A blender is one of the most exotic objects in dynamical systems: a hyperbolic invariant set whose stable or unstable manifold is so wildly folded that it intersects every curve in a nearby region. Blenders are the mechanism behind robust transitivity and persistent heterodimensional cycles — they create connections between invariant sets that survive perturbation.
Note (2026-11-13; read against the paper's full text): the essay's characterization of the covering property is not the paper's. The paper defines a blender through a family of u-dimensional discs such that, for every map in a C1-open neighbourhood and every disc in the family, some subset of the disc is mapped onto another disc of the family, so that the discs robustly intersect the stable manifold of an invariant hyperbolic set; it is a reproduction property under perturbation, not a statement that the image of any curve meets every curve in the family. The algorithm builds such families as tubes of curves around an approximate strong unstable direction and checks the conditions with interval arithmetic. The application is a family of three-dimensional Henon-like maps, where the paper verified blenders for individual parameter values up to xi = 1.716, says that for higher values its program can only be run on very small intervals of xi, and says memory limitations prevent it from covering the entire range. The closing images about shadows and curves that cannot avoid intersecting are the essay's own.
The paper on detecting and verifying blenders (arXiv: 2603.21603) presents an algorithm that rigorously proves a given dynamical system contains a blender, using computer-assisted methods.
The key insight is that the algorithm doesn't need precise knowledge of invariant manifolds or fixed points. It works with sets of curves and their images under the dynamics. If a family of curves maps back into itself in a way that covers the region — if the image of any curve in the family intersects every curve in the family — then a blender exists. This covering property can be verified with interval arithmetic from only a rough approximation of the unstable direction.
The through-claim: existence is proved by coverage, not by construction. You don't find the blender — you show it must be there because curves can't avoid intersecting. The algorithm verifies a topological property (every curve gets hit) rather than locating a geometric object (the invariant set). The blender is the shadow cast by the covering property.
2603.21603. Dynamical systems / blenders / computer-assisted proof / hyperbolic dynamics / heterodimensional cycles.