Generalized Velocity Gauge

Updated 9 July 2026

Generalized Velocity Gauge is a family of gauge conditions that parameterizes electromagnetic and light–matter coupling, interpolating between Lorenz (v=c) and Coulomb (v→∞) gauges.
It reorganizes the decomposition of scalar and vector potentials, optimizing perturbative expansions and improving convergence in quantum, relativistic, and finite-band formulations.
Its formulation preserves gauge invariance by carefully redistributing interaction terms, ensuring consistent observables across classical electrodynamics and advanced material models.

The generalized velocity gauge is a family of gauge constructions in which the representation of electromagnetic or light–matter coupling is parameterized so that familiar gauges arise as special cases. In classical electrodynamics, its defining form is

$\grad \cdot {\bf A}^{(v)}({\bf r}, t) = - {c \over v^2} {\partial \Phi^{(v)}({\bf r}, t) \over \partial t},$

with an arbitrary parameter $v$ that sets the propagation speed associated with the scalar potential; $v=c$ gives the Lorenz gauge and $v\to\infty$ gives the Coulomb gauge. In more recent quantum, atomic, and condensed-matter literature, the same expression is also used for gauge-consistent reorganizations of perturbation theory, relativistic strong-field Hamiltonians, finite-band velocity-gauge response, and Floquet or Keldysh formalisms (Yang et al., 27 Aug 2025).

1. Terminological scope

The literature is not uniform in its use of the expression “generalized velocity gauge.” In classical electromagnetics it denotes a one-parameter gauge condition for the scalar and vector potentials. In quantum theory it can denote a one-parameter family of perturbative branches, a relativistic propagation gauge, or a basis-aware velocity-gauge construction for truncated Hilbert spaces. The shared theme is not a single universal formula, but a controlled redistribution of interaction terms while preserving gauge-consistent observables.

Context	Defining construction	Special cases or role
Electromagnetic potentials	$\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$	$v=c$ : Lorenz; $v\to\infty$ : Coulomb (Yang et al., 27 Aug 2025)
Quantum transition amplitudes	One-parameter family $V^{(\lambda)}(t)$	$\lambda=0$ : velocity; $\lambda=1$ : length (Faisal, 2008)
Relativistic TDDE	Propagation-gauge Hamiltonian $v$ 0	Explicit $v$ 1 term (Kjellsson et al., 2017)
Finite-band solids	Truncated velocity gauge	Nested-commutator expansion valid in a finite basis (Tiwari et al., 26 Jan 2025)

This multiplicity of meanings has a practical consequence: statements about convergence, causality, current operators, or gauge equivalence are formulation-dependent. A claim that is precise for the electromagnetic $v$ 2-gauge need not be the relevant statement for a finite-band Floquet Hamiltonian or a relativistic Dirac solver.

2. Electromagnetic definition and governing equations

In the electromagnetic formulation, the generalized velocity gauge is defined by

$v$ 3

Using Maxwell’s equations, this leads to

$v$ 4

and

$v$ 5

The parameter $v$ 6 is therefore a gauge parameter that sets the propagation speed of the scalar potential. For finite $v$ 7, $v$ 8 propagates at speed $v$ 9; for $v=c$ 0, one recovers fully relativistically symmetric Lorenz-gauge propagation; for $v=c$ 1, the scalar potential becomes instantaneous, as in the Coulomb gauge. The same analysis notes earlier work showing that for $v=c$ 2, the vector potential contains a component propagating at speed $v=c$ 3 and another always propagating at $v=c$ 4, and that $v=c$ 5 or $v=c$ 6 corresponds to advanced or superluminal formal potentials (Yang et al., 27 Aug 2025).

This construction is explicitly interpolating. It does not change the electromagnetic fields; it changes the decomposition of those fields into scalar and vector potentials. The classical significance of the generalized velocity gauge is therefore organizational: it provides a continuous family of potential representations connecting standard textbook gauges. The same source further states that, together with the generalized Kirchhoff gauge, these constructions encompass most gauges in vector/scalar potential form in electromagnetics (Yang et al., 27 Aug 2025).

The scalar potential in the generalized velocity gauge is written with the retarded scalar Green’s function

$v=c$ 7

$v=c$ 8

which satisfies

$v=c$ 9

For the vector potential, the time-domain dyadic Green’s function is

$v\to\infty$ 0

so that

$v\to\infty$ 1

The same result is shown to be equivalent to the more compact closed form previously obtained in Yang (2005),

$v\to\infty$ 2

where the composite scalar Green’s function is defined by convolution and satisfies a fourth-order wave equation. Structurally, the dyadic Green’s function contains a transverse, propagating part $v\to\infty$ 3 and a longitudinal correction proportional to $v\to\infty$ 4, encoding the difference between $v\to\infty$ 5-propagation and light-speed propagation (Yang et al., 27 Aug 2025).

For monochromatic sources, the frequency-domain dyadic Green’s function is

$v\to\infty$ 6

with $v\to\infty$ 7. In the limit $v\to\infty$ 8, the longitudinal correction vanishes and one obtains the Lorenz-gauge dyadic Green’s function,

$v\to\infty$ 9

In the limit $\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$0, one recovers the Coulomb-gauge result,

$\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$1

The associated generalized Kirchhoff gauge is defined by

$\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$2

and is most naturally handled in the frequency domain. Its scalar potential is an “instantaneous Yukawa potential,” and its dyadic Green’s function reads

$\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$3

Here $\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$4 gives the Coulomb gauge, whereas $\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$5 yields the temporal gauge for $\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$6 (Yang et al., 27 Aug 2025).

4. Perturbative, relativistic, and atomic generalizations

In quantum perturbation theory, the generalized velocity-gauge idea appears as a one-parameter family of equivalent series representations of a single transition amplitude. Faisal starts from the minimal-coupling Schrödinger equation and derives a master perturbation expansion with a real parameter $\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$7. The interaction operator

$\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$8

reduces to the traditional velocity-gauge perturbation for $\grad \cdot {\bf A}^{(v)} = - {c \over v^2} \partial_t \Phi^{(v)}$9 and to the length-gauge perturbation for $v=c$ 0. The two standard series are therefore presented as two members of a one-parameter family of infinitely many branches of the same exact amplitude, equal in a common domain of convergence if such a domain exists (Faisal, 2008).

A different generalization appears in the dynamic Stark shift under hybrid gauge transformations, where the scalar and vector potentials are modified but the wave function is kept fixed. In that setting, gauge invariance of the correction to the Stark shift under a perturbation $v=c$ 1 requires an additional correction to the transition current in the velocity gauge,

$v=c$ 2

The formal equality

$v=c$ 3

holds only when the extra current terms $v=c$ 4 and $v=c$ 5 are included. This correction is specific to the velocity-gauge formulation because the interaction is written in terms of the canonical momentum operator (Jentschura et al., 2018).

In relativistic strong-field physics, the generalized velocity gauge is introduced for the time-dependent Dirac equation as the propagation gauge. Starting from

$v=c$ 6

the gauge transformation with

$v=c$ 7

yields

$v=c$ 8

The purpose is to reorganize the light–matter coupling so that the numerically troublesome pieces of the relativistic minimal-coupling Hamiltonian are replaced by terms that converge much faster in multipole or Taylor expansions of the laser field. In hydrogen exposed to a 15-cycle pulse with $v=c$ 9, $v\to\infty$ 0, and intensity $v\to\infty$ 1, the propagation-gauge long-wavelength approximation matches the fifth-order beyond-dipole minimal-coupling result, while the minimal-coupling expansion required $v\to\infty$ 2, about 70 times more nonzero matrix elements than the propagation-gauge calculation, which converged already at $v\to\infty$ 3 (Kjellsson et al., 2017).

In molecular electronic structure, the generalized velocity-gauge idea is used to compute oscillator strengths in $v\to\infty$ 4SCF theory without modifying the non-orthogonal ground- and excited-state Kohn–Sham determinants. The velocity-gauge oscillator strength is written in terms of the momentum matrix element,

$v\to\infty$ 5

and the practical determinant formula uses the occupied-orbital overlap matrix $v\to\infty$ 6 and its adjugate. The key claim is that the momentum operator is translationally invariant, so the velocity-gauge transition property is origin independent even when $v\to\infty$ 7. The same work reports that the use of spin-purified singlet excitation energy in the velocity-gauge transition dipole moment significantly enhances overall performance for conjugated chromophores (Shen et al., 16 Mar 2026).

5. Crystalline, Floquet, and topological formulations

For crystalline systems, a central issue is that naïve minimal coupling in a finite band model is not gauge consistent. A velocity-gauge analysis of nonlinear optical response in crystals therefore starts from the finite-band length-gauge Hamiltonian, written with the covariant derivative

$v\to\infty$ 8

and then performs a time-dependent unitary transformation to the velocity gauge. The transformed Hamiltonian becomes an infinite series in the vector potential,

$v\to\infty$ 9

with coefficients defined by nested commutators of $V^{(\lambda)}(t)$ 0 with covariant derivatives. The current operator is itself field dependent, so the response requires simultaneous expansion of both the density matrix and the velocity matrix elements. The resulting framework is stated to be valid to any order and for any finite band model, and it was illustrated for monolayer graphene (Passos et al., 2017).

A closely related basis-aware construction is the truncated velocity gauge for first-principles Floquet engineering. Rather than assuming a complete Hilbert space, the driven Hamiltonian is written as

$V^{(\lambda)}(t)$ 1

and evaluated in generalized tight-binding Hamiltonians built from maximally localized Wannier functions. In trans-polyacetylene, the truncated velocity gauge reproduces converged full-basis spectra with far fewer bands: at $V^{(\lambda)}(t)$ 2 V/Å the 6-band truncated result is already converged, whereas the conventional velocity gauge needs about 30 bands for the main spectral region and about 120 bands in the low-frequency region; at $V^{(\lambda)}(t)$ 3 V/Å, even 100 bands are insufficient in the standard velocity gauge and about 120 are required. The implementation is reported in FloqticS (Tiwari et al., 26 Jan 2025).

In topological materials, the velocity gauge is used to avoid singular Berry-connection and dipole structures that destabilize length-gauge semiconductor Bloch equations. The velocity-gauge equation

$V^{(\lambda)}(t)$ 4

decouples each $V^{(\lambda)}(t)$ 5-channel, so one may choose a grid that avoids singular points while preserving the essential response. Applied to the Haldane model, this formulation reproduces key high-harmonic-generation features, including selection rules for linear and circular drivers, the linear cut-off law scaling, and anomalous circular dichroism; a topologically trivial monolayer MoS $V^{(\lambda)}(t)$ 6 benchmark shows good qualitative agreement with the length gauge and TDDFT (Kim et al., 2021).

A current-response version of the velocity gauge has also been developed for nonlinear spectroscopy of topology in the SSH model. Starting from the full minimal-coupling Hamiltonian,

$V^{(\lambda)}(t)$ 7

and retaining both the paramagnetic and diamagnetic terms without the rotating-wave approximation, the formalism expresses linear and third-order susceptibilities through current–current correlation functions and retarded Liouvillian Green’s functions. In the topological phase of the SSH chain, the third-order response exhibits characteristic phase inversions and spectral asymmetries absent in the trivial phase, reflecting interband phase coherence and topological winding (Bittner et al., 16 May 2025).

For second-order dc nonlinear transport, recent work has emphasized that gauge equivalence requires complete treatment of field-dependent current vertices and a correlated adiabatic dc limit. One density-matrix derivation shows that the length and velocity gauges give the same intrinsic second-order dc response when the same retarded continuation is used for all external frequencies and when the velocity-gauge current includes all field-dependent vertices; apparent Fermi-sea terms cancel, leaving a Fermi-surface quantum geometric contribution determined by the band-normalized quantum metric, with no residual dc nonlinear Hall current in a fully gapped insulator in the adiabatic clean limit (Ahmad et al., 29 Jun 2026). A complementary velocity-gauge Keldysh derivation decomposes the clean-limit dc response into four terms,

$V^{(\lambda)}(t)$ 8

with lifetime scalings $V^{(\lambda)}(t)$ 9, $\lambda=0$ 0, $\lambda=0$ 1, and $\lambda=0$ 2, respectively, and makes the role of the Peierls contact vertices $\lambda=0$ 3 and $\lambda=0$ 4 explicit (Shibata, 21 Jun 2026).

A recurrent misconception is to attribute physical significance to the apparent propagation speed of an individual potential rather than to the fields. In the electromagnetic $\lambda=0$ 5-gauge, the scalar potential propagates with apparent speed $\lambda=0$ 6, and the vector potential contains both $\lambda=0$ 7-propagating and $\lambda=0$ 8-dependent pieces, but the gauge transformation

$\lambda=0$ 9

leaves

$\lambda=1$ 0

unchanged. A worked Hertz-dipole calculation makes this explicit: the $\lambda=1$ 1-dependent contributions cancel from the final $\lambda=1$ 2 and $\lambda=1$ 3, leaving the standard retarded fields (Giri et al., 2022).

A second recurrent issue is truncated or hybrid formulations. Exact gauge equivalence is a statement about the full theory. In practice, incomplete band spaces, phenomenological dephasing, omitted contact terms, or hybrid gauge transformations can break apparent equivalence. The solid-state velocity-gauge literature therefore insists on re-deriving the perturbation for the finite model rather than importing the continuum minimal-coupling form unchanged, while the nonlinear dc literature insists that $\lambda=1$ 4 and $\lambda=1$ 5 contributions are indispensable in the velocity gauge (Passos et al., 2017, Ahmad et al., 29 Jun 2026). This suggests that many reported gauge discrepancies are bookkeeping discrepancies rather than contradictions of gauge invariance.

A third issue is terminological drift across disciplines. In irrotational water-wave theory, “velocity gauge” refers to the freedom to add an arbitrary function of time to the velocity potential and to fix that freedom by choosing the gauge condition $\lambda=1$ 6 in the Cauchy–Lagrange equation. The same source explicitly states that it does not propose a new generalized gauge formalism; rather, it clarifies how the chosen gauge is preserved under Galilean transformation and why common traveling-wave ansätze may silently change gauges (Clamond, 2017). This separate usage does not belong to the electromagnetic $\lambda=1$ 7-gauge family, but it illustrates the broader mathematical idea behind gauge-fixing of potentials.

Across these literatures, the generalized velocity gauge is best understood as a family of exact or controlled reformulations whose value depends on what they preserve: interpolation between classical gauges, explicit longitudinal–transverse separation, restoration of cancellation structure in truncated expansions, retention of periodicity in crystals, or enforcement of gauge-consistent current operators. The persistent technical lesson is that observables remain gauge invariant only when the representation-dependent auxiliary terms required by the chosen formulation are kept in full.