betaFeedback

WorksPhysics and Astronomy

Collective Dynamics of Vortex Clusters in Compact Fluid Domains: From Pair Interactions to a Quadrupole Description

Abstract

Clusters of co-rotating vortices on compact fluid domains exhibit a simple collective dynamics, combining coherent global rotation with a slow breathing of the cluster size. In this work, we investigate an analytic model of vortex interactions on a doubly periodic inviscid fluid domain, based on an exact representation in terms of the Schottky-Klein prime function and its qq-representation. The two-vortex problem reduces to a single complex degree of freedom, from which explicit expressions for the orbital rotation frequency and dipole translation velocity are obtained. Building on this framework, we derive a small-cluster expansion that reveals a decomposition of the dynamics into universal planar interactions, isotropic torus corrections, and geometry-induced anisotropic modes. At leading order, the collective dynamics is encoded in a single complex quadrupole moment: its real part governs corrections to the rotation rate, while its imaginary part controls the slow breathing of the cluster. These predictions are quantitatively confirmed by direct numerical simulations, establishing a reduced description of vortex clusters on the flat torus and compact fluid domains.

Introduction

The study of point vortices in two-dimensional incompressible and inviscid flows has long served as a bridge between discrete vortex models and continuum fluid mechanics. Point vortices provide an idealized description in which vorticity is concentrated at discrete points, reducing complex fluid motion to a finite-dimensional dynamical system. Interest in vortex dipoles, clusters, and related collective structures has grown substantially in recent years, driven by experimental and theoretical developments in quantum fluids, active matter, and hydrodynamic rotor systems. In particular, vortex dipoles and clusters have been observed and characterized in trapped Bose-Einstein condensates and two-dimensional quantum fluids [2, 3, 4, 5] [1], while related collective phenomena have been explored in superfluid turbulence and vortex dynamics [6, 7, 8], anomalous vortex fluids [9], and in astrophysical settings such as vortex assemblies in the superfluid interior of neutron stars [10]. Similar collective dynamics also arise in soft-matter and active systems, including plane-confined microrotors and their mixtures [11, 12], hydrodynamically interacting rotating inclusions and rotor assemblies [13, 14], as well as vortex flows on curved membranes [15]. These developments provide strong motivation for reduced descriptions of interacting vortex ensembles.

Much of the theoretical and computational literature on vortex dipoles and clusters has focused on unbounded or effectively planar geometries, where translational symmetry and the absence of global image interactions considerably simplify the dynamics [16]. Compact periodic domains, however, introduce qualitatively new features. In such geometries, each vortex interacts with an infinite lattice of its periodic images, and the resulting motion is governed by the Green’s function of the Laplace operator on the flat torus. This global coupling modifies both few-body motion and collective many-body dynamics, making periodic domains a natural setting in which geometry and topology enter explicitly into vortex interactions. Point-vortex dynamics in doubly periodic domains has since been explored in detail [17], including analytic descriptions in finite geometries using elliptic-function methods [18], as well as studies of few-vortex configurations and dipoles emphasizing integrability and the effect of periodic boundary conditions [19, 20, 21].

A major advance in the analysis of vortex motion in multiply connected domains is the use of the Schottky-Klein prime function, which provides a compact representation of the hydrodynamic Green’s function and incorporates periodic image interactions [22, 23, 24]. This approach yields explicit interaction kernels and equations of motion for NN vortices in doubly periodic geometries [25], with an equivalent formulation in terms of the qq-digamma function ψρ(z)\psi_{\rho}(z) enabling efficient computation and extensions such as harmonic fields [26, 27].

In parallel, vortex dynamics on compact surfaces has been studied from geometric and topological perspectives, where curvature and global constraints play a central role [28]. Recent formulations based on harmonic one-forms and Hodge decomposition provide a general framework for flows on higher-genus surfaces [29], while studies of vortex pairs and dipoles reveal connections to magnetic geodesics [15, 30, 31, 32, 33]. For the flat torus, the harmonic component reduces to a constant background flow and can be removed without loss of generality [29], making it especially amenable to analytic treatment. Motivated by these developments, the present work studies vortex clusters in doubly periodic planar domains using the analytic framework of the Schottky-Klein prime function and its qq-digamma representation [26]. While periodic geometries introduce additional complexity through global topology and image interactions, they also admit simple analytic structure. Building on recent developments in this direction [26], we show that this framework extends naturally to large vortex ensembles on the flat torus. At the same time, recent experimental studies of vortex clustering in quantum fluids [5] provide additional motivation for theoretical work in such directions.

Before proceeding, we note that the present work complements and builds on a substantial body of prior studies on vortex interactions in doubly periodic domains [11, 17, 18, 19, 20, 21, 24, 25]. It advances this line of research by developing qq-function-based closed-form expressions for the interaction kernel and associated dynamical quantities, enabling analytic reductions and a coarse-grained description of large-NN vortex clusters in compact geometries. In particular, the two-vortex problem is reduced to a single complex degree of freedom, yielding explicit expressions for the orbital rotation frequency and dipole translation velocity, while for many-vortex configurations a decomposition of the dynamics emerges into universal planar interactions, isotropic torus corrections, and geometry-induced anisotropic modes. At leading order, the collective dynamics is governed by a single complex quadrupole moment: its real part controls corrections to the rotation rate, while its imaginary part determines the slow breathing of the cluster.

The paper is organized as follows. In § 2, we formulate the Hamiltonian structure of point-vortex motion on the flat torus and derive the exact equations of motion in annulus variables, together with the closed-form kernel representation. In § 3, we establish the conserved quantities and use the kernel antisymmetry to obtain an exact reduction of the two-vortex problem, leading to a complete description of binary dynamics. In § 4, we develop a small-cluster expansion based on the local form of the kernel, yielding a reduced description for compact same-sign vortex clusters. In § 5, we derive a coarse-grained expression for the collective angular velocity in terms of the cluster size and quadrupole moment and detailed numerical validation of these predictions. In § 6, we obtain an analytic evolution law for the cluster size, showing that its slow variation is governed by the imaginary part of the quadrupole moment. We conclude in § 7, and relegate technical details on the Schottky-Klein formalism and kernel expansions to the Appendices.

Model Setup

Motivated by the structure outlined above, we revisit the formulation of the NN-vortex interactions in a doubly periodic rectangular domain recently introduced by Sakajo and Krishnamurthy [25], based on the hydrodynamic Green’s function expressed in terms of the Schottky-Klein prime function. A key advance of [25] is the derivation of a closed Hamiltonian system valid for arbitrary total circulation, in which the effects of periodicity and background vorticity arise intrinsically from the compact geometry.

Building on this framework, we exploit its equivalent qq-digamma representation developed in [26] to recast the dynamics in a form that makes the analytic structure of the interaction kernel explicit, thereby enabling reductions and coarse-grained descriptions.

We consider NN point vortices in a doubly periodic rectangular domain with periods 2π2\pi and −log⁡ρ-\log\rho (0<ρ<10<\rho<1), with vortex strengths Γj\Gamma_j and complex positions wj=xj+iyjw_j=x_j+iy_j. It is well known that the canonical symplectic structure of the point vortex system gives the Hamiltonian equations

Γjx˙j=∂H∂yj,Γjy˙j=−∂H∂xj,(1)\Gamma_j\dot{x}_j=\frac{\partial H}{\partial y_j},\qquad\Gamma_j\dot{y}_j=-\frac{\partial H}{\partial x_j}, \tag*{(1)}

where HH is the interaction Hamiltonian. Introducing the complex coordinate wj=xj+iyjw_j=x_j+iy_j, we combine (1) to obtain the complex Hamiltonian form

Γjdwjdt=−2i∂H∂w‾j.(2)\Gamma_j\frac{d w_j}{d t}=-2i\frac{\partial H}{\partial\overline{w}_j}. \tag*{(2)}

Following the approach of [24, 25] and related works, we introduce annulus coordinates

νj=eiwj,wj=−ilog⁡νj.(3)\nu_j=e^{i w_j},\qquad w_j=-i\log\nu_j. \tag*{(3)}

and (2) yields

Γjdwjdt=−2νj∂H∂νj.(4)\Gamma_j\frac{d w_j}{d t}=-2\nu_j\frac{\partial H}{\partial\nu_j}. \tag*{(4)}

The vortex Hamiltonian on the flat torus can be written as the sum of pairwise interactions and a Robin self term:

H(ν1,…,νN)=−∑1≤j<k≤NΓjΓkG(νjνk;ρ)−12∑j=1NΓj2G^(νj;ρ).(5)H(\nu_1,\ldots,\nu_N)=-\sum_{1\leq j<k\leq N}\Gamma_j\Gamma_kG\left(\frac{\nu_j}{\nu_k};\sqrt{\rho}\right)-\frac{1}{2}\sum_{j=1}^{N}\Gamma_j^2\widehat{G}\left(\nu_j;\sqrt{\rho}\right). \tag*{(5)}

with the pairwise Green function expressed through the Schottky-Klein prime function PP defined in terms of a generic variable ζ\zeta (see Appendix § 10 for more details on the Schottky-Klein machinery)

G(ζ;ρ)=12πlog⁡∣P(ζ,ρ)∣−14πlog⁡∣ζ∣+14πlog⁡ρ(log⁡∣ζ∣)2,(6)G(\zeta;\sqrt{\rho})=\frac{1}{2\pi}\log\left|P(\zeta,\sqrt{\rho})\right|-\frac{1}{4\pi}\log|\zeta|+\frac{1}{4\pi\log\rho}\left(\log|\zeta|\right)^2, \tag*{(6)}

and the Robin (self) term

G^(ν;ρ)=12πlog⁡∣∏m=1∞(1−ρm)2∣.(7)\widehat{G}(\nu;\sqrt{\rho})=\frac{1}{2\pi}\log\left|\prod_{m=1}^{\infty}(1-\rho^m)^2\right|. \tag*{(7)}

For the flat torus the Robin function (7) is independent of ν\nu (a geometry–dependent constant), hence

νj∂∂νjG^(νj;ρ)=0,ν‾j∂∂ν‾jG^(νj;ρ)=0.(8)\nu_j\frac{\partial}{\partial\nu_j}\widehat{G}(\nu_j;\sqrt{\rho})=0,\qquad\overline{\nu}_j\frac{\partial}{\partial\overline{\nu}_j}\widehat{G}(\nu_j;\sqrt{\rho})=0. \tag*{(8)}

Therefore only the pairwise part of (5) contributes. Using this fact and inserting (5) and (6) into the Hamilton’s equation (4) and dividing by Γj\Gamma_j, yields the fundamental NN-vortex dynamical equation describing motion of NN point vortices of circulations Γ1,…,ΓN\Gamma_1,\ldots,\Gamma_N in a doubly-periodic rectangular domain with periods 2π2\pi and −log⁡ρ-\log\rho (0<ρ<10 < \rho< 1) (note our original flat coordinates wj=xj+iyjw_j=x_j+\mathrm{i}y_j) as

dw‾jdt=12π∑k=1k≠jNΓkK(νjνk,ρ)−14π∑k=1k≠jNΓk+12πlog⁡ρ∑k=1k≠jNΓklog⁡∣νjνk∣,(9)\frac{d\overline{w}_j}{dt}=\frac{1}{2\pi}\sum_{\substack{k=1\\k\ne j}}^N\Gamma_k K\left(\frac{\nu_j}{\nu_k},\sqrt{\rho}\right)-\frac{1}{4\pi}\sum_{\substack{k=1\\k\ne j}}^N\Gamma_k+\frac{1}{2\pi\log\rho}\sum_{\substack{k=1\\k\ne j}}^N\Gamma_k\log\left|\frac{\nu_j}{\nu_k}\right|, \tag*{(9)}

where νj=eiwj\nu_j=e^{\mathrm{i}w_j} are coordinates on the concentric annulus. A closed analytic form for the function KK, introduced in [26], is given by

K(ζ,ρ)=11−ζ+1log⁡ρ[ψρ(log⁡(1/ζ)log⁡ρ)−ψρ(log⁡(ζ)log⁡ρ)],(10)K(\zeta,\sqrt{\rho})=\frac{1}{1-\zeta}+\frac{1}{\log\rho}\left[\psi_\rho\left(\frac{\log(1/\zeta)}{\log\rho}\right)-\psi_\rho\left(\frac{\log(\zeta)}{\log\rho}\right)\right], \tag*{(10)}

where ψρ(z)\psi_\rho(z) denotes the qq-digamma function with base q=ρq=\rho, defined by the logarithmic derivative of the qq-gamma function Γρ(z)\Gamma_\rho(z),

ψρ(z)=ddzlog⁡Γρ(z),Γρ(z)=(1−ρ)1−z∏n=0∞1−ρn+11−ρn+z,(11)\psi_\rho(z)=\frac{d}{dz}\log\Gamma_\rho(z),\qquad\Gamma_\rho(z)=(1-\rho)^{1-z}\prod_{n=0}^{\infty}\frac{1-\rho^{n+1}}{1-\rho^{n+z}}, \tag*{(11)}

which converges for 0<ρ<10 < \rho< 1 and all complex zz. Noting that νk=eiwk\nu_k=e^{\mathrm{i}w_k}, the logarithmic term in (9) reduces to

log⁡∣νjνk∣=−(yj−yk).\log\left|\frac{\nu_j}{\nu_k}\right|=-(y_j-y_k).

Fundamental Motion of the Vortex Binary

The dissipationless dynamics (9) inherits the Hamiltonian structure of the flat torus and therefore admits the conserved Hamiltonian HH together with the circulation-weighted centroid

C:=∑j=1NΓjwj,C‾:=∑j=1NΓjw‾j.(12)C:=\sum_{j=1}^{N}\Gamma_j w_j,\qquad\overline{C}:=\sum_{j=1}^{N}\Gamma_j\overline{w}_j. \tag*{(12)}

The conservation of HH follows directly from (2) and its complex conjugate, while the conservation of CC is obtained by writing (9) in the form

dwˉjdt=∑k=1k≠jNΓkF(νjνk),F(ζ)=12πK(ζ,ρ)−14π+12πlog⁡ρlog⁡∣ζ∣.(13)\frac{d\bar{w}_j}{dt}=\sum_{\substack{k=1\\k\ne j}}^{N}\Gamma_k F\left(\frac{\nu_j}{\nu_k}\right),\qquad F(\zeta)=\frac{1}{2\pi}K(\zeta,\sqrt{\rho})-\frac{1}{4\pi}+\frac{1}{2\pi\log\rho}\log|\zeta|. \tag*{(13)}

Using K(1/ζ)=1−K(ζ)K(1/\zeta)=1-K(\zeta) and log⁡∣1/ζ∣=−log⁡∣ζ∣\log|1/\zeta|=-\log|\zeta| we obtain the antisymmetric relation

F(1/ζ)=−F(ζ),(14)F(1/\zeta)=-F(\zeta), \tag*{(14)}

so that the double sum in dC‾/dtd\overline{C}/dt cancels pairwise. In real variables, the centroid conservation is equivalent to the two translational invariants

∑j=1NΓjxj,∑j=1NΓjyj.(15)\sum_{j=1}^{N} \Gamma_j x_j,\qquad\sum_{j=1}^{N} \Gamma_j y_j. \tag*{(15)}

The total circulation is also trivially conserved. Unlike the planar problem, no further invariant associated with continuous rotational symmetry exists on the flat torus.

The same antisymmetry governs relative motion. Defining wjk=wj−wkw_{jk}=w_j-w_k, we have

dw‾jkdt=(Γj+Γk)F(νjνk)+∑ℓ=1ℓ≠j,kNΓℓ[F(νjνℓ)−F(νkνℓ)].(16)\frac{d\overline{w}_{jk}}{dt}=(\Gamma_j+\Gamma_k)F\left(\frac{\nu_j}{\nu_k}\right)+\sum_{\substack{\ell=1\\\ell\ne j,k}}^{N}\Gamma_\ell\left[F\left(\frac{\nu_j}{\nu_\ell}\right)-F\left(\frac{\nu_k}{\nu_\ell}\right)\right]. \tag*{(16)}

For N=2N=2 this closes exactly:

dw‾12dt=(Γ1+Γ2)F(ν1ν2).(17)\frac{d\overline{w}_{12}}{dt}=(\Gamma_1+\Gamma_2)F\left(\frac{\nu_1}{\nu_2}\right). \tag*{(17)}

We see immediately from (17) that when Γ1+Γ2=0\Gamma_1+\Gamma_2=0 the relative coordinate is frozen,

dw12dt=0,(18)\frac{dw_{12}}{dt}=0, \tag*{(18)}

so the two vortices translate rigidly as a dipole with constant separation, as shown in the lower panels of Figure 1. For generic vortex strengths, the binary dynamics depends on the total circulation and reduces to a single complex degree of freedom. Introducing the variable η\eta defined as

Image not extracted

Figure 1. Two-vortex motion on the flat torus. Top row: nonzero total circulation (chiral vortex pair), showing periodic trajectories and oscillatory inter-vortex distance. Bottom row: dipole case (Γ1=−Γ2\Gamma_1=-\Gamma_2), exhibiting rigid motion with constant separation. Green and black dots indicate the initial positions of vortices with circulations +1+1 and −1-1, respectively; the same color scheme is used for the corresponding trajectories.

η:=ν1ν2=ei(w1−w2),(19)\eta:=\frac{\nu_1}{\nu_2}=e^{i(w_1-w_2)}, \tag*{(19)}

For Γtot≠0\Gamma_{\mathrm{tot}}\ne0, the conserved centroid determines the center of motion and the binary is reconstructed from the relative coordinate Δ:=w1−w2\Delta:=w_1-w_2. Since η=eiΔ\eta=e^{i\Delta}, (17) gives

η˙=iηΓtotF(η),(20)\dot{\eta}=i\eta\Gamma_{\mathrm{tot}}F(\eta), \tag*{(20)}

and hence the quadrature

∫η0η(t)dζiζF(ζ)=Γtot(t−t0).(21)\int_{\eta_0}^{\eta(t)} \frac{d\zeta}{i\zeta F(\zeta)} = \Gamma_{\mathrm{tot}}(t-t_0). \tag*{(21)}

For Γtot≠0\Gamma_{\mathrm{tot}} \ne0, the pair undergoes nontrivial relative motion and the inter-vortex distance oscillates, as illustrated in the upper panels of Figure 1. The two-vortex problem on the flat torus is therefore completely integrable, with stationary separations determined by

F(η∗)=0.(22)F(\eta_*) = 0. \tag*{(22)}

Writing η\eta in polar form η=reiθ\eta= r e^{i\theta}, Eq. (20) yields

r˙r=−ΓtotIm⁡F(η),θ˙=ΓtotRe⁡F(η).(23)\frac{\dot{r}}{r} = -\Gamma_{\mathrm{tot}} \operatorname{Im} F(\eta), \qquad\dot{\theta} = \Gamma_{\mathrm{tot}} \operatorname{Re} F(\eta). \tag*{(23)}

With

K(η,ρ)=KR(η,ρ)+iKI(η,ρ),(24)K(\eta,\sqrt{\rho}) = K_R(\eta,\sqrt{\rho}) + i K_I(\eta,\sqrt{\rho}), \tag*{(24)}

this becomes

r˙r=−Γtot2πKI(η,ρ),(25)\frac{\dot{r}}{r} = -\frac{\Gamma_{\mathrm{tot}}}{2\pi} K_I(\eta,\sqrt{\rho}), \tag*{(25)}
θ˙=Γtot[12πKR(η,ρ)−14π+log⁡r2πlog⁡ρ].(26)\dot{\theta} = \Gamma_{\mathrm{tot}} \left[\frac{1}{2\pi}K_R(\eta,\sqrt{\rho}) - \frac{1}{4\pi} + \frac{\log r}{2\pi\log\rho}\right]. \tag*{(26)}

Thus the imaginary part of KK drives radial drift in the annulus, while the real part determines the phase evolution. Constant-rr motions satisfy KI(η,ρ)=0K_I(\eta,\sqrt{\rho}) = 0, in which case the binary rotates with angular velocity (in the η\eta variable)

Ωη=Γtot[12πKR(η,ρ)−14π+log⁡r2πlog⁡ρ].(27)\Omega_{\eta} = \Gamma_{\mathrm{tot}} \left[\frac{1}{2\pi}K_R(\eta,\sqrt{\rho}) - \frac{1}{4\pi} + \frac{\log r}{2\pi\log\rho}\right]. \tag*{(27)}

For equal like-signed vortices, Γ1=Γ2=γ\Gamma_1 = \Gamma_2 = \gamma, these reduce to

r˙r=−γπKI(η,ρ),(28)\frac{\dot{r}}{r} = -\frac{\gamma}{\pi}K_I(\eta,\sqrt{\rho}), \tag*{(28)}
Ωη=γπ[KR(η,ρ)−12+log⁡rlog⁡ρ].(29)\Omega_\eta= \frac{\gamma}{\pi}\left[K_R(\eta,\sqrt{\rho})-\frac{1}{2}+\frac{\log r}{\log\rho}\right]. \tag*{(29)}

Although θ\theta gives the natural annulus phase, the physically observed orbital frequency is more naturally defined in Euclidean coordinates. Writing

Δ=w1−w2=x12+iy12,θE=arg⁡(x12+iy12),(30)\Delta= w_1-w_2=x_{12}+iy_{12},\qquad\theta_E=\arg(x_{12}+iy_{12}), \tag*{(30)}

the Euclidean orbital frequency is

ΩE:=θ˙E=x12y˙12−y12x˙12x122+y122.(31)\Omega_E := \dot{\theta}_E = \frac{x_{12}\dot{y}_{12}-y_{12}\dot{x}_{12}}{x_{12}^{2}+y_{12}^{2}}. \tag*{(31)}

From (19),

x˙12=ΓtotFR(η),y˙12=−ΓtotFI(η),(32)\dot{x}_{12}=\Gamma_{\mathrm{tot}}F_R(\eta),\qquad\dot{y}_{12}=-\Gamma_{\mathrm{tot}}F_I(\eta), \tag*{(32)}

and therefore

ΩE=−Γtotx12FI(η)+y12FR(η)x122+y122.\Omega_E=-\Gamma_{\mathrm{tot}}\frac{x_{12}F_I(\eta)+y_{12}F_R(\eta)}{x_{12}^{2}+y_{12}^{2}}.

Using

FR=12πKR(η,ρ)−14π−y122πlog⁡ρ,FI=12πKI(η,ρ),(33)F_R=\frac{1}{2\pi}K_R(\eta,\sqrt{\rho})-\frac{1}{4\pi}-\frac{y_{12}}{2\pi\log\rho},\qquad F_I=\frac{1}{2\pi}K_I(\eta,\sqrt{\rho}), \tag*{(33)}

one obtains the exact Euclidean frequency formula

ΩE=−Γtot2πx12KI(η,ρ)+y12KR(η,ρ)−12y12−y122log⁡ρx122+y122.(34)\Omega_E=-\frac{\Gamma_{\mathrm{tot}}}{2\pi}\frac{x_{12}K_I(\eta,\sqrt{\rho})+y_{12}K_R(\eta,\sqrt{\rho})-\frac{1}{2}y_{12}-\frac{y_{12}^{2}}{\log\rho}}{x_{12}^{2}+y_{12}^{2}}. \tag*{(34)}

Equation (34) shows that the Euclidean orbital frequency is distinct from the annulus-phase frequency θ˙\dot{\theta}. For the representative equal-vortex orbit shown in Figure 2, the theoretical prediction from (34) is visually indistinguishable from the numerically extracted frequency, with residuals at the 10−710^{-7} level. The numerical decomposition reveals that the Euclidean orbital frequency is controlled by three contributions (orbit averaged),

Image not extracted

Figure 2. Theoretical vs numerical orbital frequency on the flat torus. Top: ω(t)\omega(t) (theory and numerics indistinguishable). Bottom: residual Δω∼10−7\Delta\omega\sim10^{-7}.

ΩE∝⟨x12KI+y12KR−12y12−y122log⁡ρx122+y122⟩(35)\Omega_E \propto\left\langle\frac{x_{12}K_{\mathrm{I}}+y_{12}K_{\mathrm{R}}-\frac{1}{2}y_{12}-\frac{y_{12}^{2}}{\log\rho}}{x_{12}^{2}+y_{12}^{2}} \right\rangle \tag*{(35)}

While the term proportional to y12KRy_{12}K_{\mathrm{R}} provides the dominant contribution, the correction arising from x12KIx_{12}K_{\mathrm{I}} is numerically significant, and the geometric term proportional to y122/log⁡ρy_{12}^{2}/\log\rho yields a smaller but non-negligible contribution. This demonstrates that the orbital frequency is not determined solely by the mean separation, but arises from correlated oscillations between the relative geometry and the interaction kernel. For the trajectories considered here, the fluctuation of the Euclidean radius is sufficiently weak that

⟨Ax122+y122⟩≈⟨A⟩⟨x122+y122⟩,(36)\left\langle\frac{A}{x_{12}^{2}+y_{12}^{2}} \right\rangle\approx\frac{\langle A\rangle}{\left\langle x_{12}^{2}+y_{12}^{2}\right\rangle}, \tag*{(36)}

where

A:=x12KI(η,ρ)+y12KR(η,ρ)−12y12−y122log⁡ρ.(37)A := x_{12}K_{\mathrm{I}}(\eta,\sqrt{\rho})+y_{12}K_{\mathrm{R}}(\eta,\sqrt{\rho})-\frac{1}{2}y_{12}-\frac{y_{12}^{2}}{\log\rho}. \tag*{(37)}

This yields the compact approximation

ΩE≈−Γtot⟨A⟩2π⟨x122+y122⟩(38)\Omega_E \approx-\frac{\Gamma_{\mathrm{tot}}\langle A\rangle}{2\pi\left\langle x_{12}^{2}+y_{12}^{2}\right\rangle} \tag*{(38)}

In the dipole case,

Γ1=γ,Γ2=−γ,Γtot=0,\Gamma_1=\gamma,\qquad\Gamma_2=-\gamma,\qquad\Gamma_{\mathrm{tot}}=0,

the relative coordinate is constant and both vortices move with the same velocity,

dw‾1dt=dw‾2dt=−γF(η).(39)\frac{d\overline{w}_1}{dt}=\frac{d\overline{w}_2}{dt}=-\gamma F(\eta). \tag*{(39)}

Writing dw‾/dt=x˙−iy˙d\overline{w}/dt=\dot{x}-i\dot{y}, one obtains

x˙=−γ[12πKR(η,ρ)−14π+log⁡∣η∣2πlog⁡ρ](40)\dot{x}=-\gamma\left[\frac{1}{2\pi}K_{\mathrm{R}}(\eta,\sqrt{\rho})-\frac{1}{4\pi}+\frac{\log|\eta|}{2\pi\log\rho}\right] \tag*{(40)}
y˙=γ2πKI(η,ρ).(41)\dot{y}=\frac{\gamma}{2\pi}K_{\mathrm{I}}(\eta,\sqrt{\rho}). \tag*{(41)}

Since η=ei(x12+iy12)\eta= e^{i(x_{12}+iy_{12})}, we have log⁡∣η∣=−y12\log|\eta| = -y_{12}, and the dipole velocity may be written as

x˙=−γ4π[2KR(η,ρ)−1−2y12log⁡ρ],(42)\dot{x} = -\frac{\gamma}{4\pi}\left[2K_R(\eta,\sqrt{\rho})-1-\frac{2y_{12}}{\log\rho}\right], \tag*{(42)}
y˙=γ2πKI(η,ρ).(43)\dot{y} = \frac{\gamma}{2\pi}K_I(\eta,\sqrt{\rho}). \tag*{(43)}

Thus the dipole speed on the flat torus is not determined solely by the separation magnitude, as in the planar problem, but also by the periodic image effects encoded in KRK_R and KIK_I, together with the explicit geometric correction proportional to y12/log⁡ρy_{12}/\log\rho. This sensitivity to global geometry is reflected in the rigid dipole trajectories shown in Figure 1. More generally, binaries with unequal strengths can be treated in exactly the same way: the dynamics again reduces to the single complex variable η=ν1/ν2\eta= \nu_1/\nu_2. Representative trajectories for unequal like-signed and opposite-signed pairs are shown in Figure 3.

Image not extracted

Figure 3. Motion of vortices of unequal strengths. Green, red, and black dots indicate the initial positions of vortices with circulations +1+1, +5+5, and −5-5, respectively; the same color scheme is used for the corresponding trajectories.

Same Sign Cluster Dynamics

Clusters of like-signed vortices on the flat torus exhibit a simple but nontrivial collective behavior. To leading order, the cluster undergoes a coherent rotation driven by the mutual pairwise interactions, while its overall size remains nearly constant. Superimposed on this rotation is a slower modulation of the cluster size, corresponding to a weak breathing mode induced by the compact geometry. These two effects—rotation and slow breathing—capture the dominant large-scale dynamics of tightly bound vortex clusters.

To quantify this collective behavior systematically, we derive a small-cluster expansion of the vortex motion on the flat torus by expanding the closed analytic form of the interaction kernel

K(ζ,ρ)=11−ζ+1log⁡ρ[ψρ(log⁡(1/ζ)log⁡ρ)−ψρ(log⁡(ζ)log⁡ρ)],(44)K(\zeta,\sqrt{\rho}) = \frac{1}{1-\zeta}+\frac{1}{\log\rho}\left[\psi_{\rho}\left(\frac{\log(1/\zeta)}{\log\rho}\right)-\psi_{\rho}\left(\frac{\log(\zeta)}{\log\rho}\right)\right], \tag*{(44)}

where ψρ\psi_{\rho} is the qq-digamma function. Introducing multiplicative and additive variables as before (repeated here for convenience)

ζ=e−z,z=iw,w=x+iy,\zeta= e^{-z},\qquad z=iw,\qquad w=x+iy,

it is convenient to expand in z=log⁡ζz = \log\zeta about coincidence (z→0z \to0). This yields

K(z)=1z+12+(112−2ψρ(1)(1)log⁡2ρ)z+O(z3),(45)K(z) = \frac{1}{z} + \frac{1}{2} + \left(\frac{1}{12} - \frac{2\psi_\rho^{(1)}(1)}{\log^2 \rho}\right)z + O(z^3), \tag*{(45)}

where ψρ(1)(1)=QPolyGamma⁡(1,1;ρ)\psi_\rho^{(1)}(1) = \operatorname{QPolyGamma}(1,1;\rho) (see Appendix § 11 for a detailed and rather subtle derivation). Passing to the physical coordinate z=iwz = iw, the kernel becomes

K(w)=−iw+12+(i12−2iψρ(1)(1)log⁡2ρ)w+O(w3).(46)K(w) = -\frac{i}{w} + \frac{1}{2} + \left(\frac{i}{12} - \frac{2i\psi_\rho^{(1)}(1)}{\log^2 \rho}\right)w + O(w^3). \tag*{(46)}

The interaction kernel entering the equations of motion is

F(w,wˉ)=12πK(w)−14π+i4πlog⁡ρ(w−wˉ),F(w,\bar{w}) = \frac{1}{2\pi}K(w) - \frac{1}{4\pi} + \frac{i}{4\pi\log\rho}(w-\bar{w}),

which is equivalent to the form

F(ζ)=12πK(ζ,ρ)−14π+12πlog⁡ρlog⁡∣ζ∣.(47)F(\zeta) = \frac{1}{2\pi}K(\zeta,\sqrt{\rho}) - \frac{1}{4\pi} + \frac{1}{2\pi\log\rho}\log|\zeta|. \tag*{(47)}

Substituting the local expansion gives

F(w,wˉ)=−i2πw+A(ρ)w+B(ρ)wˉ+⋯ ,(48)F(w,\bar{w}) = -\frac{i}{2\pi w} + A(\rho)w + B(\rho)\bar{w} + \cdots, \tag*{(48)}

with

A(ρ)=i24πlog⁡2ρ[log⁡ρ(6+log⁡ρ)−24ψρ(1)(1)],B(ρ)=−i4πlog⁡ρ.(49)A(\rho) = \frac{i}{24\pi\log^2 \rho}\left[\log\rho(6+\log\rho)-24\psi_\rho^{(1)}(1)\right], \qquad B(\rho) = -\frac{i}{4\pi\log\rho}. \tag*{(49)}

The leading term recovers the planar interaction,

F(w,wˉ)∼−i2πw,w→0,(50)F(w,\bar{w}) \sim-\frac{i}{2\pi w}, \qquad w \to0, \tag*{(50)}

while the coefficients A(ρ)A(\rho) and B(ρ)B(\rho) encode the torus geometry. We now consider a compact cluster of NN vortices of equal circulation Γ\Gamma and introduce internal coordinates

wj=R+ξj,∑j=1Nξj=0.w_j = R + \xi_j,\qquad\sum_{j=1}^{N} \xi_j = 0.

Using the conservation law R˙=0\dot{R}=0 and substitution into the equations of motion yields

ξ˙j=∑k≠jΓ[−i2π(ξj−ξk)+A(ρ)(ξj−ξk)+B(ρ)(ξˉj−ξˉk)]=−iΓ2π∑k≠j1ξˉj−ξˉk+NΓA(ρ) ξj+NΓB(ρ) ξˉj.(51)\begin{aligned} \dot{\xi}_j &= \sum_{k\ne j} \Gamma\left[-\frac{i}{2\pi(\xi_j-\xi_k)} + A(\rho)(\xi_j-\xi_k) + B(\rho)(\bar{\xi}_j-\bar{\xi}_k)\right] \\ &= -\frac{i\Gamma}{2\pi}\sum_{k\ne j}\frac{1}{\bar{\xi}_j-\bar{\xi}_k} + N\Gamma A(\rho)\,\xi_j + N\Gamma B(\rho)\,\bar{\xi}_j. \tag*{(51)} \end{aligned}

This expression shows that the dynamics of a compact same-sign vortex cluster decomposes into a universal planar interaction together with geometry-induced corrections determined by the torus modulus ρ\rho. The B(ρ)B(\rho) term contributes an isotropic component to the collective motion, while the A(ρ)A(\rho) term generates an anisotropic deformation that couples directly to the cluster shape.

Coarse-Grained Angular Velocity of a Clustered Configuration

We can now extract a collective angular velocity of the vortex cluster directly from the reduced dynamics without assuming any cluster symmetry. Defining

I:=∑j=1N∣ξj∣2,Q:=∑j=1Nξj2,(52)I := \sum_{j=1}^{N} |\xi_j|^2,\qquad Q := \sum_{j=1}^{N} \xi_j^2, \tag*{(52)}

we now proceed to calculate the rotation rate

Ω(t)=∑j=1NIm⁡(ξˉjξ˙j)I=−Im⁡(∑j=1Nξjξˉ˙j)I.(53)\Omega(t) = \frac{\sum_{j=1}^{N}\operatorname{Im}(\bar{\xi}_j\dot{\xi}_j)}{I} = -\frac{\operatorname{Im}\left(\sum_{j=1}^{N}\xi_j\dot{\bar{\xi}}_j\right)}{I}. \tag*{(53)}

Using the reduced dynamics of the previous section (presented here for clarity),

ξˉ˙j=−iΓ2π∑k≠j1ξˉj−ξˉk+NΓA(ρ)ξj+NΓB(ρ)ξˉj,(54)\dot{\bar{\xi}}_j=-\frac{\mathrm{i}\Gamma}{2\pi}\sum_{k\ne j}\frac{1}{\bar{\xi}_j-\bar{\xi}_k}+N\Gamma A(\rho)\xi_j+N\Gamma B(\rho)\bar{\xi}_j, \tag*{(54)}

we obtain

∑j=1Nξˉ˙jξj=−iΓ2π∑j=1N∑k≠jξˉjξˉj−ξˉk+NΓA(ρ)Q+NΓB(ρ)I.(55)\sum_{j=1}^{N}\dot{\bar{\xi}}_j\xi_j=-\frac{\mathrm{i}\Gamma}{2\pi}\sum_{j=1}^{N}\sum_{k\ne j}\frac{\bar{\xi}_j}{\bar{\xi}_j-\bar{\xi}_k}+N\Gamma A(\rho)Q+N\Gamma B(\rho)I. \tag*{(55)}

The double sum evaluates to 12N(N−1)\frac{1}{2}N(N-1), yielding

Ω(t)=ΓN(N−1)4πI−Im⁡(NΓA(ρ)Q)I−Im⁡(NΓB(ρ)I)I.(56)\Omega(t)=\frac{\Gamma N(N-1)}{4\pi I}-\frac{\operatorname{Im}(N\Gamma A(\rho)Q)}{I}-\frac{\operatorname{Im}(N\Gamma B(\rho)I)}{I}. \tag*{(56)}

Writing R2=I/NR^2=I/N, B(ρ)=−i/(4πlog⁡ρ)B(\rho)=-\mathrm{i}/(4\pi\log\rho), and A(ρ)=iAI(ρ)A(\rho)=\mathrm{i}A_I(\rho), this reduces to

Ω(t)=Γ(N−1)4πR2+NΓ4πlog⁡ρ−NΓAI(ρ)Re⁡(Q)I,(57)\Omega(t)=\frac{\Gamma(N-1)}{4\pi R^2}+\frac{N\Gamma}{4\pi\log\rho}-\frac{N\Gamma A_I(\rho)\operatorname{Re}(Q)}{I}, \tag*{(57)}

where

AI(ρ)=124πlog⁡2ρ[log⁡ρ(6+log⁡ρ)−24ψρ(1)(1)].(58)A_I(\rho)=\frac{1}{24\pi\log^2\rho}\left[\log\rho(6+\log\rho)-24\psi_\rho^{(1)}(1)\right]. \tag*{(58)}

Thus the collective angular velocity decomposes into a universal planar term, an isotropic torus-induced shift, and an anisotropic correction governed by the quadrupole moment. In the nearly isotropic cluster Q≈0Q \approx0, one obtains

Ω(t)≈Γ(N−1)4πR2+NΓ4πlog⁡ρ.(59)\Omega(t) \approx\frac{\Gamma(N-1)}{4\pi R^2} + \frac{N\Gamma}{4\pi\log\rho}. \tag*{(59)}

The leading term in the coarse-grained angular velocity plays the role of a many-body analogue of the rigid two-rotor frequency in an unbounded 3D fluid (see Appendix § 12 and Ref. [11] for a typical soft matter setup), with the cluster radius replacing the pair separation and the effective strength set by Γ(N−1)\Gamma(N-1). Notably, the scaling differs: Ω∼R−2\Omega\sim R^{-2} in the planar vortex system, as opposed to Ω∼D−3\Omega\sim D^{-3} for 3D rotors. The additional torus-induced shift and quadrupolar correction are specific to the periodic geometry and to the internal deformability of the cluster. We compare the coarse-grained prediction for the angular velocity with direct numerical simulations of the full vortex motion on the flat torus. The simulations are initialized with N=50N=50 equal-circulation vortices randomly distributed within a compact disk in the fundamental cell of a rectangular torus with ρ=e−4π\rho=e^{-4\pi}, ensuring a tightly clustered, anisotropic configuration with Q(0)≠0Q(0)\ne0. The Hamiltonian variation satisfies ΔH≲10−4\Delta H\lesssim10^{-4} over the duration of the simulation. The theoretically derived formula

Ωcg(t)=Γ(N−1)4πR2(t)+NΓ4πlog⁡ρ−NΓAI(ρ)Re⁡(Q(t))I(t),(60)\Omega_{\rm cg}(t)=\frac{\Gamma(N-1)}{4\pi R^2(t)}+\frac{N\Gamma}{4\pi\log\rho}-\frac{N\Gamma A_I(\rho)\operatorname{Re}(Q(t))}{I(t)}, \tag*{(60)}

with I(t)=∑j∣ξj∣2I(t)=\sum_j|\xi_j|^2, R2(t)=I(t)/NR^2(t)=I(t)/N, and Q(t)=∑jξj2Q(t)=\sum_j\xi_j^2 can be compared with a numerically extracted solution

Ωsim(t)=∑j=1NIm⁡(ξˉjξ˙j)∑j=1N∣ξj∣2,(61)\Omega_{\rm sim}(t)=\frac{\sum_{j=1}^{N}\operatorname{Im}(\bar{\xi}_j\dot{\xi}_j)}{\sum_{j=1}^{N}|\xi_j|^2}, \tag*{(61)}

and compare it with the full Ωcg(t)\Omega_{\rm cg}(t) and the isotropic approximation

Ωiso(t)=Γ(N−1)4πR2(t)+NΓ4πlog⁡ρ.(62)\Omega_{\rm iso}(t)=\frac{\Gamma(N-1)}{4\pi R^2(t)}+\frac{N\Gamma}{4\pi\log\rho}. \tag*{(62)}

As shown in Figure 5, Ωcg(t)\Omega_{\rm cg}(t) is in near-perfect agreement with Ωsim(t)\Omega_{\rm sim}(t) over the full evolution, while Ωiso(t)\Omega_{\rm iso}(t) exhibits a clear deviation. The residual Ωsim−Ωcg\Omega_{\rm sim}-\Omega_{\rm cg} remains at the level of 10−310^{-3}, whereas the deviation from the isotropic prediction is an order of magnitude larger, demonstrating that the quadrupolar correction is essential for quantitative accuracy. A physically motivated way to decompose the rotation rate is

Image not extracted

Figure 5. Comparison between numerical simulations and coarse-grained theory for a compact vortex cluster on the flat torus (ρ=e−4π\rho= e^{-4\pi}; N=50N = 50 equal-circulation vortices randomly initialized in a disk). Top left: Ωsim(t)\Omega_{\mathrm{sim}}(t) vs Ωtheory(t)\Omega_{\mathrm{theory}}(t) and Ωiso(t)\Omega_{\mathrm{iso}}(t). Top right: residuals. Middle left: decomposition into pair, torus, and quadrupolar contributions. Middle right: Ωsim−Ωiso\Omega_{\mathrm{sim}}-\Omega_{\mathrm{iso}} vs ΩA(t)\Omega_A(t). Bottom: difference, showing that the quadrupolar term captures the deviation from isotropic theory.

Ω(t)=Ωpair(t)+ΩB+ΩA(t),(63)\Omega(t)=\Omega_{\rm pair}(t)+\Omega_B+\Omega_A(t), \tag*{(63)}

with

Ωpair(t)=Γ(N−1)4πR2(t),ΩB=NΓ4πlog⁡ρ,ΩA(t)=−NΓAI(ρ)Re⁡(Q(t))I(t).(64)\Omega_{\mathrm{pair}}(t) = \frac{\Gamma(N-1)}{4\pi R^2(t)}, \qquad\Omega_B = \frac{N\Gamma}{4\pi\log\rho}, \qquad\Omega_A(t) = -\frac{N\Gamma A_I(\rho)\operatorname{Re}(Q(t))}{I(t)}. \tag*{(64)}

In Figure 5, we find that Ωpair\Omega_{\mathrm{pair}} provides the dominant contribution, while ΩB\Omega_B produces a constant geometric offset. The remaining time-dependent modulation is entirely captured by ΩA(t)\Omega_A(t), which accurately reproduces the difference Ωsim−Ωiso\Omega_{\mathrm{sim}}-\Omega_{\mathrm{iso}}. The residual

(Ωsim−Ωiso)−ΩA(65)(\Omega_{\mathrm{sim}}-\Omega_{\mathrm{iso}})-\Omega_A \tag*{(65)}

remains small throughout the evolution, confirming that higher-order corrections are negligible in the compact-cluster dynamics.

The size evolution is independently validated in Figure 4, where the numerical results for dR2/dtdR^2/dt and R2(t)R^2(t) agree closely with the analytic prediction −2ΓAI(ρ)Im⁡(Q)-2\Gamma A_I(\rho)\operatorname{Im}(Q) (derived in next section). These results demonstrate that the leading-order theory accurately captures both the collective rotation and the weak breathing of the cluster, with the quadrupole moment Q(t)Q(t) providing the key dynamical control of deviations from isotropic motion.

Image not extracted

Figure 4. Numerical test of size-evolution for a compact vortex cluster on the flat torus (ρ=e−4π\rho= e^{-4\pi}; N=50N = 50 equal-circulation vortices randomly initialized within a disk in the fundamental cell). From left to right: real-space trajectories showing that the configuration remains compact; comparison of the simulated dR2/dtdR^2/dt with the theoretical prediction −2ΓAI(ρ)Im⁡Q(t)-2\Gamma A_I(\rho)\operatorname{Im}Q(t); and comparison of the simulated R2(t)R^2(t) with the integrated theoretical prediction. The close agreement in the last two panels confirms that the weak breathing of the cluster is accurately governed by the imaginary part of the quadrupole moment. The Hamiltonian variation satisfies ΔH≲10−4\Delta H \lesssim10^{-4} over the duration of the simulation.

Analytic evolution of the cluster size

The reduced dynamics of the previous section also implies that the cluster size is tied to the quadrupole moment. Defining

I(t):=∑j=1N∣ξj∣2,R2(t)=I(t)N,I(t) := \sum_{j=1}^{N} |\xi_j|^2, \qquad R^2(t) = \frac{I(t)}{N},

we differentiate to obtain

I˙=2Re⁡(∑j=1Nξjξˉ˙j).\dot{I} = 2\operatorname{Re}\left(\sum_{j=1}^{N}\xi_j\dot{\bar{\xi}}_j\right).

Using the reduced equation of motion,

ξˉ˙j=−iΓ2π∑k≠j1ξˉj−ξˉk+NΓA(ρ)ξj+NΓB(ρ)ξˉj,\dot{\bar{\xi}}_j=-\frac{i\Gamma}{2\pi}\sum_{k\ne j}\frac{1}{\bar{\xi}_j-\bar{\xi}_k}+N\Gamma A(\rho)\xi_j+N\Gamma B(\rho)\bar{\xi}_j,

we find

∑j=1Nξjξˉ˙j=−iΓ2π∑j=1N∑k≠jξjξj−ξk+NΓA(ρ)Q+NΓB(ρ)I,(66)\sum_{j=1}^{N}\xi_j\dot{\bar{\xi}}_j=-\frac{i\Gamma}{2\pi}\sum_{j=1}^{N}\sum_{k\ne j}\frac{\xi_j}{\xi_j-\xi_k}+N\Gamma A(\rho)Q+N\Gamma B(\rho)I, \tag*{(66)}

where Q=∑j=1Nξj2Q=\sum_{j=1}^{N}\xi_j^2. The double sum evaluates to 12N(N−1)\frac{1}{2}N(N-1), so that the pair contribution is purely imaginary and does not affect I˙\dot{I}. The term proportional to B(ρ)=−i/(4πlog⁡ρ)B(\rho)=-i/(4\pi\log\rho) is also purely imaginary. The only contribution to I˙\dot{I} therefore arises from the term involving A(ρ)=iAI(ρ)A(\rho)=iA_I(\rho), giving

I˙=−2NΓAI(ρ)Im⁡(Q).(67)\dot{I}=-2N\Gamma A_I(\rho)\operatorname{Im}(Q). \tag*{(67)}

It follows that

dR2dt=−2ΓAI(ρ)Im⁡(Q).(68)\frac{dR^2}{dt}=-2\Gamma A_I(\rho)\operatorname{Im}(Q). \tag*{(68)}

Thus the cluster size evolves solely through the imaginary part of the quadrupole moment. In particular, when Im⁡(Q)\operatorname{Im}(Q) is small or oscillatory with small mean, the cluster exhibits only weak breathing and remains approximately of constant size. Together with the angular-velocity law, this shows that the same complex quadrupole governs both collective rotation and size modulation, with Re⁡(Q)\operatorname{Re}(Q) controlling rotation and Im⁡(Q)\operatorname{Im}(Q) controlling the breathing dynamics. To quantitatively validate the size evolution law (68), we compare its predictions with direct numerical simulations of vortex motion on the flat torus. The simulations are performed for a compact cluster of N=50N=50 vortices of equal circulation, initialized randomly within a small disk inside the fundamental domain, with modulus ρ=e−4π\rho=e^{-4\pi}. The initial configuration is thus strongly localized, ensuring that the small-cluster expansion underlying the reduced dynamics is applicable throughout the evolution.

The results are summarized in Figure 4. The leftmost panel shows the real-space trajectories, confirming that the vortices remain tightly clustered and do not disperse across the torus. The Hamiltonian variation satisfies ΔH≲10−4\Delta H\lesssim10^{-4} over the duration of the simulation. The third panel provides a direct comparison between the numerical time derivative dR2/dtdR^2/dt and the theoretical prediction −2ΓAI(ρ)Im⁡Q(t)-2\Gamma A_I(\rho)\operatorname{Im} Q(t). The agreement is excellent: the two curves coincide in phase, amplitude, and turning points, demonstrating that the instantaneous rate of change of the cluster size is correctly captured by the reduced quadrupole description.

An even more stringent test is obtained by integrating (68) in time. The rightmost panel compares the measured R2(t)R^2(t) with the reconstructed theoretical prediction

Rtheory2(t)=R2(t0)−2ΓAI(ρ)∫t0tIm⁡Q(s) ds,R^2_{\mathrm{theory}}(t) = R^2(t_0) - 2\Gamma A_I(\rho) \int_{t_0}^{t} \operatorname{Im} Q(s)\,\mathrm{d}s,

and again shows near-perfect agreement over the full evolution. This confirms that not only the instantaneous dynamics, but also the cumulative effect of the quadrupolar term is accurately described by the theory.

Together, these results show that the weak temporal variation of the cluster size is governed by the imaginary part of the quadrupole moment, consistent with (68). Combined with the angular-velocity relation, this indicates that the leading collective dynamics of compact vortex clusters is encoded in the complex quadrupole Q(t)Q(t): its real part controls deviations from the shape-independent collective rotation rate, while its imaginary part governs the slow breathing of the cluster.

Conclusion

We have investigated an exact Hamiltonian formulation of point-vortex motion on the flat torus, governed by a closed interaction kernel that incorporates both the singular planar interaction and global geometric effects. The antisymmetry property of the kernel under inversion underlies the conservation laws and enables reductions of the dynamics. In particular, the two-vortex problem is completely integrable, with a clear distinction between rigid dipole motion and nontrivial chiral dynamics.

A local expansion of the kernel yields a reduced description of compact same-sign vortex clusters, in which the dynamics separates into universal planar, isotropic torus, and anisotropic contributions. This leads to a coarse-grained formulation in terms of collective variables, where the evolution is governed by the second moments of the configuration. The leading correction to the rotation rate is controlled by the real part of the quadrupole moment, while the slow evolution of the cluster size is governed by its imaginary part, identifying the complex quadrupole as the key dynamical quantity encoding deviations from isotropic motion.

Numerical simulations confirm these predictions. For compact clusters, the theoretical angular velocity and size evolution laws accurately reproduce both the mean behavior and the subleading corrections, demonstrating that anisotropy provides the dominant deviation from isotropic dynamics.

These results establish a unified framework linking exact Hamiltonian structure, reduced dynamics, and emergent collective behavior for vortices on the flat torus and periodic fluid domains in general. The formalism presented here will be of interest to the growing body of work on vortex and rotor clusters in superfluid and soft matter systems [2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 16] [1]. The formulation naturally suggests extensions to dissipative dynamics, interacting clusters, and more general compact geometries, and provides a pathway toward continuum and kinetic descriptions of vortex matter [5, 9] in compact domains.

Acknowledgments

We are very thankful to Takashi Sakajo, Suryateja Gavva, Naomi Oppenheimer and Haim Diamant. R.S is supported by DST INSPIRE Faculty fellowship, India (Grant No.IFA19-PH231). Both authors acknowledge support from NFSG and OPERA Research Grant from Birla Institute of Technology and Science, Pilani (Hyderabad Campus).

Data Availability

The data that support the findings of this study are available within the article. The computational codes used in this work are available from the authors upon reasonable request.

Recap of the Schottky-Klein machinery

The Schottky-Klein prime function is a special function defined on multiply-connected circular domains [22]. For the annulus

Dζ={ζ∈C∣ρ<∣ζ∣<1},D_{\zeta} = \{\zeta\in\mathbb{C} \mid\rho< |\zeta| < 1\},

the prime function is given by the infinite product

P(ζ,ρ)=(1−ζ)∏k=1∞(1−ρkζ)(1−ρk/ζ).(69)P(\zeta,\sqrt{\rho}) = (1-\zeta)\prod_{k=1}^{\infty}(1-\rho^{k}\zeta)(1-\rho^{k}/\zeta). \tag*{(69)}

Note that P(ζ,ρ)P(\zeta,\sqrt{\rho}) has a simple zero at ζ=1\zeta= 1 in DζD_{\zeta}. The associated KK-function, defined as the logarithmic derivative of P(ζ,ρ)P(\zeta,\sqrt{\rho}), is

K(ζ,ρ)=ζP′(ζ,ρ)P(ζ,ρ),(70)K(\zeta,\sqrt{\rho}) = \frac{\zeta P'(\zeta,\sqrt{\rho})}{P(\zeta,\sqrt{\rho})}, \tag*{(70)}

where the prime denotes differentiation with respect to the first argument, i.e. P′(ζ,ρ)=dP(ζ,ρ)dζP'(\zeta,\sqrt{\rho}) = \frac{dP(\zeta,\sqrt{\rho})}{d\zeta}. A closed analytic form for the above has been provided in [26]:

K(ζ,ρ)=11−ζ+1log⁡ρ[ψρ(log⁡(1/ζ)log⁡ρ)−ψρ(log⁡(ζ)log⁡ρ)],(71)K(\zeta,\sqrt{\rho}) = \frac{1}{1-\zeta} + \frac{1}{\log\rho}\left[\psi_{\rho}\left(\frac{\log(1/\zeta)}{\log\rho}\right)-\psi_{\rho}\left(\frac{\log(\zeta)}{\log\rho}\right)\right], \tag*{(71)}

where ψρ(z)\psi_{\rho}(z) denotes the qq-digamma function with base q=ρq=\rho, defined by the logarithmic derivative of the qq-gamma function Γρ(z)\Gamma_{\rho}(z),

ψρ(z)=ddzlog⁡Γρ(z),Γρ(z)=(1−ρ)1−z∏n=0∞1−ρn+11−ρn+z,(72)\psi_{\rho}(z) = \frac{d}{dz}\log\Gamma_{\rho}(z), \qquad\Gamma_{\rho}(z) = (1-\rho)^{1-z}\prod_{n=0}^{\infty}\frac{1-\rho^{n+1}}{1-\rho^{n+z}}, \tag*{(72)}

which converges for $0 < \rho < 1$ and all complex zz away from its poles. In this product representation, each factor in the denominator (1−ρn+z)(1-\rho^{n+z}) corresponds to the lattice of poles of Γρ(z)\Gamma_{\rho}(z). The poles of Γρ(z)\Gamma_{\rho}(z) arise from the zeros of the denominator term $(1-\rho^{n+z}) = 0, which yield

ρn+z=1⇒(n+z)ln⁡ρ=2πik,k∈Z.(73)\rho^{n+z}=1 \Rightarrow(n+z)\ln\rho=2\pi i k,\qquad k\in\mathbb{Z}. \tag*{(73)}

Thus, the poles are located at

zn,k=−n+2πikln⁡ρ,n,k∈Z.(74)z_{n,k}=-n+\frac{2\pi i k}{\ln\rho},\qquad n,k\in\mathbb{Z}. \tag*{(74)}

These poles lie on a rectangular lattice in the complex zz–plane, with unit spacing along the real direction and vertical spacing 2π/∣ln⁡ρ∣2\pi/\lvert\ln\rho\rvert.

Local expansion of the kernel K(z)K(z)

We derive the small-zz expansion of the kernel

K(z)=11−ez+1log⁡ρ[ψρ(−zlog⁡ρ)−ψρ(zlog⁡ρ)],0<ρ<1.(75)K(z)=\frac{1}{1-e^z}+\frac{1}{\log\rho}\left[\psi_\rho\left(-\frac{z}{\log\rho}\right)-\psi_\rho\left(\frac{z}{\log\rho}\right)\right],\qquad0<\rho<1. \tag*{(75)}

Writing L=log⁡ρL=\log\rho and x=z/Lx=z/L, we expand each contribution about z=0z=0.

The elementary term gives

11−ez=−1z+12−z12+O(z3).(76)\frac{1}{1-e^z}=-\frac{1}{z}+\frac{1}{2}-\frac{z}{12}+O(z^3). \tag*{(76)}

For the qq-digamma part, we use the shift identity

ψρ(u+1)−ψρ(u)=−(log⁡ρ)ρu1−ρu,(77)\psi_\rho(u+1)-\psi_\rho(u)=-\frac{(\log\rho)\rho^u}{1-\rho^u}, \tag*{(77)}

which yields

ψρ(−x)−ψρ(x)=[ψρ(1−x)−ψρ(1+x)]+L[e−Lx1−e−Lx−eLx1−eLx].(78)\psi_\rho(-x)-\psi_\rho(x)=\left[\psi_\rho(1-x)-\psi_\rho(1+x)\right]+L\left[\frac{e^{-Lx}}{1-e^{-Lx}}-\frac{e^{Lx}}{1-e^{Lx}}\right]. \tag*{(78)}

Expanding for small xx gives

ψρ(1−x)−ψρ(1+x)=−2ψρ(1)(1)x+O(x3),\psi_\rho(1-x)-\psi_\rho(1+x)=-2\psi_\rho^{(1)}(1)x+O(x^3),

and

L[e−Lx1−e−Lx−eLx1−eLx]=2x+L26x+O(x3).L\left[\frac{e^{-Lx}}{1-e^{-Lx}}-\frac{e^{Lx}}{1-e^{Lx}}\right]=\frac{2}{x}+\frac{L^2}{6}x+O(x^3).

Combining these results and using x=z/Lx=z/L, we obtain

1L[ψρ(−x)−ψρ(x)]=2z+(16−2ψρ(1)(1)log⁡2ρ)z+O(z3).(79)\frac{1}{L}\left[\psi_{\rho}(-x)-\psi_{\rho}(x)\right]=\frac{2}{z}+\left(\frac{1}{6}-\frac{2\psi_{\rho}^{(1)}(1)}{\log^2\rho}\right)z+O(z^3). \tag*{(79)}

Adding the two contributions yields

K(z)=1z+12+(112−2ψρ(1)(1)log⁡2ρ)z+O(z3),(80)K(z)=\frac{1}{z}+\frac{1}{2}+\left(\frac{1}{12}-\frac{2\psi_{\rho}^{(1)}(1)}{\log^2\rho}\right)z+O(z^3), \tag*{(80)}

which is the local expansion quoted in the main text.

Two rotors in an infinite 3D fluid

We consider two rotors aligned with z^\hat{z}, with strengths Γ1\Gamma_1 and Γ2\Gamma_2 and positions Xi=(xi,yi,zi)\mathbf{X}_i=(x_i,y_i,z_i). The induced velocity is

U(X,Xj)=Γjz^×X−Xj∥X−Xj∥3,(81)\mathbf{U}(\mathbf{X},\mathbf{X}_j)=\Gamma_j\hat{z}\times\frac{\mathbf{X}-\mathbf{X}_j}{\lVert\mathbf{X}-\mathbf{X}_j\rVert^3}, \tag*{(81)}

leading to

X˙1=Γ2z^×X1−X2R3,X˙2=Γ1z^×X2−X1R3,(82)\dot{\mathbf{X}}_1=\Gamma_2\hat{z}\times\frac{\mathbf{X}_1-\mathbf{X}_2}{R^3},\qquad\dot{\mathbf{X}}_2=\Gamma_1\hat{z}\times\frac{\mathbf{X}_2-\mathbf{X}_1}{R^3}, \tag*{(82)}

with R=∥X1−X2∥R=\lVert\mathbf{X}_1-\mathbf{X}_2\rVert. The cross-product structure implies z˙1=z˙2=0\dot{z}_1=\dot{z}_2=0, so the motion is planar.

Introducing complex coordinates zi=xi+iyiz_i=x_i+\mathrm{i}y_i and w=z1−z2w=z_1-z_2, we obtain

w˙=i(Γ1+Γ2)w∣w∣3.(83)\dot{w}=\mathrm{i}(\Gamma_1+\Gamma_2)\frac{w}{\lvert w\rvert^3}. \tag*{(83)}

Hence ∣w∣=D|w| = D is constant, so the pair is rigid, and

ddt(Γ1z1+Γ2z2)=0.(84)\frac{d}{dt}\left(\Gamma_1 z_1+\Gamma_2 z_2\right)=0. \tag*{(84)}

For Γ1+Γ2≠0\Gamma_1+\Gamma_2\ne0, the conserved center of vorticity

C=Γ1z1+Γ2z2Γ1+Γ2(85)C=\frac{\Gamma_1 z_1+\Gamma_2 z_2}{\Gamma_1+\Gamma_2} \tag*{(85)}

is fixed, and the solution is

w(t)=DeiΩt,Ω=Γ1+Γ2D3.(86)w(t)=D e^{i\Omega t},\qquad\Omega=\frac{\Gamma_1+\Gamma_2}{D^3}. \tag*{(86)}

Writing

z1=C+R1eiΩt,z2=C−R2eiΩt,(87)z_1=C+R_1e^{i\Omega t},\qquad z_2=C-R_2e^{i\Omega t}, \tag*{(87)}

the radii satisfy

R1=Γ2Γ1+Γ2D,R2=Γ1Γ1+Γ2D.(88)R_1=\frac{\Gamma_2}{\Gamma_1+\Gamma_2}D,\qquad R_2=\frac{\Gamma_1}{\Gamma_1+\Gamma_2}D. \tag*{(88)}

Thus unequal strengths lead to circular motion about CC with unequal radii.

For Γ1=Γ2=Γ\Gamma_1=\Gamma_2=\Gamma, one recovers symmetric rotation with R1=R2=D/2R_1=R_2=D/2 and Ω=2Γ/D3\Omega=2\Gamma/D^3.

For Γ1+Γ2=0\Gamma_1+\Gamma_2=0, one finds w˙=0\dot{w}=0 and

z˙1=z˙2=−iΓw0D3,(89)\dot{z}_1=\dot{z}_2=-i\Gamma\frac{w_0}{D^3}, \tag*{(89)}

so the pair translates rigidly without rotation.

Infinite-plane limit of the orbital frequency

In this appendix we verify that the exact expression for the Euclidean orbital angular frequency of a vortex binary on the flat torus reduces, in the infinite-plane limit, to the classical result for point vortices in an unbounded domain. We begin from the exact formula derived in the main text for the orbital angular frequency in Euclidean coordinates,

ΩE=−Γtot2πx12KI(η,ρ)+y12KR(η,ρ)−12y12−y122log⁡ρx122+y122(90)\Omega_E=-\frac{\Gamma_{\mathrm{tot}}}{2\pi}\frac{x_{12}K_I(\eta,\sqrt{\rho})+y_{12}K_R(\eta,\sqrt{\rho})-\frac{1}{2}y_{12}-\frac{y_{12}^{2}}{\log\rho}}{x_{12}^{2}+y_{12}^{2}} \tag*{(90)}

where Δ=w1−w2=x12+iy12\Delta=w_1-w_2=x_{12}+iy_{12}, r2=x122+y122r^2=x_{12}^2+y_{12}^2, and K=KR+iKIK=K_R+iK_I denotes the interaction kernel. We now consider the infinite-plane limit, corresponding to ρ→0\rho\to0, so that log⁡ρ→−∞\log\rho\to-\infty. In this limit, the explicit geometric contribution vanishes,

y122log⁡ρ→0,(91)\frac{y_{12}^{2}}{\log\rho}\to0, \tag*{(91)}

and the orbital frequency reduces to

ΩE=−Γtot2πx12KI+y12KR−12y12r2.(92)\Omega_E=-\frac{\Gamma_{\mathrm{tot}}}{2\pi}\frac{x_{12}K_I+y_{12}K_R-\frac{1}{2}y_{12}}{r^2}. \tag*{(92)}

To evaluate this expression, we use the local expansion of the kernel for small relative separation, as derived in Appendix B of the main text. Writing w=x12+iy12w=x_{12}+iy_{12}, the kernel admits the expansion

K(w)=−iw+12+O(w),(93)K(w)=-\frac{i}{w}+\frac{1}{2}+O(w), \tag*{(93)}

where the higher-order terms encode torus corrections that vanish in the infinite-plane limit. We now compute the real and imaginary parts of the leading-order contribution. Using

1w=x12−iy12r2,(94)\frac{1}{w}=\frac{x_{12}-iy_{12}}{r^2}, \tag*{(94)}

we obtain

−iw=−i(x12−iy12)r2=−y12−ix12r2.(95)-\frac{i}{w}=-\frac{i(x_{12}-iy_{12})}{r^2}=\frac{-y_{12}-ix_{12}}{r^2}. \tag*{(95)}

Thus,

KR=12−y12r2,KI=−x12r2.(96)K_R=\frac{1}{2}-\frac{y_{12}}{r^2},\qquad K_I=-\frac{x_{12}}{r^2}. \tag*{(96)}

Substituting into the numerator of (D3), we find

x12KI+y12KR−12y12=x12(−x12r2)+y12(12−y12r2)−12y12=−x122r2+12y12−y122r2−12y12.(97)x_{12}K_I+y_{12}K_R-\frac{1}{2}y_{12}=x_{12}\left(-\frac{x_{12}}{r^2}\right)+y_{12}\left(\frac{1}{2}-\frac{y_{12}}{r^2}\right)-\frac{1}{2}y_{12} = -\frac{x_{12}^2}{r^2}+\frac{1}{2}y_{12}-\frac{y_{12}^2}{r^2}-\frac{1}{2}y_{12}. \tag*{(97)}

The linear terms in y12y_{12} cancel exactly, yielding

x12KI+y12KR−12y12=−x122+y122r2=−1.(98)x_{12}K_I+y_{12}K_R-\frac{1}{2}y_{12}=-\frac{x_{12}^2+y_{12}^2}{r^2}=-1. \tag*{(98)}

Substituting back into (D3), we obtain

ΩE=(−Γtot2π)(−1r2)=Γtot2πr2.(99)\Omega_E=\left(-\frac{\Gamma_{\mathrm{tot}}}{2\pi}\right)\left(-\frac{1}{r^2}\right)=\frac{\Gamma_{\mathrm{tot}}}{2\pi r^2}. \tag*{(99)}

To make contact with the classical planar result, we now derive the orbital frequency directly from the Biot–Savart law in the infinite plane. Consider two point vortices of strengths Γ1\Gamma_1 and Γ2\Gamma_2 located at positions r1\mathbf{r}_1 and r2\mathbf{r}_2 in an unbounded two-dimensional incompressible fluid. The velocity induced at r1\mathbf{r}_1 by the vortex at r2\mathbf{r}_2 is given by the planar Biot–Savart law,

r˙1=Γ22πz^×(r1−r2)∣r1−r2∣2,r˙2=Γ12πz^×(r2−r1)∣r1−r2∣2.(100)\dot{\mathbf{r}}_1=\frac{\Gamma_2}{2\pi}\frac{\hat{\mathbf{z}}\times(\mathbf{r}_1-\mathbf{r}_2)}{\lvert\mathbf{r}_1-\mathbf{r}_2\rvert^2},\qquad \dot{\mathbf{r}}_2=\frac{\Gamma_1}{2\pi}\frac{\hat{\mathbf{z}}\times(\mathbf{r}_2-\mathbf{r}_1)}{\lvert\mathbf{r}_1-\mathbf{r}_2\rvert^2}. \tag*{(100)}

Introducing the relative coordinate

r:=r1−r2,r=∣r∣,\mathbf{r}:=\mathbf{r}_1-\mathbf{r}_2,\qquad r=\lvert\mathbf{r}\rvert,

we obtain

r˙=r˙1−r˙2=Γ2z^×r2πr2−Γ1z^×(−r)2πr2=Γ1+Γ22πz^×rr2=Γtot2πz^×rr2.(101)\begin{aligned} \dot{\mathbf{r}}&=\dot{\mathbf{r}}_1-\dot{\mathbf{r}}_2\\ &=\frac{\Gamma_2\hat{\mathbf{z}}\times\mathbf{r}}{2\pi r^2}-\frac{\Gamma_1\hat{\mathbf{z}}\times(-\mathbf{r})}{2\pi r^2}\\ &=\frac{\Gamma_1+\Gamma_2}{2\pi}\frac{\hat{\mathbf{z}}\times\mathbf{r}}{r^2} =\frac{\Gamma_{\mathrm{tot}}}{2\pi}\frac{\hat{\mathbf{z}}\times\mathbf{r}}{r^2}. \tag*{(101)} \end{aligned}

This shows that the relative velocity is everywhere perpendicular to r\mathbf{r}, implying that the separation rr is constant in time:

ddtr2=2r⋅r˙=0.(102)\frac{d}{dt}r^2 = 2\mathbf{r}\cdot\dot{\mathbf{r}} = 0. \tag*{(102)}

The motion is therefore a rigid rotation. To extract the angular frequency, we use the kinematic definition

ΩE=r×r˙r2.(103)\Omega_{\mathrm{E}} = \frac{\mathbf{r}\times\dot{\mathbf{r}}}{r^2}. \tag*{(103)}

Substituting the expression for r˙\dot{\mathbf{r}}, we obtain

r×r˙=Γtot2πr2r×(z^×r)=Γtot2πr2[z^(r⋅r)−r(r⋅z^)]=Γtot2πr2z^r2,(104)\begin{aligned} \mathbf{r}\times\dot{\mathbf{r}} &= \frac{\Gamma_{\mathrm{tot}}}{2\pi r^2}\mathbf{r}\times(\hat{\mathbf{z}}\times\mathbf{r})\\ &= \frac{\Gamma_{\mathrm{tot}}}{2\pi r^2}\left[\hat{\mathbf{z}}(\mathbf{r}\cdot\mathbf{r})-\mathbf{r}(\mathbf{r}\cdot\hat{\mathbf{z}})\right]\\ &= \frac{\Gamma_{\mathrm{tot}}}{2\pi r^2}\hat{\mathbf{z}}r^2, \tag*{(104)} \end{aligned}

where we used the vector identity a×(b×c)=b(a⋅c)−c(a⋅b)\mathbf{a}\times(\mathbf{b}\times\mathbf{c})=\mathbf{b}(\mathbf{a}\cdot\mathbf{c})-\mathbf{c}(\mathbf{a}\cdot\mathbf{b}) and r⋅z^=0\mathbf{r}\cdot\hat{\mathbf{z}}=0. Therefore,

ΩE=∣r×r˙∣r2=Γtot2πr2.(105)\Omega_{\mathrm{E}} = \frac{|\mathbf{r}\times\dot{\mathbf{r}}|}{r^2} = \frac{\Gamma_{\mathrm{tot}}}{2\pi r^2}. \tag*{(105)}

This establishes explicitly that the two-vortex system in the infinite plane undergoes uniform circular motion with constant angular frequency proportional to Γtot\Gamma_{\mathrm{tot}} and inversely proportional to the square of the separation. In the dipole case Γtot=0\Gamma_{\mathrm{tot}}=0, the relative velocity vanishes, and the pair translates rigidly with constant separation, consistent with the standard planar dynamics. This calculation shows that the exact torus expression (D1) reduces smoothly to the planar result, with the explicit geometric term vanishing and the constant contribution in the kernel canceling the corresponding term in the numerator. The recovery of the classical limit provides a nontrivial consistency check of the formulation.

References

  1. [2]D. V. Freilich, D. M. Bianchi, A. M. Kaufman, T. K. Langin and D. S. Hall, Real-time dynamics of single vortex lines and vortex dipoles in a Bose-Einstein condensate, Science 329 (2010) 1182-1185.DOI
  2. [3]A. C. White, C. F. Barenghi and N. P. Proukakis, Creation and Characterization of Vortex Clusters in Atomic Bose-Einstein Condensates, Physical Review A 86, 013635, (2012).DOI
  3. [4]A. C. White, B. P. Anderson, and V. S. Bagnato, “Vortices and turbulence in trapped atomic condensates”, Proc. Natl. Acad. Sci. U.S.A. 111, 4719-4726 (2014).DOI
  4. [5]G. Gauthier, M. T. Reeves, X. Yu, A. S. Bradley, M. Baker, T. A. Bell, H. Rubinsztein-Dunlop, M. J. Davis, and T. W. Neely, Giant vortex clusters in a two-dimensional quantum fluid, Science 364, 1264-1267 (2019).DOI
  5. [6]S. J. Rooney, P. B. Blakie, B. P. Anderson and A. S. Bradley, Suppression of Kelvon-induced decay of quantized vortices in oblate Bose-Einstein condensates, Phys. Rev. A 84, 023637 (2011).DOI
  6. [7]R. H. Goodman, P. G. Kevrekidis and R. Carretero-González, Dynamics of Vortex Dipoles in Anisotropic Bose-Einstein Condensates, SIAM J. Appl. Dyn. Syst. 14, no. 2, 699-729 (2015).
  7. [8]G. W. Stagg, N. G. Parker, and C. F. Barenghi, Ultraquantum turbulence in a quenched homogeneous Bose gas, Phys. Rev. A 94, 053632 (2016).DOI
  8. [9]P. Wiegmann and A. Abanov, “Anomalous hydrodynamics of two-dimensional vortex fluids,” Phys. Rev. Lett., vol. 113, 034501, 2014.pubmed.ncbi.nlm.nih.gov/25083647
  9. [10]B. Haskell and A. Melatos, Models of pulsar glitches, Int. J. Mod. Phys. D 24, 1530008 (2015).DOI
  10. [11]E. Lushi and P. M. Vlahovska, “Periodic and chaotic orbits of plane-confined micro-rotors in creeping flows,” Journal of Nonlinear Science, vol. 25, pp. 1111-1123, 2015.DOI
  11. [12]K. Yeo, E. Lushi, and P. M. Vlahovska, “Collective dynamics in a binary mixture of hydrodynamically coupled microrotors,” Phys. Rev. Lett., vol. 114, 188301, 2015.pubmed.ncbi.nlm.nih.gov/26001020
  12. [13]N. Oppenheimer, D. B. Stein, and M. J. Shelley, “Rotating membrane inclusions crystallize through hydrodynamic and steric interactions,” Phys. Rev. Lett., vol. 123, 148101, 2019.DOI
  13. [14]N. Oppenheimer, D. B. Stein, M. Y. B. Zion, and M. J. Shelley, “Hyperuniformity and phase enrichment in vortex and rotor assemblies,” Nature Communications vol. 13, 804, 2022.DOI
  14. [15]R. Samanta and N. Oppenheimer, “Vortex flows and streamline topology in curved biological membranes,” Physics of Fluids, vol. 33, 051906, 2021.DOI
  15. [16]P. K. Newton and G. Chamoun, Vortex lattice theory: A particle interaction perspective, SIAM Rev. 51, 501-542 (2009).DOI
  16. [17]J. B. Weiss and J. C. McWilliams, “Nonergodicity of point vortices in a square doubly periodic domain,” Physics of Fluids A, vol. 3, no. 5, pp. 835-844, 1991.
  17. [18]I. Kunin, F. Hussain, and M. Zhou, “Dynamics of a pair of vortices in a rectangle,” International Journal of Engineering Science, vol. 32, pp. 1835-1844, 1994.DOI
  18. [19]M. A. Stremler and H. Aref, Motion of three point vortices in a periodic parallelogram, J. Fluid Mech. 392, 101-128 (1999).DOI
  19. [20]M. A. Stremler, On relative equilibria and integrable dynamics of point vortices in periodic domains, Theor. Comput. Fluid Dyn. 24, 25-37 (2010).
  20. [21]A. C. H. Tsang and E. Kanso, “Dipole interactions in doubly periodic domains,” Journal of Nonlinear Science, vol. 23, no. 6, pp. 971-991, 2013.arxiv.org/abs/1304.6795
  21. [22]D. G. Crowdy and J. S. Marshall, Computing the Schottky-Klein prime function on the Schottky double of planar domains, Computational Methods and Function Theory 7, 293-308 (2007).
  22. [23]D. G. Crowdy, E. H. Kropf, C. C. Green, and M. M. S. Nasser, The Schottky-Klein prime function: a theoretical and computational tool for applications, IMA Journal of Applied Mathematics 81, 589-628 (2016).DOI
  23. [24]C. C. Green and J. S. Marshall, “Green’s function for the Laplace-Beltrami operator on a toroidal surface,” Proc. R. Soc. A, vol. 469, 20120479, 2012.DOI
  24. [25]V. S. Krishnamurthy and T. Sakajo, “The N-vortex problem in a doubly periodic rectangular domain with constant background vorticity,” Physica D: Nonlinear Phenomena, vol. 448, 133728, 2023.DOI
  25. [26]Aswathy K. R., U. Maurya, S. T. Gavva, and R. Samanta, “Dynamics of vortex clusters on a torus,” Phys. Fluids, vol. 37, 093324, 2025.DOI
  26. [27]Aswathy K. R., U. Maurya, S. T. Gavva, and R. Samanta, “Dynamics of vortex clusters on a torus including harmonic fields,” In preparation.
  27. [28]S. Boatto and J. Koiller, “Vortices on closed surfaces,” arXiv:0802.4313 (2008); also in Geometry, Mechanics and Dynamics, Fields Institute Communications.arxiv.org/abs/0802.4313
  28. [29]C. Grotta-Ragazzo, B. Gustafsson, and J. Koiller, “On the interplay between vortices and harmonic flows: Hodge decomposition of Euler’s equations in 2D,” Regular and Chaotic Dynamics, vol. 29, pp. 241-303, 2024 (arXiv:2309.12582).
  29. [30]B. Gustafsson, “Vortex pairs and dipoles on closed surfaces,” Journal of Nonlinear Science, vol. 32, p. 62, 2022.DOI
  30. [31]T. D. Drivas, D. Glukhovskiy, and B. Khesin, “Singular vortex pairs follow magnetic geodesics,” International Mathematics Research Notices, vol. 2024, no. 14, pp. 10880-10894, 2024 (arXiv:2401.08512).DOI
  31. [32]U. Maurya, S. T. Gavva, A. Saha, and R. Samanta, “Vortex dynamics in tubular fluid membranes,” Physics of Fluids, vol. 37, 073109, 2025.DOI
  32. [33]K. Banthia and R. Samanta, “A self propelled vortex dipole model on a surface of variable negative curvature,” Journal of Physics A: Mathematical and Theoretical, accepted manuscript (2026), doi:10.1088/1751-8121/ae56a1.DOI

Paper details

vlnc:W_gygbry79
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents