Topological Kuramoto Model: Dynamics & Multistability

• The Topological Kuramoto Model is a framework that integrates graph and cell complex topology with coupled oscillator dynamics, using winding numbers to classify phase-locked states.
• It employs gradient flows and Morse theory to analyze synchronization, multistability, and topological phase transitions in oscillator networks.
• The approach has practical applications in power grids, neuroscience, and engineered oscillatory systems by predicting behavior under complex connectivity.

The Topological Kuramoto Model designates a class of generalizations of the classical Kuramoto system that explicitly incorporates topological data—such as cycles, higher-dimensional cells, winding numbers, homology class, and symmetry—when analyzing the dynamics, multistability, and synchronization of coupled oscillator networks. The model includes gradient flows on graph and cell complex configuration spaces, and utilizes algebraic-topological invariants (notably winding numbers) to parameterize and classify equilibrium states, bifurcations, and dynamical phenomena. This framework exposes universal mechanisms underpinning multistability, phase transitions, and the emergence of robust or exotic patterns across a wide range of network topologies.

1. Topological Formulation: From Graphs to Cell Complexes

The classical Kuramoto model describes $N$ oscillators with phases $\theta_i \in S^1$ and adjacency matrix $A$. The dynamics

$$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$

are invariant under global phase shifts (the diagonal $S^1$ action), and organize on the $N$-torus phase space $\mathbb{T}^N$. This symmetry allows reduction to a quotient phase space (relative phases) $Q \cong \mathbb{T}^{N-1}$ (Burns et al., 21 May 2026).

The topological generalization emerges upon realizing that on a network with nontrivial cycles, the steady states (fixed points) decompose into distinct classes, each labeled by integer-valued winding numbers that count the net phase twist around each independent cycle (Delabays et al., 2015, Ferguson, 2017, Medvedev et al., 15 Jun 2025). For a graph $G$ with cycle space of dimension $c = |E| - |V| + 1$, each equilibrium can be assigned a winding vector $\theta_i \in S^1$0, corresponding to the total $\theta_i \in S^1$1 phase increment along the $\theta_i \in S^1$2 fundamental cycles.

Generalizing further, for a regular cell or simplicial complex, phases can be attached to $\theta_i \in S^1$3-cells and the coupling defined via coboundary (discrete exterior derivative) and boundary operators. In this context, the equations of motion become (Bačić et al., 7 Oct 2025, Nurisso et al., 2023): $\theta_i \in S^1$4 where $\theta_i \in S^1$5 and $\theta_i \in S^1$6 are incidence matrices, $\theta_i \in S^1$7 coupling matrices, and all nonlinearities act componentwise.

2. Winding Numbers, Loop Currents, and Homotopy Classes

In networks with nontrivial cycle topology, the difference of any two phase-locked solutions is a combination of quantized loop currents (Delabays et al., 2015): $\theta_i \in S^1$8 where $\theta_i \in S^1$9 is a continuous parameter and $A$0 is the topological winding number around the ring. The set of physically distinct phase-locked states is indexed by $A$1; solutions with different $A$2 cannot be deformed into one another through continuous phase shifts without introducing a discontinuity of $A$3.

Abstractly, for each fundamental cycle or higher-dimensional cell, one may define winding numbers for the phase variables, generalizing to homotopy invariants for maps from a network or fractal (e.g., the Sierpinski gasket) to the circle. On a post-critically finite fractal $A$4, each equilibrium class is uniquely labeled by a multicomponent degree vector $A$5, which determines a unique harmonic (Dirichlet energy minimizing) map; there is exactly one stable equilibrium in each homotopy class in the continuum limit (Medvedev et al., 15 Jun 2025).

In cell complexes, the winding constraints manifest as "topological nonlinear Kirchhoff conditions," which govern the multistability structure and feasibility of fixed points via admissible integer winding vectors (Bačić et al., 7 Oct 2025).

3. Morse Theory, Cell Decomposition, and Robustness

In the all-to-all case, the global dynamics admit a complete Morse-theoretic classification on the reduced (modulo $A$6) phase space $A$7. The Kuramoto vector field is a gradient flow: $A$8 with symmetric potential

$A$9

whose critical points are characterized as follows (Burns et al., 21 May 2026):

• Global minima (synchrony): $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$0, Morse index 0.
• Saddles: for subsets $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$1 of $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$2 nodes with $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$3 ($$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$4), Morse index $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$5.
• Global maxima (source): phases $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$6 where the centroid vanishes.

The unstable manifolds (indexed by $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$7) tile $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$8 as toroidal "templates," forming a Morse–Smale complex corresponding to the lower skeleton of the torus. Homological invariants (Betti numbers, deformation retracts, Alexander duality) quantify the attractor and its structure. This cell decomposition and the associated Morse indices are robust under $$\dot\theta_i = \omega_i + K \sum_{j} A_{ij}\sin(\theta_j - \theta_i)$$9-small perturbations of the flow: all critical manifolds persist as normally-hyperbolic sets and no new low-potential bifurcations emerge.

4. Topological Multistability and Bounds

The number of stable fixed points in the topological Kuramoto model is governed by winding number constraints and cell structure. On a ring of $S^1$0 nodes, the number of linearly stable phase-locked solutions satisfies (Delabays et al., 2015, Bačić et al., 7 Oct 2025): $S^1$1 and each physically distinct solution is characterized by a winding number $S^1$2 such that $S^1$3.

For general cell complexes, the argument extends and the set of normal fixed points is in bijection with certain admissible integer winding vectors $S^1$4 subject to polyhedral bounds determined by the number of boundary cells: $S^1$5 Cell complexes consequently support a much richer multistability structure than simplicial complexes; e.g., a cube admits $S^1$6 normal fixed points, the dodecahedron $S^1$7, as determined by the counts of vertices and faces (Bačić et al., 7 Oct 2025).

On general graphs, a full classification of steady states is achieved via winding-number maps, with a Weyl-type asymptotic: $S^1$8 for a graph refined by subdividing edges, where $S^1$9 is the cycle-space dimension. When all edge-differences are restricted to $N$0, all labeled fixed points are linearly stable (Ferguson, 2017).

5. Topological Defects and Exotic Phase Transitions

Topological currents in Kuramoto networks have direct analogs in condensed matter physics: persistent (quantized) currents in superconducting rings, vortices in 2D XY magnets, and Josephson junction arrays (Delabays et al., 2015, Flovik et al., 2016). These defects appear as quantized windings of the phase field, characterized locally by discrete contour integrals of the phase difference. In 2D oscillator lattices with disorder or noise (Rouzaire et al., 2022, Flovik et al., 2016), these defects do not bind (as in the classical BKT vortex scenario) but remain free and undergo superdiffusive random walks, driven by stress along dynamic domain boundaries.

The periodic forcing of Kuramoto networks can induce a novel topological phase transition ("SPOR") in the order-parameter space, in which the order-parameter trajectory acquires or loses a unit winding about a singularity; this transition is not associated with a local bifurcation, symmetry breaking, or critical slowing down, but is purely topological and defined by a singular change in global invariants (Wright et al., 2020). In quantum engineering contexts, self-organized classical Kuramoto fields can drive quantum spin chains into Floquet phases with emergent (space/time) translation symmetry or quantized topological pumps set by band Chern numbers (Bastidas, 2024).

6. Extensions: Higher-Order, Fractal, and Simplicial Topologies

The topological Kuramoto framework extends naturally to higher-order cell (simplicial) complexes and fractal graphs. In edge-based and face-based models, oscillator phases are assigned to $N$1-cells and coupled through higher-order Laplacians and boundary/coboundary operators (Nurisso et al., 2023, Bačić et al., 7 Oct 2025). Each phase-locked state is then subject to generalized Kirchhoff–type conditions and classified by multi-index winding vectors.

On post-critically finite fractals such as the Sierpinski gasket, the space of stable equilibria is indexed by all possible degree vectors in the countable family of non-contractible cycles at each refinement stage; in the continuum limit, there is a unique minimal energy equilibrium in each homotopy class (Medvedev et al., 15 Jun 2025). This yields a hierarchical and self-similar family of stable patterns unachievable on classical lattices.

The Kuramoto model has been further generalized to $N$2-dimensional toroidal phase spaces (each oscillator described by $N$3 angular coordinates), where the synchronization transition becomes first-order (saddle-node), generating irreducible bistability and discontinuities absent in $N$4 versions (Novaes, 1 May 2026).

7. Implications, Methods, and Applications

The topological Kuramoto paradigm enables:

• Algorithmic determination of all phase-locked states via solving affine-linear flow equations and enforcing nonlinear Kirchhoff (winding) conditions (Bačić et al., 7 Oct 2025).
• Quantitative prediction of multistability bounds for arbitrary topologies, essential for power grid engineering, neuroscience, and oscillatory material design.
• Classification of phase transitions: whether local (bifurcation-theoretic), global (topological), or mixed.
• Control and inference: The theory enables reconstructing network topology from dynamical response (especially via pacemaker detuning experiments (Prignano et al., 2011)) and the design of target states via frustration tuning (Nurisso et al., 2023).

A key insight is that the combinatorial topology of the underlying network (cycle space, rank, boundary cell count, homology type) fundamentally organizes and constrains every aspect of the collective dynamics, from the number and stability of phase-locked patterns to the existence and structure of bifurcations, domain formation, and topological defects.

---

References:

(Delabays et al., 2015, Ferguson, 2017, Medvedev et al., 15 Jun 2025, Bačić et al., 7 Oct 2025, Nurisso et al., 2023, Flovik et al., 2016, Wright et al., 2020, Rouzaire et al., 2022, Prignano et al., 2011, Bastidas, 2024, Novaes, 1 May 2026, Burns et al., 21 May 2026) and related works as cited in the details above.