Skip to content

Kuramoto Theory — Mathematical Foundations

The Kuramoto Model

Yoshiki Kuramoto (1975) proposed the simplest model of synchronization in a population of coupled oscillators:

dθ_i/dt = ω_i + (K/N) Σ_{j=1}^N sin(θ_j - θ_i)

where: - θ_i ∈ [0, 2π): phase of oscillator i - ω_i: natural frequency (drawn from distribution g(ω)) - K: global coupling strength - N: number of oscillators

Order Parameter

The complex order parameter z = R·exp(iΨ) summarizes the collective state:

z = (1/N) Σ_{j=1}^N exp(iθ_j)
  • R = |z| ∈ [0, 1]: coherence (0 = incoherent, 1 = perfect sync)
  • Ψ = arg(z): mean phase

The Kuramoto model can be rewritten using z:

dθ_i/dt = ω_i + K·R·sin(Ψ - θ_i)

Each oscillator is pulled toward the mean phase Ψ with force proportional to both K and R. This creates a positive feedback loop: partial sync (R > 0) creates a stronger pull, which increases sync further.

Critical Coupling

For a symmetric unimodal frequency distribution g(ω), there exists a critical coupling K_c below which the incoherent state (R = 0) is stable:

K_c = 2 / (π · g(0))

For Lorentzian g(ω) with half-width Δ centered at ω₀:

K_c = 2Δ

Above K_c, the order parameter grows as R ~ √(K - K_c).

Ott-Antonsen Reduction

For Lorentzian frequency distributions, the infinite-N Kuramoto model reduces exactly to a single complex ODE (Ott & Antonsen, 2008):

dz/dt = -(Δ + iω₀)z + (K/2)(z - |z|²z)

Steady-state: R_ss = √(1 - 2Δ/K) for K > K_c.

This is implemented in scpn_phase_orchestrator.upde.reduction and used by the MPC supervisor for O(1) prediction.

Generalized Coupling

SPO extends the basic model with:

In these SCPN forms the coupling enters as a per-pair matrix K_ij (the K_nm matrix), which absorbs the 1/N normalisation written explicitly as (K/N) in the classic form above: to reproduce a uniform global coupling of strength K, set every K_ij = K/N; a heterogeneous K_nm encodes the network topology directly. The normalisation is therefore the caller's responsibility, not applied implicitly.

Sakaguchi-Kuramoto (Phase Lags)

dθ_i/dt = ω_i + Σ_j K_ij sin(θ_j - θ_i - α_ij)
The phase lag α_ij models transport delays or asymmetric coupling.

Stuart-Landau (Amplitude Dynamics)

dθ_i/dt = ω_i + Σ_j K_ij sin(θ_j - θ_i)
dr_i/dt = (μ_i - r_i²)r_i + ε Σ_j K^r_ij r_j cos(θ_j - θ_i)
Adds amplitude r_i with Hopf bifurcation: μ > 0 → r → √μ (active), μ < 0 → r → 0 (quiescent).

Simplicial (Higher-Order)

dθ_i/dt = ω_i + (σ₁/N) Σ_j K_ij sin(θ_j - θ_i)
                + (σ₂/N²) Σ_{j,k} sin(θ_j + θ_k - 2θ_i)
3-body interactions produce explosive (first-order) synchronization transitions (Gambuzza et al. 2023, Nature Physics).

Inertial (Second-Order)

m_i θ̈_i + d_i θ̇_i = P_i + Σ_j K_ij sin(θ_j - θ_i)
The swing equation for power grid stability. Inertia m_i represents rotating mass of generators.

Spectral Alignment Function

The Synchrony Alignment Function (Skardal & Taylor, 2016) provides a closed-form approximation of R from the Laplacian eigenstructure of K:

R ≈ 1 - (1/2N) Σ_{j=2}^N λ_j⁻² ⟨v^j, ω⟩²

where λ_j are eigenvalues and v^j eigenvectors of the graph Laplacian L = D - K. This enables coupling topology optimization without ODE integration (10x faster).

Inverse Problem

Given observed phases θ_i(t), infer K and ω by minimising:

L(K, ω) = Σ_t (1 - cos(θ_predicted(t) - θ_observed(t))) + λ·‖K‖₁

The L1 penalty on K promotes sparsity (topology discovery). Gradients computed via JAX autodiff through the ODE solver.

How this maps to controls in the product

Each mathematical form links to a concrete engineering path:

  • pairwise and lagged couplings feed the standard UPDE derivatives,
  • amplitude terms map to Stuart-Landau and related imprint experiments,
  • higher-order terms map to simplicial coupling studies,
  • inertial terms map to swing-domain workflows where stability boundaries are evaluated against grid-like operating constraints.

This model page is used to keep mathematical assumptions visible next to control behavior. The same equations inform guard thresholds, action constraints, and the conditions under which a policy is allowed to adjust coupling or phase forcing.

References

  • Kuramoto Y (1975). Self-entrainment of a population of coupled non-linear oscillators. Lecture Notes in Physics 39:420-422.
  • Ott E, Antonsen TM (2008). Low dimensional behavior of large systems of globally coupled oscillators. Chaos 18:037113.
  • Skardal PS, Taylor D (2016). Optimal synchronization of directed complex networks. Chaos 26:094807.
  • Gambuzza LV et al. (2023). Stability of synchronization in simplicial complexes. Nature Physics 19:1427-1434.
  • Filatrella G et al. (2008). Analysis of a power grid using a Kuramoto-like model. European Physical Journal B 61:485-491.
  • O'Keeffe KP et al. (2017). Oscillators that sync and swarm. Nature Communications 8:1504.