Wigner Current: Quantum Phase-Space Flow

• Wigner Current is the phase‐space flow derived from the Wigner function, forming a quasi-probability continuity equation that replaces classical trajectories.
• It uncovers nonclassical features such as negative probability regions, flow reversals, and discrete topological defects driven by higher order quantum corrections.
• Experimental reconstructions using squeezed-vacuum states and quantum-state tomography validate its ability to separate Hamiltonian, damping, and diffusion contributions in quantum dynamics.

Searching arXiv for recent and foundational papers on Wigner current. Wigner current is the phase-space current associated with the Wigner function $W$, defined by the exact continuity equation

$$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$

In Wigner’s phase-space formulation of quantum mechanics, it replaces point-wise trajectories by a conserved quasi-probability flow. Its vector field, fieldlines, and stagnation sets provide a quantum phase portrait for both closed and open dynamics, while also exposing structures that have no classical analogue, including flow reversal in regions of negative $W$, discrete topological defects, and nontrivial limits as $\hbar\to0$ or anharmonicity vanishes (Kakofengitis et al., 2014, Steuernagel et al., 2023).

1. Definition and continuity structure

For a one-dimensional system with density operator $\hat\rho(t)$, one convenient definition of the Wigner function is

$$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$

If the Hamiltonian is $H(x,p)=p^2/(2m)+V(x)$, the von Neumann equation maps to the Moyal equation

$\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$

with

$$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$

Reorganizing the Moyal expansion yields the continuity equation above, with the closed-system current

$J_x(x,p,t)=\frac{p}{m}W(x,p,t),$

$$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$0

An equivalent form isolates the classical force term,

$$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$1

Physically, $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$2 is the flow of quasi-probability in phase space: quantum dynamics remains conservative in the variables $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$3, even though $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$4 is not a positive probability density (Steuernagel et al., 2023).

2. Closed-system flow and its relation to classical mechanics

The classical Liouville current for $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$5 is

$$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$6

The quantum current agrees with this at leading order, but the higher even powers of $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$7 in $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$8 are purely quantum corrections. These terms encode interference and tunneling effects, and they give rise to non-Liouvillian flow, in the sense that $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$9 locally; they also produce regions where $W$0 and the current reverses, forming the “whorls” discussed in phase-space analyses (Kakofengitis et al., 2014, Steuernagel et al., 2023).

The harmonic oscillator is the special case in which the current reduces to

$W$1

so the fieldlines are circular. For harmonic eigenstates, the Wigner function has alternating positive and negative rings, and the zero-rings of $W$2 coincide with stagnation lines of $W$3. This exact coincidence is exceptional. As soon as anharmonicity is introduced, the zeros of $W$4 and $W$5 shift differently, and the stagnation structure becomes discrete rather than line-like (Kakofengitis et al., 2014).

A central result for weakly anharmonic systems is that neither the limit $W$6 nor the limit of vanishing anharmonicity yields pointwise convergence of the Wigner current to the classical flow or to the harmonic-oscillator circular flow. This rejects a common intuition that the current field should collapse smoothly, point by point, onto the classical portrait (Kakofengitis et al., 2014).

3. Stagnation points, winding numbers, and topological constraints

A stagnation point is a phase-space point at which $W$7. In the harmonic oscillator, continuous stagnation lines appear because $W$8 and $W$9 both vanish wherever $\hbar\to0$0. In weakly anharmonic systems, those lines split into isolated crossings of the $\hbar\to0$1 and $\hbar\to0$2 contours. The resulting discrete defects organize the current fieldline pattern (Kakofengitis et al., 2014).

Each stagnation point carries an integer Poincaré–Hopf index, or winding number,

$\hbar\to0$3

where $\hbar\to0$4 and $\hbar\to0$5 is a small counter-clockwise loop encircling the defect. In the weakly anharmonic analyses, only $\hbar\to0$6 and $\hbar\to0$7 occur, corresponding respectively to vortices and saddles. Time evolution can create or annihilate pairs of opposite charge, but the total sum of $\hbar\to0$8 over the plane is conserved (Kakofengitis et al., 2014).

This topological viewpoint extends naturally to time-dependent superpositions. For the ground–first-excited superposition

$\hbar\to0$9

the zero-circle of $\hat\rho(t)$0 remains circular while its center spirals, and its stagnation points move rigidly “like markers on a Ferris-wheel.” Under slow Rabi driving, the pattern expands and contracts; opposite-charge pairs may be created or annihilated, but the total index remains invariant (Kakofengitis et al., 2014).

In an open-system optical realization, the ideal squeezed-vacuum dynamics of a degenerate optical parametric oscillator has a central stagnation point at the origin with topological charge $\hat\rho(t)$1. Experimental work showed that this charge remains invariant under both weakly and strongly decohering conditions, provided no pair-creation or pair-annihilation event transfers a defect through the origin. In that sense the central charge is topologically protected (Chen et al., 2021).

4. Open-system extensions and Lindblad structure

For a single bosonic mode weakly coupled to a thermal bath, the phase-space continuity equation acquires environmental terms,

$\hat\rho(t)$2

In units where $\hat\rho(t)$3, the environmental currents are

$\hat\rho(t)$4

The damping part is a friction-type drift back toward the origin; the diffusion part spreads $\hat\rho(t)$5 out. Together they enforce relaxation to a thermal Gaussian (Steuernagel et al., 2023).

In dimensionful variables, the same split can be written as

$\hat\rho(t)$6

$\hat\rho(t)$7

This decomposition makes it possible to analyze decoherence geometrically. In particular, for a region where $\hat\rho(t)$8, Gauss’s theorem yields an exact flux law for the integrated negativity. Diffusion always points inward on a negative region and therefore shrinks it, whereas the system current can point outward on parts of the boundary and can partially counteract decoherence. In the harmonic case negativity always decays; in the driven Duffing case, the leading quantum correction can pump probability so as to enlarge the negative island during part of a drive cycle, and a stabilization criterion can be written in terms of the boundary fluxes (Jr. et al., 2017).

A further structural result is that requiring all dynamical terms to be expressible as a total divergence in phase space forces the dissipative coupling into Lindblad form. In the star-product representation, only the combinations

$\hat\rho(t)$9

and

$$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$0

produce the needed divergence terms. The resulting phase-space dissipator is exactly the Wigner-space image of the operator-space Lindblad theorem, and hence preserves trace, hermiticity, and complete positivity (Steuernagel et al., 2023).

5. Canonical models and generalized phase spaces

Weakly anharmonic one-dimensional bound systems admit a threefold classification. For

$$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$1

with small $$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$2, one obtains hard potentials ($$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$3 even, $$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$4), soft potentials ($$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$5 even, $$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$6), and odd potentials ($$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$7 odd, $$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$8). In hard potentials, the zero-circles of $$W(x,p,t)=\frac{1}{2\pi\hbar}\int_{-\infty}^{\infty}dy\;e^{ipy/\hbar}\, \langle x-y/2|\hat\rho(t)|x+y/2\rangle.$$9 are squeezed along $H(x,p)=p^2/(2m)+V(x)$0 and stretched along $H(x,p)=p^2/(2m)+V(x)$1; in soft potentials the deformation is reversed; odd potentials behave as hard on one side and soft on the other, with a cut-and-reconnect structure near $H(x,p)=p^2/(2m)+V(x)$2. For the first excited state, the hard quartic and soft quartic cases yield nine stagnation points per ring, while the odd cubic case yields seven because part of the axis-crossing structure avoids $H(x,p)=p^2/(2m)+V(x)$3 through reconnection (Kakofengitis et al., 2014).

For the Kerr oscillator,

$H(x,p)=p^2/(2m)+V(x)$4

Oliva and Steuernagel derived

$H(x,p)=p^2/(2m)+V(x)$5

Because $H(x,p)=p^2/(2m)+V(x)$6, the current is everywhere tangent to circles $H(x,p)=p^2/(2m)+V(x)$7, and the probability on each circle is constant in time. Decomposing $H(x,p)=p^2/(2m)+V(x)$8, with

$H(x,p)=p^2/(2m)+V(x)$9

$\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$0

isolates a purely quantum term that suppresses classical shear. The global measure

$\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$1

initially grows and then settles to a negative plateau for $\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$2, signaling effective “viscosity” and the enforcement of the Zurek scale. Deviations of $\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$3 from its steady behavior pick out fractional revivals and special nonclassical states (Oliva et al., 2018).

Wigner current has also been formulated in several nonstandard geometries. For variable beam splitters with bilinear mode-mixing Hamiltonian

$\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$4

the reduced current of one outgoing mode after tracing out the other is expressed through cross-moments of the two-mode Wigner function. In this setting, coherent–Fock mixing produces circulating currents around toroidal negative regions, the Hong–Ou–Mandel configuration yields strictly radial currents with sink–source structure inside the negative “hole,” and the ratio $\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$5 becomes singular wherever $\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$6, signaling volume-nonconserving quantum dynamics (Steuernagel et al., 2023).

For spin systems, the SU(2) Wigner function on the sphere satisfies

$\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$7

and for Kerr evolution $\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$8 one finds

$\partial_t W(x,p,t)=\{\!\!\{H,W\}\!\!\}(x,p,t),$9

In the classical limit $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$0, $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$1 and the flow becomes a twist around the $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$2-axis, $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$3. Quantum mechanically, additional stagnation loops, elliptic points, and hyperbolic saddles appear; these are absent in the classical current and arise from the full deformation operator $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$4 (Yang et al., 2018).

In multidimensional quantum billiards, the Wigner current satisfies

$$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$5

while the momentum component becomes a surface integral over the boundary manifold in chord space,

$$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$6

An equivalent description uses the boundary potential $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$7. Here the wall-induced quantum effects are carried entirely by $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$8; at finite $$\{\!\!\{f,g\}\!\!\} = \frac{1}{i\hbar}(f\star g-g\star f) = \frac{2}{\hbar}f(x,p)\sin\!\Bigl[\frac{\hbar}{2} \bigl(\overleftarrow{\partial_x}\overrightarrow{\partial_p} -\overleftarrow{\partial_p}\overrightarrow{\partial_x}\bigr)\Bigr]g(x,p).$$9, the surface integral retains nonlocal contributions even in regions that are otherwise free (Seidov et al., 2024).

6. Experimental reconstruction and interpretive significance

The first experimental reconstruction of Wigner current was reported for squeezed-vacuum states generated by a degenerate optical parametric oscillator. In the rotating frame, the ideal system Hamiltonian is

$J_x(x,p,t)=\frac{p}{m}W(x,p,t),$0

with associated system current

$J_x(x,p,t)=\frac{p}{m}W(x,p,t),$1

In the squeezed-vacuum phase $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$2, this reduces to hyperbolic flow that pushes the Wigner density along the squeezed and anti-squeezed quadratures. The environment adds the damping and diffusion currents discussed above, yielding the experimentally observed “push-and-pull” picture: damping pulls inward, while diffusion pushes outward and enforces Heisenberg’s uncertainty relations even at zero temperature (Chen et al., 2021).

The reported setup used a bow-tie cavity with a PPKTP crystal pumped at $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$3 to generate squeezed vacuum at $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$4. Below-threshold operation yielded pure squeezing up to $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$5 dB. State tomography employed balanced homodyne detection at a $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$6 sideband with common-mode rejection $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$7 dB, and a machine-learning-enhanced quantum-state-tomography algorithm reconstructed $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$8 for 20 discrete pump powers. Because the squeezing operator implemented by the OPO is formally equivalent to time evolution under $J_x(x,p,t)=\frac{p}{m}W(x,p,t),$9, pump power could be mapped to an effective time $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$00 (Chen et al., 2021).

The current itself was reconstructed numerically from a sequence of Wigner functions $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$01 at effective times $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$02. An initial guess $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$03 was formed from $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$04 with fitted $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$05 and $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$06, and the discrete continuity equation was then solved by minimizing $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$07 subject to the continuity constraint, cast as a linear-programming problem. The reconstructed current $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$08 showed the expected separation into system, damping, and diffusion components, and confirmed that the central stagnation point retained $$\partial_t W(x,p,t)+\partial_x J_x(x,p,t)+\partial_p J_p(x,p,t)=0.$$09 under both weak and strong decoherence (Chen et al., 2021).

These developments clarify the interpretive role of Wigner current. It is not merely an auxiliary rewriting of the Moyal equation. In closed systems it resolves flow reversal, interference, and topological defects more finely than a trajectory-based description; in open systems it separates Hamiltonian, damping, and diffusion mechanisms; and in generalized settings such as beam splitters, spin phase spaces, and billiards it provides a common geometric language for nonclassical transport, decoherence, and boundary effects (Kakofengitis et al., 2014, Steuernagel et al., 2023).