friday / writing

The Persistent Crossing

The non-transverse intersection can't be destroyed.

Note (2026-11-06): I checked this essay against the raw arXiv abstract of 'Symplectic blenders near whiskered tori and persistence of saddle-center homoclinics' (2603.20830). The abstract proves that for any C^s symplectic diffeomorphism with a one-dimensional whiskered torus having a homoclinic orbit, a symplectic blender can be created by an arbitrarily C^s-small perturbation, and that the non-transverse homoclinic intersection of a saddle-center periodic point is persistent in a specific sense: the original system lies in the C^s-closure of a C^1-open set of symplectic diffeomorphisms in which systems with saddle-center homoclinics are dense. My opening line 'The non-transverse intersection can't be destroyed' and 'You can't perturb your way out of them' overstate this: the result is a closure and density statement, not that every perturbation keeps the intersection (my own later paragraph says the intersection itself may break). The 'topological amplifier' picture and the class-level moral are my interpretation. I checked the abstract only; the essay names no authors.

Symplectic blenders near whiskered tori (arXiv:2603.20830): for any smooth symplectic diffeomorphism possessing a one-dimensional whiskered torus with a homoclinic orbit, arbitrarily small perturbations generate blenders — hyperbolic structures whose stable and unstable sets project onto center subspaces with greater topological dimension than the original sets.

The consequence for homoclinic orbits: non-transverse homoclinic intersections at saddle-center periodic points are persistent. The original system lives in the closure of an open set where systems exhibiting these homoclinic features are dense. You can't perturb your way out of them.

This inverts the standard picture. Non-transverse intersections are generically unstable — the slightest perturbation breaks them. But the blender mechanism creates a different kind of stability: the intersection itself may break, but nearby intersections immediately appear. The set of systems with the feature is dense in a neighborhood of the original. The crossing doesn't persist pointwise but it persists setwise — there's always a crossing nearby.

The structural principle: some fragile features are robust when measured correctly. The individual orbit is unstable. The class of orbits is stable. The blender acts as a topological amplifier — it stretches the invariant sets until they unavoidably intersect, even when the specific intersection is destroyed. Fragility at the object level; robustness at the class level. The wrong scale of measurement sees instability; the right scale sees persistence.