Riddled Basin

A riddled basin is a basin of attraction in which every neighborhood of any point in the basin contains points belonging to the basin of a different attractor. The basin is not merely fractal or Wada; it is structurally porous, with holes belonging to other basins densely interwoven throughout its interior. This means that the basin has no interior in the topological sense: it is not an open set, and no matter how deep inside the basin a trajectory appears to be, an infinitesimal perturbation can cause it to escape to a different attractor.

The phenomenon was first identified by Alexander, Yorke, and others in the 1990s in the context of coupled chaotic systems. Two identical chaotic maps, coupled symmetrically, can synchronize to a common chaotic attractor. The basin of this synchronized state appears large and stable. But transverse to the synchronization manifold — the subspace where the two systems are identical — the dynamics are unstable. The basin of the synchronized state is riddled with points that belong to the basin of desynchronization. The result is that a trajectory can appear synchronized for an arbitrarily long time and then suddenly diverge.

Mathematical Structure

Riddled basins arise in systems with an invariant manifold that contains an attractor. The manifold is invariant because the dynamics on it are self-contained: trajectories that start on the manifold remain on it. The attractor on the manifold is stable in the manifold directions but unstable in the transverse directions. The transverse instability creates a set of points that escape from the manifold, and this set is dense in the basin of the manifold attractor.

The density is the key property. In a fractal basin boundary, the boundary itself is fractal, but the interior of the basin is safe: points deep in the interior remain in the basin under small perturbations. In a riddled basin, there is no safe interior. Every point in the basin is arbitrarily close to a point in another basin. The only sense in which the basin is "large" is measure: the riddled basin can occupy a large fraction of phase space in terms of Lebesgue measure, even though it has no topological interior.

This measure-topology mismatch is the defining characteristic of riddled basins. It is also the source of their physical relevance: a system can spend most of its time in a riddled basin, appearing stable, and then suddenly transition to a different attractor with no warning.

Physical Examples

Coupled chaotic oscillators. The paradigmatic example is two coupled logistic maps or coupled Rössler systems. When the coupling is strong enough, the oscillators synchronize to a common chaotic trajectory. But the synchronization is fragile: the basin of the synchronized state is riddled with points that lead to desynchronization. The riddling is caused by a transverse blowout bifurcation: the transverse Lyapunov exponent becomes positive, destabilizing the manifold.

Power grids. In models of interconnected power systems, the synchronous state — where all generators rotate at the same frequency — is an attractor on a manifold. The basin of this synchronized state can be riddled with states that lead to cascading failure. The implication is that the grid can appear stable for long periods and then suddenly lose synchronization due to a perturbation that would seem harmless by traditional stability analysis.

Neural networks. In networks with symmetric connectivity, the synchronized or balanced states can have riddled basins. Small perturbations that break the symmetry can send the network into a different dynamical regime. This may be relevant to the sudden onset of seizures or other pathological states in neural tissue.

Relation to Wada and Fractal Boundaries

Riddled basins are distinct from Wada basins, though they share the property of extreme topological complexity. A Wada basin has a boundary that is shared by three or more basins: every point on the boundary is on the boundary of all of them. A riddled basin has a boundary that is shared by at least two basins, but the sharing is denser: the other basin's points are inside the first basin, not merely on its boundary.

In a sense, riddled basins are the interior analogue of Wada boundaries. Wada boundaries are shared boundaries; riddled basins are shared interiors. Both represent the breakdown of the intuitive notion that a basin is a safe region separated from other basins by a boundary. In both cases, the geometry of the basin structure makes prediction impossible: no finite-precision measurement can determine which attractor will be reached.

Implications for Robustness

Riddled basins pose a fundamental challenge to the concept of robustness. A system is typically called robust if it returns to its attractor after small perturbations. But in a riddled basin, the attractor is not robust in any practical sense: the system can be arbitrarily close to the attractor and still be perturbed away from it. The only robustness that remains is measure-theoretic: the probability of a random perturbation causing escape may be small, but it is never zero.

This means that riddled basins force a probabilistic conception of stability. A system with a riddled basin is not stable in the classical sense; it is metastable, with a lifetime that depends on the noise level and the measure of the riddling. The engineering implications are severe: a system that appears stable by all standard metrics may still be vulnerable to rare but catastrophic transitions.