, the basic data are a finite-dimensional smooth loop $L$, its tangent algebra $\mathfrak{l}$, the pseudoautomorphism group $\Psi$, a principal $\Psi$-bundle $P \to M$, and associated bundles $Q$, $Q'$, and $A$ [2303.00664]. A configuration is a pair $(s,\omega)$, where $s \in \Gamma(Q)$ is a section and $\omega$ is a connection on $P$. The torsion of $(s,\omega)$ is the $A$-valued 1-form
$$
T^{(s,\omega)} := (s^*\theta)\circ \mathrm{proj}_H,
$$
equivalently
$$
T^{(s,\omega)}|_p = (R_{(s_p)^{-1}})_* (d^\omega s)|_p,
$$
and the projected curvature is
$$
\hat{F}^{(s,\omega)} := \phi_s(F^{(\omega)}).
$$
These satisfy the structural equation
$$
\hat{F}^{(s,\omega)}
=
d^\omega T^{(s,\omega)} - \frac12 [T^{(s,\omega)},T^{(s,\omega)}]^{(s)}.
$$

Fixing a Riemannian metric $g$ on $M$ and a $\Psi$-invariant, nondegenerate inner product on $\mathfrak{l}$ determines the adjoint $(d^\omega)^*$. The Coulomb–Hodge gauge condition is
$$
(d^\omega)^* T^{(s,\omega)} = 0,
$$
equivalently $\delta T^{(s,\omega)} = 0$ with $\delta = (d^\omega)^*$ [2303.00664]. Under loop gauge transformations with $\omega$ fixed, sections $A \in \Gamma(Q')$ act by $s \mapsto As$, and the torsion transforms affinely:
$$
T^{(As,\omega)} = (\rho_A^{(s)})^{-1} d^\omega A + \mathrm{Ad}_A^{(s)} T^{(s,\omega)}.
$$
This generalizes the classical Lie-theoretic formula $A \mapsto g^{-1}dg + g^{-1}Ag$ [2303.00664].

The analytical mechanism is local and elliptic. Writing a 1-parameter loop gauge transformation as $A_t = \exp_s(t\xi)$ with $\xi \in \Gamma(A)$, the torsion satisfies
$$
\frac{d}{dt} T^{(A_t s,\omega)}
=
[\xi, T^{(A_t s,\omega)}]^{(A_t s)} + d^\omega \xi.
$$
Using the corresponding evolution operator $U_{t\xi}^{(s)}$, one obtains
$$
T^{(\exp_s \xi \cdot s,\omega)}
=
U_\xi^{(s)} T^{(s,\omega)}
+
U_\xi^{(s)}
\left[
\int_0^1 (U_{\tau \xi}^{(s)})^{-1} d\tau
\right]
d^\omega \xi,
$$
and packages the gauge equation as a smooth Banach-space map
$$
G(a,\xi)
:=
(d^\omega)^*
\left[
U_\xi^{(s)} a
+
U_\xi^{(s)}
\left(
\int_0^1 (U_{\tau\xi}^{(s)})^{-1} d\tau
\right)
d^\omega \xi
\right].
$$
At $(a,\xi)=(0,0)$, the linearization in the second variable is
$$
D_2 G|_{(0,0)}(\eta) = (d^\omega)^* d^\omega \eta =: \Delta^{(\omega)} \eta,
$$
the bundle Laplacian on $0$-forms [2303.00664].

On a closed manifold, $\Delta^{(\omega)}$ is elliptic and Fredholm of index $0$. Restricting to the $L^2$-orthogonal complement $K^\perp$ of $\ker \Delta^{(\omega)}$ makes it an isomorphism, which permits an Implicit Function Theorem argument. The main existence theorem states that for $k \in \mathbb{N}$ and $r \ge 1$ with $kr > n$, there exist $\delta \in (0,1]$ and $K \in (0,\infty)$ such that if
$$
\|T^{(s,\omega)}\|_{W^{k-1,r}} < \delta,
$$
then there exists $A \in W^{k,r}(Q')$ with
$$
(d^\omega)^* T^{(As,\omega)} = 0
$$
and
$$
\|T^{(As,\omega)}\|_{W^{k-1,r}}
\le
K \|T^{(s,\omega)}\|_{W^{k-1,r}}
\left(1 + \|T^{(s,\omega)}\|_{W^{k-1,r}}^{\,k-1}\right).
$$
If additionally $(k-1)r \ge n$, then the solution $A$ is smooth [2303.00664].

The continuity statement is correspondingly precise: the Implicit Function Theorem yields a $C^k$ map $\xi : B_\delta \to W^{k,r}(A)$ and a Lipschitz estimate
$$
\|\xi(a_1)-\xi(a_2)\|_{W^{k,r}}
\le
2\beta N \|a_1-a_2\|_{W^{k-1,r}},
$$
for suitable constants $\beta,N$ [2303.00664]. Uniqueness holds within the local slice $K^\perp$, i.e. modulo the finite-dimensional kernel of the Laplacian.

A distinguished application is $G_2$ geometry, where the loop of unit octonions $UO$ yields $\Psi = \mathrm{Spin}(7)$, $\Psi' = \mathrm{SO}(7)$, $H = G_2$, and $\mathfrak{l} \cong \mathrm{Im}\,\mathbb{O} \cong \mathbb{R}^7$ [2303.00664]. In this specialization the torsion satisfies
$$
\nabla_a \varphi_{bcd} = -2 T_a{}^e \psi_{ebcd},\qquad
\nabla_a \psi_{bcde} = 8 T_{a[b}\varphi_{cde]},
$$
and the Coulomb–Hodge condition becomes the divergence-free torsion equation $\mathrm{div}\, T^{(V)} = 0$ for an appropriate unit octonion section $V$.

## 5. Elliptic–hyperbolic gauge fixing in Einstein–Yang–Mills

For the Einstein–Yang–Mills system, continuous Coulomb–Hodge structure appears in the constant mean extrinsic curvature spatial harmonic generalized Coulomb gauge, abbreviated CMCSHGC [2112.14273]. The unknowns are a spacetime metric $\hat{g}$ on $M \simeq \mathbb{R} \times \Sigma$, a Yang–Mills connection $\hat{A}_\mu$, curvature $\hat{F}_{\mu\nu}$, and the induced ADM variables on spacelike slices $\Sigma_t$: metric $g_{ij}$, extrinsic curvature $K_{ij}$, lapse $N$, shift $X^i$, spatial potential $A_i$, temporal component $A_0$, electric field $E_i = \hat{F}_{0i}$, and magnetic part $F_{ij}$ [2112.14273].

The gauge has three components. The constant mean curvature condition requires $\tau(t,x)=\tau(t)$ to be spatially constant. The spatial harmonic condition relative to a background metric $\gamma$ is
$$
V^i := g^{jk}\big(\Gamma^i_{jk}(g)-\Gamma^i_{jk}(\gamma)\big)=0.
$$
The generalized Coulomb condition relative to $(\gamma,\hat{A})$ is
$$
g^{ij} D_j^{\hat{A}[\gamma]}(A_i-\hat{A}_i)=0.
$$
The Yang–Mills Gauss constraint is
$$
g^{ij}(\nabla_i E_j + [A_i,E_j])=0.
$$
To avoid derivative loss, the temporal gauge variable is replaced by
$$
\phi := A_0 - A_j X^j,
$$
which satisfies an elliptic equation obtained by differentiating the generalized Coulomb condition in time [2112.14273].

Under this gauge, the full system becomes elliptic–hyperbolic. The lapse solves
$$
- g^{ij}\nabla_i \nabla_j N
+
\left(
|K|^2
+
\frac{n-2}{n-1} g^{ij} E_i \cdot E_j
+
\frac{1}{2(n-1)} g^{ij} g^{kl} F_{ik}\cdot F_{jl}
\right) N
= 1.
$$
The shift satisfies
$$
\Delta_g X^i - R^i{}_j X^j
=
-2 K^{ij}\nabla_j N
- 2N g^{ik} g^{lj} F_{kl}\cdot E_j
+ \tau \nabla^i N
+ (2N K^{jk} - 2\nabla^j X^k)\big(\Gamma^i{}_{jk}(g)-\Gamma^i{}_{jk}(\gamma)\big),
$$
and the shifted Yang–Mills potential $\phi$ obeys a second elliptic equation whose leading term is
$$
g^{ij} D_i^{A[\gamma]} D_j^{A[\gamma]} \phi
+
g^{ij}[\hat{A}_i-A_i, D_j^A \phi]
+
\text{lower order terms}
=0
$$
[2112.14273].

The hyperbolic subsystem evolves $(g,K,A,E)$. In particular,
$$
\partial_t g_{ij} = -2N K_{ij} + \mathcal{L}_X g_{ij},
$$
and
$$
\partial_t A_i
=
N E_i + \partial_i \phi + [A_i,\phi]
+ X^k \nabla_k^{[\gamma]} A_i
+ A_j \nabla_i^{[\gamma]} X^j.
$$
For the electric field,
$$
\partial_t E_i
=
N (\mathrm{tr}_g K) E_i
- (\nabla_j N) F_i{}^j
- N(\nabla_j F_i{}^j - \nabla_i \mathcal{C})
+ N[F_i{}^j,A_j]
+ [E_i,\phi]
+ X^k \nabla_k E_i
+ E_k \nabla_i X^k
- 2N K_i{}^j E_j,
$$
where
$$
\mathcal{C} := g^{ij} D_j^{\hat{A}[\gamma]}(A_i-\hat{A}_i).
$$
The identity
$$
\nabla_j F_i{}^j = \Delta_g^\gamma A_i + \nabla_i \mathcal{C} + \mathcal{SL}_i
$$
isolates the elliptic image of $A_i$ and removes the derivative-loss obstruction [2112.14273].

The local well-posedness theorem states that for initial data
$$
(g_0,K_0,A_0,E_0) \in H^s \times H^{s-1} \times H^s \times H^{s-1},
\qquad s > \frac{n}{2} + 1,
$$
on a closed $\Sigma$, satisfying the constraints and gauge conditions, there exists $T>0$ and a unique solution
$$
(g,K,A,E) \in C([0,T]; H^s \times H^{s-1} \times H^s \times H^{s-1}),
$$
with
$$
(N,X,\phi) \in C([0,T]; H^{s+1} \times H^{s+1} \times H^{s+1}),
$$
and the solution map is continuous in these topologies [2112.14273]. The gauge and constraint quantities propagate because the paper derives a closed wave/transport system for the gauge-error variables $(T,S,H,M,G,L)$; if they vanish initially, they remain zero for the lifespan of the solution.

The continuation criterion is likewise explicit: if $T^*$ is the maximal time of existence, then either $T^*=\infty$, or
$$
\lim_{t\to T^{*-}}
\sup \max\{ C(\hat{\Lambda}(t)), \|D\hat{g}(t)\|_{L^\infty}, \|DA(t)\|_{L^\infty}, \|E(t)\|_{L^\infty}\}
= \infty
$$
[2112.14273]. Here the Coulomb–Hodge ingredient is the generalized divergence-free Yang–Mills condition, while the “Hodge” coordinate condition applies to the spatial metric itself.

## 6. Topological transport and interpolating quantum gauge theories

In Berry-geometry transport on the Brillouin torus, the Hodge–de Rham decomposition provides a coordinate-free regularization of topological bands. For an isolated 2D Bloch band, the Berry curvature 2-form is
$$
\mathcal{F}_n = \Omega_n(k)\, dk_x \wedge dk_y,
$$
with Chern number
$$
C_n = \frac{1}{2\pi} \int_{BZ} \mathcal{F}_n.
$$
On a closed oriented 2-manifold, any 2-form decomposes as
$$
\mathcal{F} = d\mathcal{A} + \mathcal{H},
$$
where $\mathcal{A}\in \Omega^1(T^2)$ is a globally defined Hodge potential and
$$
\mathcal{H} = \frac{2\pi C}{\mathrm{Vol}(BZ)} \,\mathrm{vol}_{BZ}
$$
is the harmonic 2-form fixed by the Chern number [2606.31409]. Because the singular topological content is isolated in $\mathcal{H}$, the potential $\mathcal{A}$ can be globally smooth and single-valued even when the quantum Berry connection is obstructed by Dirac strings.

The continuous Coulomb–Hodge gauge is then
$$
\delta \mathcal{A}=0,\qquad
\oint_{\gamma_i} \mathcal{A}=0,\quad i=x,y.
$$
Applying $\delta$ to $\mathcal{F}=d\mathcal{A}+\mathcal{H}$ yields the elliptic PDE
$$
\Delta \mathcal{A} = \delta \mathcal{F},
$$
since $\delta\mathcal{H}=0$ and $\delta \mathcal{A}=0$ [2606.31409]. Existence and uniqueness follow from Hodge theory on the compact boundaryless torus, with the vanishing cycle integrals removing the harmonic 1-form freedom.

The transport formulas become geometrically regularized. At $T=0$, the anomalous Hall conductivity for a single band satisfies
$$
\sigma_{xy}
=
-\frac{e^2}{\hbar}
\int_{BZ}
\frac{d^2k}{(2\pi)^2}
f_n(k)\Omega_n(k),
$$
and with $\mathcal{F}=d\mathcal{A}+\mathcal{H}$ one obtains
$$
\sigma_{xy}
=
\frac{e^2}{h}
\left[
C\nu + \frac{1}{2\pi}\oint_{FS}\mathcal{A}
\right],
$$
where
$$
\nu := \frac{1}{\mathrm{Vol}(BZ)}\int_{BZ} f\, \mathrm{vol}_{BZ}.
$$
For a filled band, $\nu=1$ and the first term is the TKNN quantized Hall response, while the second is a Fermi-surface line integral of the smooth 1-form $\mathcal{A}$ [2606.31409]. The same formalism reproduces the stability of integration by parts for the Berry-curvature dipole, because derivatives of the curvature are replaced by expressions involving $d(\partial_a f_0)\wedge \mathcal{A}$.

The relation to the Maximally Localized Wannier Function gauge is explicit in trivial bands. When $C=0$, $\mathcal{H}=0$ and $d\mathcal{A}=\mathcal{F}$; then $\delta\mathcal{A}=0$ coincides with the momentum-space Coulomb gauge $\nabla_k \cdot A_{\mathrm{Berry}}=0$, and
$$
\oint_{\gamma_i}\mathcal{A}=0
\quad \Longleftrightarrow \quad
W_i = \exp\!\left(i\oint_{\gamma_i} A_{\mathrm{Berry}}\right)=1
$$
matches the MLWF gauge-fixing condition [2606.31409]. For topological bands with $C\neq 0$, MLWFs are obstructed, but the Hodge potential remains globally smooth and computationally usable.

A different realization of continuity appears in perturbative Yang–Mills theory through $\theta$-interpolating gauges. The gauge-fixing functional
$$
F[A] = \partial_i A^i + \theta^2 \partial_0 A^0
$$
defines
$$
L_{GF} = - \frac{1}{2\theta^2}[F[A]]^2
$$
in the Feynman-to-Coulomb flow gauge [2006.11594], and
$$
L_{gf} = -\frac{1}{2\alpha^2}\big[\partial_i A_i^a + \theta^2 \partial_0 A_0^a\big]^2
$$
in the Landau-to-Coulomb interpolating gauge [2103.01683]. As $\theta \to 0$, both reduce to the Coulomb condition $\partial_i A^i=0$, while $\theta=1$ yields the covariant Feynman or Landau gauges, respectively. The common deformed denominator
$$
K_\theta^2 \equiv \mathbf{k}^2 - \theta^2 k_0^2
$$
regularizes the $k_0$ behavior of propagators, so that energy divergences are absent for $\theta \neq 0$ and strict Coulomb gauge is recovered only after renormalization and the limit $\theta \to 0$ [2006.11594, 2103.01683].

These papers establish BRST-controlled renormalizability for all fixed $\theta$, with anisotropic renormalization of spatial and temporal components and multiplicative renormalization of the interpolation parameter:
$$
\theta_R^2 = \left(\frac{Z_5}{Z_6}\right)^{1/2} \theta^2
$$
in the Feynman-to-Coulomb flow gauge [2006.11594], and
$$
\theta_R^2 = \left(\frac{Z_5'}{Z_6'}\right)^{1/2} \theta^2
$$
in the Landau-to-Coulomb flow gauge [2103.01683]. In the latter case the all-orders identity
$$
Z_0' = Z_7'
$$
implies a $\theta$-independent Yang–Mills $\beta$-function [2103.01683]. The differential-form interpretation stated in these works is that Coulomb gauge is precisely the Hodge condition $\delta A=0$ on spatial slices, while the interpolating functional adds a temporal term that vanishes in the Coulomb limit [2006.11594, 2103.01683].

## 7. Scope, assumptions, and recurring limitations

Across these usages, the gauge is never merely an algebraic constraint; it is coupled to geometric or analytic assumptions that determine existence and uniqueness. In Hamiltonian electromagnetism, the spatial manifold is taken to be $\mathbb{R}^3$ with fields decaying sufficiently fast at infinity so that surface terms vanish and $\nabla^{-2}$ exists as the standard Green operator [2110.11682]. On more general manifolds, harmonic 1-forms and boundary conditions affect the decomposition and uniqueness. In non-associative gauge theory, the manifold is closed, the loop $L$ is compact, the pseudoautomorphism group $\Psi$ is finite-dimensional, and the Coulomb–Hodge slice is proved only for sufficiently small torsion in $W^{k-1,r}$ with $kr>n$ [2303.00664]. In Einstein–Yang–Mills, the manifold $\Sigma$ is closed, the gauge group is compact semisimple, and short-time existence relies on elliptic isomorphism properties that require the evolving fields to remain close to chosen backgrounds on successive time slabs [2112.14273]. In Brillouin-zone transport, the Brillouin zone is a compact boundaryless torus, and the Hodge problem is solved after explicitly projecting out harmonic zero modes by vanishing holonomies [2606.31409]. In interpolating Yang–Mills gauges, the perturbative construction controls energy divergences but does not address nonperturbative Gribov copies [2006.11594].

A recurrent misconception is that Coulomb gauge simply means “transverse variables only.” The literature considered here is more precise. In Hamiltonian electromagnetism, the longitudinal sector is not discarded; it is retained as the constrained instantaneous Coulomb sector determined by Gauss’s law [2110.11682]. In CMCSHGC, the generalized Coulomb condition does not eliminate the need for Gauss law or elliptic determination of $\phi$; rather, it is the condition that makes those structures analytically tractable [2112.14273]. In topological transport, the co-closed gauge does not remove the topological flux; it isolates it in the harmonic piece $\mathcal{H}$ [2606.31409]. In non-associative gauge theory, the object being gauge-fixed is torsion rather than an ordinary Lie-algebra-valued connection 1-form [2303.00664].

Another recurring point is that “continuous” should not be read uniformly. Depending on context, it means continuous Green-kernel projections, continuous dependence in Sobolev spaces, continuous-in-time evolution, continuous interpolation in a gauge parameter, or a globally smooth nonsingular representative on a compact manifold. A plausible implication is that the expression names a methodological family rather than a single standardized gauge doctrine. What remains invariant across the cited work is the emphasis on co-closedness, orthogonal decomposition, and a precise separation between constrained and propagating, harmonic and exact, or topological and geometric sectors.

Source: https://www.emergentmind.com/topics/continuous-coulomb-hodge-gauge