The Lagrangian and symplectic structures of the Kuramoto oscillator model Journal: J. Phys. A: Math. Gen.
Sherwin Kouchekian, Razvan Teodorescu · 2026
Read The Lagrangian and symplectic structures of the Kuramoto oscillator model Journal: J. Phys. A: Math. Gen. by Sherwin Kouchekian, Razvan Teodorescu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-00113.txt.
Introduction
Sherwin Kouchekian and Razvan Teodorescu revisit the Kuramoto model, a classic system of coupled oscillators used to study how rhythms fall into sync. After fifty years of study, the model still stands as a key example of a phase change far from equilibrium. It also seemed to lack a variational form, since its equations did not look compatible with a Lagrangian. Their paper shows that view is wrong.
Treating the model as a mean-field spin system on a sphere, the pair give it a proper variational description. Small disturbances around unstable balance points then match known spin models. Disturbances out of the plane match a semiclassical model from quantum physics. The result links synchronization to a pairing mechanism studied in that field.
Catalog record
| CREATOR | Sherwin Kouchekian, Razvan Teodorescu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-00113
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Lagrangian and symplectic structures of the Kuramoto oscillator model Journal: J. Phys. A: Math. Gen. Sherwin Kouchekian, Razvan Teodorescu Abstract Despite being under intense scrutiny for 50 years, the Kuramoto oscillator model has remained a quintessential representative of non-equilibrium phase transitions. One of the reasons for its enduring relevance is the apparent lack of an optimization formulation, due to the fact that (superficially), the equations of motion seem to not be compatible with a Lagrangian structure. We show that, as a mean-field classical (twisted) spin model on S 2 S^{2} , the Kuramoto model can be described variationaly. Based on this result perturbation analysis around (unstable) Kuramoto equilibria are shown to be equivalent to low-energy fluctuations of mean-field Heisenberg spin models. Intriguingly, off-plane perturbations around these equilibria configurations turn out to be described by a semiclassical Gaudin model, pointing to the fact that oscillator synchronization maps to the spin pairing mechanism investigated by Richardson and subsequently by others. , pacs 05.30, 05.40, 05.45 1 Introduction {sec:intro} Since its introduction 50 years ago, the Kuramoto oscillator model has been considered a fundamental example for dynamical systems exhibiting non-equilibrium phase transitions, in this case the synchronization phase transition. After reviewing the main characteristics of this system, we show that, in the original angular variables, the Langevin-type Kuramoto equations are not compatible with a Lagrangian structure, which is a main reason the problem was not given a variational formulation throughout the years. We then explore a complex bilinear structure which allows to retrieve most of the known exact results about this model. While still not Lagrangian, this exploration leads to the formulation of the problem as the mean-field limit of a spin system in \mathbb{R}^{3} , whose two-dimensional restriction reduces to the Kuramoto model. We therefore derive its corresponding Lagrangian and Hamiltonian, which allows to explore the planar and off-planar perturbations around Kuramoto equilibria. The planar perturbations confirm the known stability analysis of synchronization states, as a direct corollary of the exact Hamiltonian for the generalized model. The off-plane perturbations turn out to be described by a semiclassical Gaudin model with energy spectrum given by the centered frequency distribution of the Kuramoto oscillators. This result establishes a natural relationship between Kuramoto oscillator synchronization, on one hand, and (Anderson) pseudo-spin pairing in the semiclassical Gaudin model, on the other. 2 A brief review of the Kuramoto model {first} 2.1 Oscillators interacting via a mean-field attractive force - a first analysis of the phase-locking and synchronization phenomena. Let \theta_{j}(t),\,j=1,2,\ldots,N represent the angular variables of N oscillators, with (arbitrary) corresponding natural frequencies \omega_{j}\in\mathbb{R}_{+} . Denoting by \lambda\geq 0 the coupling constant of the mean-field attractive interaction between the oscillators, the Kuramoto equations take the Langevin form \dot{\theta_{j}}=\omega_{j}+\frac{\lambda}{N}\sum_{k=1}^{N}\sin(\theta_{k}-\theta_{j}):=F_{j}(\{\theta_{k}\}_{k=1}^{N}),\,\,j=1,2,\ldots,N, (1) and where \dot{f} represents the time derivative of the function f(t) . Two oscillators, identified by their indices, \{j,\ell\} , are said to be phase-locked if their angular difference is a constant: \dot{\theta_{j}}-\dot{\theta_{\ell}}=0\Leftrightarrow\omega_{j}+\frac{\lambda}{N}\sum_{k=1}^{N}\sin(\theta_{k}-\theta_{j})=\omega_{\ell}+\frac{\lambda}{N}\sum_{k=1}^{N}\sin(\theta_{k}-\theta_{\ell}) (2) If oscillators j,\ell are phase-locked, their phase difference, denoted by \Delta_{j\ell}:=\theta_{j}-\theta_{\ell}=-\Delta_{\ell j} , will be constant. This difference can be identified to the phase difference between the two angles, stationary in a co-rotating frame turning at an angular speed equal to the common value \dot{\theta_{j}}=\dot{\theta_{\ell}} . It is then an obvious proof by contradiction that, for phase-locked oscillators of different natural frequencies, \omega_{j}\neq\omega_{\ell} , their two phases cannot coincide: \dot{\theta_{j}}=\dot{\theta_{\ell}},\,\,\omega_{j}\neq\omega_{\ell}\Rightarrow\Delta_{j\ell}\neq 0. (3) Let us notice that a trivial estimate performed in Eq. 2 yields the following necessary condition for the phase locking of two oscillators, j,\ell , when N\geq 3 : \omega_{j}-\omega_{\ell}=\lambda\left[\frac{\sum_{k=1}^{N}\sin(\theta_{k}-\theta_{\ell})-\sin(\theta_{k}-\theta_{j})}{N}\right]\Rightarrow\lambda\geq\frac{N|\omega_{j}-\omega_{\ell}|}{2(N-1)} (4) We conclude that, for a system of N\geq 3 oscillators in which all the frequencies are distinct, for values of the coupling constant below a critical value \lambda_{c} \lambda<\lambda_{c}:=\frac{N}{2(N-1)}\min_{1\leq j\neq\ell\leq N}|\omega_{j}-\omega_{\ell}|, the system will always remain unlocked, while for values of the coupling constant exceeding a synchronization value \lambda_{s} \lambda\geq\lambda_{s}:=\frac{N}{2(N-1)}\max_{1\leq j\neq\ell\leq N}|\omega_{j}-\omega_{\ell}|, the system may become completely phase locked, for special configurations of the relative phases, \{\Delta_{j\ell}\}_{j,\ell=1}^{N} . In particular, a Kuramoto system in which oscillators have the same, common frequency \omega_{j}=\Omega,\,\forall\,j=1,2,\ldots,N (for which \lambda_{c}=0=\lambda_{s} ) will always possibly fully phase-lock at arbitrarily small values of the coupling constant, \lambda>0 . In the special case N=2 , we obtain the reduction \lambda_{c}=\lambda_{s}=|\omega_{1}-\omega_{2}|, with the phase-locking condition \sin\Delta_{12}=\frac{\omega_{1}-\omega_{2}}{\lambda}. In the general case N\geq 3,\,0<\lambda_{c}<\lambda<\lambda_{s} , the system will exhibit a mixture of the two phases, with a relative proportion of phase-locked oscillators depending on the precise distribution of the frequencies, relative to the values \lambda,\lambda_{c},\lambda_{s} . These features (the system exhibiting a mixture of phases while in a dynamic, time-dependent state, and possibly also depending on initial conditions through the relative phases \Delta_{j\ell} ) have since been generalized as generic characteristics of non-equilibrium phase transitions. It should be emphasized that this class of phenomena, in contrast with the much better-established theory of equilibrium phase transitions, in which the existence of a well-defined globally convex functional (a thermodynamic potential) and its degree of smoothness allow to classify and indeed fully describe the critical properties of a system (such as in first-order phase transitions, second-order, etc.) fundamentally lacks a variational formulation, as we demonstrate in the next section. 2.2 Analyzing variational presentations of the Kuramoto model in two dimensions 2.2.1 Non-Lagrangian nature of the original Kuramoto equations. Let us first note that, superficially, Eqs. 1 cannot be realized as Euler-Lagrange equations for a Lagrangian density {L}(\theta_{j},\dot{\theta}_{j}) : taking one derivative in Eqs. 1 , we obtain \ddot{\theta_{j}}=\frac{\lambda}{N}\sum_{k=1}^{N}\cos(\Delta_{kj})\left[\omega_{k}-\omega_{j}+\frac{\lambda}{N}\sum_{\ell=1}^{N}\sin(\Delta_{\ell k})-\sin(\Delta_{\ell j})\right] (5) Assuming that Eqs. 5 are Euler-Lagrange equations for {L}(\theta_{j},\dot{\theta}_{j}) , it would follow that {L}(\theta_{j},\dot{\theta}_{j})=\sum_{j=1}^{N}\frac{1}{2}\dot{\theta}^{2}_{j}+\Phi(\{\theta_{k}\}_{k=1}^{N}), (6) since the Lagrangian cannot contain terms of the form \dot{\theta_{j}^{2}} (which would generate a first-order derivative in Eqs. 5 ), as they would only have a boundary contribution to the action, therefore no effect under the Euler-Lagrange first variation. We are therefore seeking a function \Phi(\{\theta_{k}\}_{k=1}^{N}) such that \frac{\partial\Phi}{\partial\theta_{j}}=\frac{\lambda}{N}\sum_{k=1}^{N}\cos(\Delta_{kj})\left[\omega_{k}-\omega_{j}+\frac{\lambda}{N}\sum_{\ell=1}^{N}\sin(\Delta_{\ell k})-\sin(\Delta_{\ell j})\right] (7) We prove that this is inconsistent by verifying that, for general j\neq q , \frac{\partial^{2}\Phi}{\partial\theta_{j}\partial\theta_{q}}\neq\frac{\partial^{2}\Phi}{\partial\theta_{q}\partial\theta_{j}}. (8) From Eq. 7 , we have \frac{\partial^{2}\Phi}{\partial\theta_{q}\partial\theta_{j}}=\frac{\lambda}{N}(\omega_{j}-\omega_{q})\sin(\Delta_{qj})+\left(\frac{\lambda}{N}\right)^{2}\sum_{k,\ell=1}^{N}\sin(\Delta_{jk})\left[\sin(\Delta_{lk})-\sin(\Delta_{\ell j})\right][\delta_{kq}-\delta_{jq}] +\left(\frac{\lambda}{N}\right)^{2}\left\{\sum_{k,\ell=1}^{N}\cos(\Delta_{kj})\left[\cos(\Delta_{\ell k})(\delta_{\ell q}-\delta_{kq})-\cos(\Delta_{\ell j})(\delta_{\ell q}-\delta_{jq})\right]\right\} Therefore, for j\neq q , the difference \frac{\partial^{2}\Phi}{\partial\theta_{j}\partial\theta_{q}}-\frac{\partial^{2}\Phi}{\partial\theta_{q}\partial\theta_{j}}=\left(\frac{\lambda}{N}\right)^{2}\cos(\Delta_{qj})\sum_{k=1}^{N}\left[\cos(\Delta_{kj})-\cos(\Delta_{kq})\right], which cannot be 0 unless for very special values of \{\theta_{j}\},j\in\Sigma_{N} . We conclude that, in its original angular variables, the close-form conditions do not hold (Eqs. 8 ), and the Kuramoto equations Eq. 5 are not of Euler-Lagrange form. 2.2.2 A first Poisson bracket structure for Kuramoto oscillators. Introducing the notation z_{j}(t):=e^{i\theta_{j}(t)},\,\,j\in\Sigma_{N}:=\{1,2,\ldots,N\} , we arrive at the following formulation for the Kuramoto equations: \dot{z}_{j}=z_{j}\left[i\omega_{j}+\frac{\lambda}{2N}\sum_{k=1}^{N}(z_{k}\bar{z}_{j}-\bar{z}_{k}z_{j})\right] (9) Defining the following antisymmetric and symmetric bilinear forms on \mathbb{C}\times\mathbb{C}\to\mathbb{R} , [u,v]:=\frac{1}{2i}(u\bar{v}-v\bar{u}),\quad\{u,v\}:=\frac{1}{2}(u\bar{v}+v\bar{u}), (10) as well as the complex order parameter , r(t)=\frac{1}{N}\sum_{k=1}^{N}z_{k}(t), we arrive at the Kuramoto equations in complex form: \dot{z}_{j}=iz_{j}(\omega_{j}+\lambda[r,z_{j}]) (11) We will denote the modulus and argument of the complex order parameter as r(t)=|r|(t)e^{i\theta_{0}(t)}\Rightarrow 0\leq|r|(t)\leq 1, and locking of oscillators with the order parameter variable is defined the same way as already introduced. Theorem 1 {Thm_properties} Let \alpha\in\mathbb{R} and u,v\in\mathbb{C}^{*} . We list the following properties of the bilinear forms defined in Eq. 10 : [u,v]=-[v,u],\quad[u,[v,w]]+[w,[u,v]]+[v,[w,u]]=0, (12) \{u,v\}=\{v,u\},\quad[\alpha u,v]=\alpha[u,v],\quad[i\alpha u,v]=i\alpha\{u,v\}, (13) [u,v]=0\Leftrightarrow\arg\frac{u}{v}\in\{0,\pi\},\quad\{u,v\}=0\Leftrightarrow\arg\frac{u}{v}\in\left\{-\frac{\pi}{2},\frac{\pi}{2}\right\}. (14) The proof is based on elementary computations. This formalism allows to retrieve in a convenient manner conditions for phase-locking of two or several oscillators: Theorem 2