Gross–Pitaevskii Vortex Dynamics

• Gross–Pitaevskii vortex dynamics is a framework describing the motion, interaction, and reconnection of quantized vortices in Bose–Einstein condensates, defined by zeros of the complex order parameter and phase winding.
• The approach integrates analytical methods, reduced-order models, and direct simulations to connect point-vortex systems, filament dynamics, and turbulent decay phenomena.
• It underlines the impact of background density, trapping geometry, and driving protocols on vortex behavior, with applications ranging from vortex lattices to quantum turbulence.

Gross–Pitaevskii vortex dynamics concerns the motion, interaction, deformation, and reconnection of quantized vortices in solutions of the Gross–Pitaevskii (GP) equation. In the GP description of a dilute, zero-temperature Bose–Einstein condensate, the condensate is represented by a single complex order parameter $\psi$, and vortices appear as isolated zeros of $|\psi|$ around which the phase winds by $2\pi n$. In two spatial dimensions these defects are point-like in the Jacobian or vorticity measure; in three dimensions they form vortex filaments that may close into rings or more complicated topological structures such as torus knots. Across asymptotic analysis, reduced-order models, direct simulations, and tracking algorithms, the subject links Hamiltonian point-vortex motion, filament dynamics, reconnection physics, Kelvin-wave cascades, and the effects of trapping, inhomogeneity, dissipation, and multicomponent structure (Proment et al., 2013).

1. Governing equations, conserved quantities, and vortex observables

In dimensional form, the zero-temperature GP equation for a condensate wavefunction $\psi$ is

$$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$

with boson mass $m$, $s$-wave scattering length $a_s$, and external potential $V_{\rm ext}$. In the homogeneous setting $V_{\rm ext}\equiv 0$, one convenient nondimensionalization uses the healing length $|\psi|$0, the time scale $|\psi|$1, and density scale $|\psi|$2, yielding

$|\psi|$3

The corresponding GP energy functional is

$|\psi|$4

and excitation energies are measured relative to the uniform ground state (Proment et al., 2013).

Trapped condensates introduce a confining potential and a different natural scaling. For an axisymmetric harmonic trap,

$|\psi|$5

one standard dimensionless form is

$|\psi|$6

with $|\psi|$7 preserved (Reichl et al., 2013).

Several asymptotic and rigorous studies instead use the small-core form

$|\psi|$8

where $|\psi|$9 is the vortex-core scale. In this regime the Ginzburg–Landau energy and the supercurrent

$2\pi n$0

play a central role, while the Jacobian

$2\pi n$1

encodes vortex positions $2\pi n$2 and degrees $2\pi n$3 (Corso et al., 2024).

The hydrodynamic representation $2\pi n$4 identifies $2\pi n$5 as the condensate density and $2\pi n$6 as the superfluid velocity away from cores. For computational geometry, the pseudo-vorticity field

$2\pi n$7

is especially useful: it is tangent to the vortex filament and remains smooth even though the physical vorticity is singular on the nodal set. This quantity underlies spectrally accurate filament-tracking algorithms in periodic domains (Villois et al., 2016).

2. Reduced vortex laws: point vortices, inhomogeneity, and filaments

In the small-core limit on the plane, the leading-order dynamics of well-separated vortices are governed by the classical Helmholtz–Kirchhoff or Kirchhoff–Onsager point-vortex system. For degree-$2\pi n$8 vortices at positions $2\pi n$9, one asymptotic form is

$\psi$0

with Hamiltonian

$\psi$1

A rigorous construction on arbitrarily large finite time intervals establishes $\psi$2-vortex GP solutions whose vortex positions admit an asymptotic expansion in $\psi$3, with the first correction determined by a linear wave equation for an outer phase field; this justifies the formal picture in which vortex accelerations radiate and the radiation reacts back on the vortex motion (Pino et al., 24 Jul 2025).

A complementary finite-$\psi$4 theorem proves that, under well-prepared initial data and with $\psi$5, GP vortices on $\psi$6 move according to the same Kirchhoff–Onsager law while their empirical vorticity converges, in a hydrodynamic limit, to weak solutions of the two-dimensional incompressible Euler equations (Jerrard et al., 2013).

Inhomogeneity changes the leading-order balance. For a non-homogeneous background density with Thomas–Fermi profile $\psi$7, the vortex law becomes

$\psi$8

so each vortex moves independently along level sets of the background density and classical pairwise interactions are subleading after the relevant $\psi$9 time rescaling (Jerrard et al., 2013). A distinct critical regime occurs when $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$0; then vortex–vortex interaction and background forcing both survive at order one, and the effective law is a mixed Hamiltonian system combining the renormalized interaction energy $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$1 with the inhomogeneity term $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$2 (Kurzke et al., 2015).

For nearly parallel three-dimensional filaments, the GP equation rigorously reduces to the Klein–Majda–Damodaran system

$$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$3

where $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$4 describes the transverse displacement of the $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$5th filament. The associated Hamiltonian is

$$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$6

which preserves the symplectic structure inherited from GP flow (Jerrard et al., 2020). In an inhomogeneous rotating condensate within the Thomas–Fermi and anelastic approximations, a non-canonical Hamiltonian for the macroscopically averaged vorticity yields an exact PDE for static vortex-lattice configurations and, for an anisotropic Gaussian density, admits spatially uniform vorticity solutions governed by a closed ODE for a vector $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$7 (Ruban, 2016).

3. Rings, dipoles, and torus knots

Vortex dipoles furnish one of the clearest examples of how GP vortex dynamics depends on initial data beyond mere core separation. For two antiparallel singly quantized vortices at distance $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$8, improved two-point Padé approximants give substantially more accurate stationary single-vortex profiles than earlier Fetter, Kerr, or Berloff ansätze. In periodic-domain GP simulations, the dipole may either annihilate into sound or propagate indefinitely as a solitary wave, and the outcome depends strongly on the initial modulus profile and on the phase contours. Relaxing the modulus under the diffusive Ginzburg–Landau flow while holding the phase fixed produces low-energy initial data that annihilate faster and follow smoother trajectories. Generalizing the usual circular phase to elliptical phase contours introduces two parameters $$i\hbar\,\partial_t\psi = -\frac{\hbar^2}{2m}\nabla^2\psi + \frac{4\pi\hbar^2 a_s}{m}|\psi|^2\psi + V_{\rm ext}\psi,$$9 and produces a sharp annihilation-versus-solitary-wave transition. For small separation and circular phase, the linearized annihilation time is $m$0, while numerically relaxed profiles give

$m$1

The AnSol boundary satisfies $m$2 for small $m$3 with $m$4 and $m$5 as $m$6 with $m$7 (Rorai et al., 2012).

Vortex rings in elongated harmonic traps display long oscillation periods that can substantially exceed the axial trap period. Phase imprinting of a $m$8 jump across $m$9 creates a dark soliton that quickly undergoes snake instability into a vortex ring, while a direct phase ansatz with prescribed $s$0 produces a ring of controlled minimum radius. In the Thomas–Fermi regime, a semiclassical estimate gives

$s$1

and simulations show that the ring period grows with interaction strength, decreases with increasing trap aspect ratio $s$2, and increases monotonically with $s$3. At fixed $s$4, $s$5 decreases from $s$6 at $s$7 to $s$8 at $s$9, and increases from $a_s$0 at $a_s$1 to $a_s$2 at $a_s$3. Slight axial anisotropy causes the ring to split into two vortex lines of opposite circulation and later recombine, yet the fundamental oscillation frequency remains essentially unchanged (Reichl et al., 2013).

Torus knots extend the GP vortex family beyond rings. For coprime integers $a_s$4, the torus knot $a_s$5 winds poloidally and toroidally around an underlying torus. An ab initio construction places $a_s$6 vortices and $a_s$7 antivortices in each meridional plane using Berloff’s Padé approximation for a single two-dimensional vortex, thereby producing a single closed $a_s$8 nodal set. Numerical evaluation of the excitation energy in a cubic box shows that for the trefoil $a_s$9 and its dual $V_{\rm ext}$0, the minima lie on straight-line fits

$V_{\rm ext}$1

At fixed $V_{\rm ext}$2, $V_{\rm ext}$3 is significantly less energetic than $V_{\rm ext}$4. Propagation is primarily along the torus axis, and the normalized axial speed obeys the fit

$V_{\rm ext}$5

with

$V_{\rm ext}$6

The characteristic axial size oscillates by about $V_{\rm ext}$7 around its initial value. When $V_{\rm ext}$8 is too large, both $V_{\rm ext}$9 and $V_{\rm ext}\equiv 0$0 undergo $V_{\rm ext}\equiv 0$1 simultaneous self-reconnections and decay into exactly $V_{\rm ext}\equiv 0$2 primary rings, followed by further Kelvin-wave-driven reconnections (Proment et al., 2013).

4. Trapping, rotation, nucleation, and stability windows

Rotating trapped condensates admit a reduced ODE description for multiple co-rotating vortices. In a two-dimensional harmonic trap with anisotropy parameter $V_{\rm ext}\equiv 0$3, matched asymptotics yields a finite-dimensional system for the vortex centers $V_{\rm ext}\equiv 0$4 that quantitatively reproduces stable vortex-crystal configurations, especially for multiple vortices. In the isotropic many-vortex limit, the theory produces an effective vortex-crystal density and lattice radius, together with an asymptotic estimate for the maximum vortex number as a function of the rotation rate. In anisotropic traps, a two-vortex pair aligned with the long axis is linearly stable, whereas a pair on the short axis is unstable; in the strongly anisotropic limit the vortices align along the long axis and the analysis yields an effective one-dimensional density and a corresponding maximum admissible number of vortices (Xie et al., 2017).

Sudden changes of interaction sign produce a different trapped-vortex problem. In the two-dimensional GP equation with harmonic trapping and optional optical lattice, switching the nonlinearity from repulsive to attractive leads a single-charged vortex to one of three outcomes: collapse, persistent survival as a breathing vortex, or loss of topological charge followed by relaxation into a fundamental soliton. For zero optical lattice depth, the critical norm for collapse of the $V_{\rm ext}\equiv 0$5 vortex is $V_{\rm ext}\equiv 0$6. At $V_{\rm ext}\equiv 0$7, the numerically identified thresholds are $V_{\rm ext}\equiv 0$8, $V_{\rm ext}\equiv 0$9, and ultimate collapse for $|\psi|$00. Increasing the optical-lattice strength expands the survival window to larger norms (Chen et al., 2012).

Annular geometries make nucleation and boundary injection particularly explicit. In a two-dimensional ring potential with Gaussian radial profile and a rotating sinusoidal perturbation,

$|\psi|$01

a finite-volume Strang-splitting computation with benchmark parameters $|\psi|$02, $|\psi|$03, $|\psi|$04, $|\psi|$05, $|\psi|$06, $|\psi|$07, and $|\psi|$08 observed no vortices for $|\psi|$09 over $|\psi|$10, whereas for $|\psi|$11 vortices nucleated from both the inner and outer edges and, by $|\psi|$12, produced 12 vortices: six of index $|\psi|$13 near the outer edge and six of index $|\psi|$14 near the inner edge (Chauleur et al., 2024).

These results delimit several distinct notions of stability in GP vortex dynamics. In rotating traps, stability refers to persistent co-rotating equilibria or crystals; in quenched attractive systems it refers to survival against collapse and against topological demotion; in annular stirring problems it refers to nucleation thresholds, edge-mediated injection, and the long-time arrangement of defects. A plausible implication is that GP vortices are best viewed not as a single universal defect class but as topological excitations whose effective dynamics depend sensitively on the competition among core energy, background density, trapping geometry, and forcing protocol.

5. Tracking vortex filaments, turbulent tangles, and reduced-order numerics

Direct numerical study of GP vortex dynamics depends on the ability to identify vortex cores accurately in the complex field. In periodic domains, a robust method is to locate points where $|\psi|$15 and $|\psi|$16 by Newton–Raphson iteration in the plane normal to the pseudo-vorticity $|\psi|$17. One first finds low-density seed points, constructs the normal plane using the local tangent $|\psi|$18, iterates to machine precision, and then marches along the filament with step size $|\psi|$19 until closure or periodic re-entry. From the reconstructed curves one computes curvature

$|\psi|$20

torsion

$|\psi|$21

and Kelvin-wave occupations from Fourier transforms of transverse displacements (Villois et al., 2016).

Applied to a GP-driven turbulent tangle generated from a Taylor–Green configuration, this tracking framework resolves the total vortex-line density

$|\psi|$22

and shows Vinen decay,

$|\psi|$23

with best-fit coefficient $|\psi|$24, in quantitative agreement with low-temperature He II experiments. The same data reveal rare linked rings during decay, curvature PDFs with exponential tails at large $|\psi|$25, torsion PDFs with symmetric $|\psi|$26 tails, and a Kelvin-wave spectrum

$|\psi|$27

in the inertial range, consistent with the L’vov–Nazarenko weak-wave-turbulence prediction (Villois et al., 2016).

For two-dimensional small-core problems, vortex tracking also enables reduced-order simulation. A numerical strategy based on the reduced Hamiltonian point-vortex system integrates the ODEs for vortex centers, reconstructs a smoothed GP field from the canonical harmonic map and precomputed radial core profile, and proves rigorous supercurrent-error estimates of the form

$|\psi|$28

up to the first vortex-collision time. In practice, the reconstructed field reproduces the density dips and far-field phase at a tiny fraction of the cost of a full GP solve when $|\psi|$29 is small (Corso et al., 2024).

Boundary conditions can themselves be a dynamical tool. Quasi-periodic boundary conditions generalize ordinary periodicity by keeping $|\psi|$30 strictly periodic while allowing prescribed phase windings compatible with net vorticity. In this framework the phase jump over a lattice translation $|\psi|$31 is fixed by the total vortex count $|\psi|$32, and Kelvin’s circulation theorem together with the boundary matching implies that the net vortex number is conserved. Real-time simulations use a third-order Adams–Bashforth integrator and nearest-neighbour vortex matching, and they reveal bulk phenomena unavailable in periodic boxes with zero net circulation, including perfectly periodic vortex arrays, vortex depinning, pair nucleation, and Kármán vortex streets. The same formulation is proposed as a toy model for neutron-star bulk vortex dynamics (Magistrelli et al., 18 Sep 2025).

6. Multicomponent, stochastic, and nonequilibrium generalizations

In multicomponent condensates beyond the Thomas–Fermi regime, variational reductions show that vortex cores may acquire inertia. For a two-dimensional $|\psi|$33-component GP system with one vortex per component, elimination of phase-gradient collective coordinates yields an effective Hamiltonian

$|\psi|$34

with effective magnetic field, harmonic confinement, and short-range singular repulsion. The resulting Newton–Lorentz equations differ fundamentally from massless point-vortex dynamics and allow chaos in the three-vortex problem, which is not expected for the corresponding inertia-free planar point vortices (1212.00165).

A related binary-mixture model treats vortices of one component with small filled cores of a second component. A time-dependent variational derivation produces the point-vortex Lagrangian

$|\psi|$35

so the canonical momentum contains an effective vector-potential contribution. For a single positive vortex in a rigid circular boundary, the massless limit gives uniform precession, whereas a sufficiently large filled core destabilizes that precession. For realistic GP simulations with $|\psi|$36, $|\psi|$37 $|\psi|$38Na atoms, and $|\psi|$39 $|\psi|$40K atoms, the extracted radial-oscillation frequency is approximately $|\psi|$41, in excellent agreement with the analytic prediction $|\psi|$42 (Richaud et al., 2020).

Stochastic and driven variants of GP dynamics alter both vortex nucleation and spectral transfer. In a two-dimensional Bose–Hubbard lattice with synthetic magnetic field, white noise, and dissipation, the stochastic GP equation generates vortices in the bulk and leads to stable steady states after a transition period. The incompressible kinetic-energy spectrum shows a Kolmogorov $|\psi|$43 law in the infrared during the transition and a $|\psi|$44 ultraviolet core regime, while steady states exhibit an infrared $|\psi|$45 law. The appearance of the $|\psi|$46 regime correlates empirically with a fastest-vortex speed $|\psi|$47 (Kato et al., 2017).

Generalized GP equations for strongly nonequilibrium quantum fluids produce another qualitative shift. In the driven-dissipative model

$|\psi|$48

vortex motion becomes self-accelerated, long-range vortex–antivortex interaction becomes repulsive, and annihilation can be dramatically slowed. In finite samples, relaxation may leave metastable vortex–antivortex clusters whose geometry depends on the sample shape, pump profile, and nonequilibrium strength; at sufficiently strong driving, self-accelerated vortices may also nucleate new vortex–antivortex pairs (Gladilin et al., 2018).

Taken together, these extensions indicate that “Gross–Pitaevskii vortex dynamics” denotes a family of related dynamical regimes rather than a single asymptotic law. In conservative homogeneous settings the dominant structures are Hamiltonian point vortices and filaments; in trapped or inhomogeneous condensates the background density enters the leading motion; in three dimensions reconnections, Kelvin waves, and topology become unavoidable; and in multicomponent, stochastic, or driven-dissipative systems vortices may acquire inertia, altered interaction laws, or self-acceleration. This suggests that the unifying object is not a particular reduced equation, but the evolution of phase singularities in a nonlinear dispersive order parameter whose core scale, background profile, and forcing determine the effective defect mechanics.