Mean-field approach to finite-size fluctuations in the Kuramoto-Sakaguchi model

Oleh E. Omel’chenko, Georg A. Gottwald · 2025

Read Mean-field approach to finite-size fluctuations in the Kuramoto-Sakaguchi model by Oleh E. Omel’chenko, Georg A. Gottwald (2025) 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-2511-03700.txt.

Introduction

Oleh E Omel'chenko and Georg A Gottwald derive exact statistics for the flicker that finite size causes in synchronized oscillator crowds. The Kuramoto-Sakaguchi model, a standard for coupled rhythms with phase lag, settles into partial synchrony in the infinite limit. Real crowds are finite, so their shared order parameter wobbles around that limit.

An explicit solution of a key single-oscillator equation unlocks covariance formulas for the wobble, plus the variance of its strength, all written purely in model parameters. No prior picture of the synchronized state is assumed. Simulations of the full model confirm the formulas on both sides of the sync threshold. Power grids and neural rhythms both lean on such models, so honest finite-size error bars matter.

Catalog record

CREATOROleh E. Omel’chenko, Georg A. Gottwald
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE44 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-03700

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Mean-field approach to finite-size fluctuations in the Kuramoto-Sakaguchi model Oleh E. Omel’chenko, Georg A. Gottwald Abstract We develop an ab initio approach to describe the statistical behavior of finite-size fluctuations in the Kuramoto-Sakaguchi model. We obtain explicit expressions for the covariance function of fluctuations of the complex order parameter and determine the variance of its magnitude entirely in terms of the equation parameters. Our results rely on an explicit complex-valued formula for solutions of the Adler equation. We present analytical results for both the sub- and the super-critical case. Moreover, our framework does not require any prior knowledge about the structure of the partially synchronized state. We corroborate our results with numerical simulations of the full Kuramoto-Sakaguchi model. The proposed methodology is sufficiently general such that it can be applied to other interacting particle systems. Introduction. Synchronization is a generic phenomenon inherent in interacting oscillators that determines their ability to exhibit a wide range of complex dynamical behavior 26 ; 43 ; 37 ; 1 ; 2 ; 12 ; 39 . It has been observed in a plethora of natural and engineered systems, including pace-maker cells of circadian rhythms 46 , networks of neurons 3 , chemical oscillators 24 ; 45 and power grids 14 . The main features of synchronization are described by the paradigmatic Kuramoto-Sakaguchi (KS) model 40 \dot{\psi}_{i}=\omega_{i}+\frac{\displaystyle K}{\displaystyle N}\sum\limits_{j=1}^{N}\sin(\psi_{j}-\psi_{i}-\lambda) (1) for the collective behavior of a system of N all-to-all coupled phase oscillators. In this model, \omega_{i} are natural frequencies drawn randomly and independently from a given distribution g(\omega) and K and \lambda are real parameters quantifying the coupling strength and the phase lag, respectively. It is often convenient to rewrite ( 1 ) in the corotating frame \theta_{i}=\psi_{i}-\Omega t with a global rotation frequency \Omega as \dot{\theta}_{i}=\omega_{i}-\Omega+K\>\mathrm{Im}\left(Z(t)e^{-i\theta_{i}}e^{-i\lambda}\right), (2) which reveals that each oscillator interacts with the other oscillators only via the mean field. Importantly, the degree of synchronization is quantified by the mean field, also coined the complex order parameter, Z(t)=\frac{\displaystyle 1}{\displaystyle N}\sum\limits_{j=1}^{N}e^{i\theta_{j}(t)}. (3) A fully synchronized state with \theta_{1}=\theta_{2}=\dots=\theta_{N} is characterized by r(t)=|Z(t)|=1 , whereas a completely disordered (incoherent) state is characterized by r(t)\sim\mathcal{O}(1/\sqrt{N}) . It is well known that for sufficiently large values of the coupling strength K the system becomes fully synchronized, whereas for sufficiently small values of K the oscillators behave independently and form an incoherent state. For intermediate values of the coupling strength, partially synchronized states are observed in which r(t) fluctuates around a constant value between 0 and 1 . On a microscopic level the oscillator population splits into two groups: coherent oscillators \mathcal{C} , which are phase locked, and rogue oscillators \mathcal{R} , which drift with respect to the mean field. In the thermodynamic limit of infinitely many oscillators each oscillator interacts with the other oscillators via a constant collective mean. This limit is well understood and described by mean-field theory 40 ; 31 ; 32 ; 33 . In particular, the rotation frequency of the coherent oscillators \Omega_{\infty} and the constant order parameter r_{\infty} are determined by a self-consistency relation \frac{\displaystyle 1}{\displaystyle K}e^{i\lambda}=i\int_{-\infty}^{\infty}g(\Omega_{\infty}+Kr_{\infty}s)h(s)ds, (4) where h(s)=\left\{\begin{array}[]{lcl}(1-\sqrt{1-s^{-2}})s&\mbox{ for }&|s|>1\\[2.84526pt] s-i\sqrt{1-s^{2}}&\mbox{ for }&|s|\leq 1.\end{array}\right. (5) However, recently finite-size effects have gained much attention. Numerical experiments have lead to a better understanding of finite-size effects and provide clear evidence for their significance 8 ; 9 ; 6 ; 30 ; 21 ; 42 ; 48 ; 44 . In particular, several finite-size phenomena were observed such as a stochastic drift of oscillators which disappears in the thermodynamic limit 15 ; 28 ; 17 ; 18 and emergent random chimera switching in a deterministic two-population KS model 23 . Capturing finite-size effects analytically is a notoriously hard problem. Relaxation rates and multi-oscillator were obtained using kinetic theory 19 ; 4 ; 47 and finite-size scaling near criticality has been successfully quantified 22 ; 21 ; 20 ; 35 . A general scheme for analyzing fluctuations in systems of coupled phase oscillators was proposed in 25 ; 10 ; 7 . In particular, accurate approximations were obtained in the subcritical case of the KS model ( 1 ) with \lambda=0 and a Lorentzian frequency distribution 7 . However, this approach relies on several approximations which do not carry over to the case of more general frequency distributions such as Gaussians as we will show (see Appendix). In this Letter we will capture finite-size effects in a first principle way and provide explicit expressions for the covariance function and the variance of the order parameter. Our focus will be the hitherto unexplored parameter range away from criticality. Recently, it was numerically recognized that fluctuations \zeta(t)=\sqrt{N}(Z(t)-\langle Z(t)\rangle) (6) around the thermodynamic mean \langle Z(t)\rangle , where \langle\cdot\rangle denotes time averaging, approximately obey a Gaussian process 48 ; 23 . Fluctuations in Z(t) , with a variance that decays as 1/N and vanishes in the thermodynamic limit, lead to fluctuations in other macroscopic variables, including the order parameter with fluctuations \delta(t)=\sqrt{N}(|Z(t)|-\langle|Z(t)|\rangle)=\sqrt{N}(r(t)-\langle r(t)\rangle). (7) Gaussian processes are entirely determined by their mean, accessible via classical mean-field theory, and their covariance function R(\tau)=\langle\zeta(t)\overline{\zeta(t+\tau)}\rangle, (8) where the bar denotes the complex-conjugate. Once determined, this suggests to approximate the deterministic dynamics of the synchronized oscillators \theta_{i} , which are deterministically driven by Z(t) , by an effective stochastic equation for \theta_{i} , which is driven by a Gaussian process 16 . Previous work 48 ; 23 has shown the efficacy of such an approach, however, these works relied on approximating the general Gaussian process by an Ornstein-Uhlenbeck process the parameters of which were estimated only numerically and needed to be recalculated for each set of equation parameters. Moreover, a precise a priori knowledge of which oscillators are synchronized and which are not was required. In this Letter, we find explicit analytical expressions for the covariance function ( 8 ) and the variance of finite-size fluctuations of the order parameter V=\langle\delta^{2}(t)\rangle , given entirely in terms of equation parameters K , \lambda and g(\omega) . We present expressions for both the partially synchronized and the completely incoherent state. Remarkably, our expressions do not require any a priori knowledge about which oscillators partake in the synchronized cluster and which do not. Our main results can be formulated as follows. For partially synchronized states we predict R_{\mathrm{ps}}(\tau)=Kr_{\infty}\int\limits_{-\infty}^{\infty}\frac{\displaystyle(\overline{h^{2}(s)}-1)(1-|h^{2}(s)|)g(\Omega_{\infty}+Kr_{\infty}s)}{\displaystyle\overline{h^{2}(s)}-\exp(iKr_{\infty}s\sqrt{1-s^{-2}}\>\tau)}ds (9) and V_{\mathrm{ps}}=\frac{\displaystyle 1}{\displaystyle 2}R_{\mathrm{ps}}(0)+\mathcal{O}\left(\frac{\displaystyle 1}{\displaystyle\sqrt{N}}\right)\quad\mbox{for}\quad N\gg 1. (10) For completely incoherent states we find R_{\mathrm{incoh}}(\tau)=\int_{-\infty}^{\infty}g(\omega)e^{-i\omega\tau}d\omega (11) and \sqrt{N}\langle r(t)\rangle=\frac{\displaystyle\sqrt{\pi}}{\displaystyle 2}\quad\mbox{and}\quad V_{\mathrm{incoh}}=1-\frac{\displaystyle\pi}{\displaystyle 4}\quad\mbox{for}\quad N\gg 1. (12) Our analytical results and their efficacy in capturing finite-size fluctuations are illustrated in Figs. 1 and 2 , where we compare our predictions with simulations of the full KS model ( 1 ) for a network of 1,000 oscillators. Fig. 1 shows the covariance function R(\tau) for a partially synchronized state described by ( 9 ) and for a completely incoherent state described by ( 11 ). Fig. 2 (a) shows the averaged order parameter \langle r(t)\rangle as a function of the coupling strength K . The order parameter of the full KS model ( 1 ) is very well reproduced by classical mean-field theory ( 4 ) for values of the coupling strength K\geq K_{\rm{crit}} . Moreover, our extension of the mean-field results to the completely incoherent state given by ( 12 ) captures the value of

Read the complete text