Generalized frustration in the multidimensional Kuramoto model

TL;DR

This paper extends the multidimensional Kuramoto model to arbitrary dimensions, analyzing how generalized coupling matrices influence synchronization and stability, revealing dimension-dependent transition behaviors and the emergence of active states.

Contribution

It generalizes the analysis of the Kuramoto model with matrix coupling to any dimension, exploring synchronization states and their stability across different dimensions.

Findings

Synchronization depends on the eigenvalues of the coupling matrix.

Even dimensions exhibit continuous transitions and active states.

Odd dimensions show discontinuous transitions with suppressed active states.

Abstract

The Kuramoto model was recently extended to arbitrary dimensions by reinterpreting the oscillators as particles moving on the surface of unit spheres in a D-dimensional space. Each particle is then represented by a D-dimensional unit vector. For $D=2$ the particles move on the unit circle and the vectors can be described by a single phase, recovering the original Kuramoto model. This multidimensional description can be further extended by promoting the coupling constant between the particles to a matrix that acts on the unit vectors, representing a type of generalized frustration. In a recent paper we have analyzed in detail the role of the coupling matrix for $D=2$. Here we extend this analysis to arbitrary dimensions, presenting a study of synchronous states and their stability. We show that when the natural frequencies of the particles are set to zero, the system converges either to a stationary synchronized state with well defined phase, or to an effective two-dimensional dynamics, where the synchronized particles rotate on the sphere. The stability of these states depend on the eigenvalues and eigenvectors of the coupling matrix. When the natural frequencies are not zero, synchronization depends on whether $D$ is even or odd. In even dimensions the transition to synchronization is continuous and rotating states are replaced by active states, where the order parameter rotates while its module oscillates. If $D$ is odd the phase transition is discontinuous and active states are suppressed, occurring only for a restricted class of coupling matrices.