ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let T ⁣:M→MT\colon M \to M be a smooth diffeomorphism of a compact Riemannian manifold MM with a compact TT-invariant attractor X⊂MX \subset M and a natural measure μ\mu on XX of information dimension ID⁡(μ)=D\operatorname{ID}(\mu)=D. Fix a generic observable h ⁣:M→Rh\colon M \to \mathbb R, k∈Nk\in\mathbb N, and a small δ>0\delta>0. Let ϕh\phi_h be the kk-delay coordinate map corresponding to hh, and let σh,ε\sigma_{h,\varepsilon} denote the associated prediction error for sufficiently small ε>0\varepsilon>0. Schroer–Sauer–Ott–Yorke conjecture. The following hold: (i) if k<Dk<D, then

μ({x∈X:σh,ε(ϕh(x))>δ})≥C\mu\bigl(\{x\in X:\sigma_{h,\varepsilon}(\phi_h(x))>\delta\}\bigr)\geq C

for some C>0C>0; (ii) if D<k<2DD<k<2D and ϕh\phi_h is not injective, then

C1εk−D≤μ({x∈X:σh,ε(ϕh(x))>δ})≤C2εk−DC_1\varepsilon^{k-D}\leq\mu\bigl(\{x\in X:\sigma_{h,\varepsilon}(\phi_h(x))>\delta\}\bigr)\leq C_2\varepsilon^{k-D}

for some C1,C2>0C_1,C_2>0; (iii) if D<k<2DD<k<2D and ϕh\phi_h is injective, then

μ({x∈X:σh,ε(ϕh(x))>δ})=0;\mu\bigl(\{x\in X:\sigma_{h,\varepsilon}(\phi_h(x))>\delta\}\bigr)=0;

(iv) if k>2Dk>2D, then

μ({x∈X:σh,ε(ϕh(x))>δ})=0.\mu\bigl(\{x\in X:\sigma_{h,\varepsilon}(\phi_h(x))>\delta\}\bigr)=0.

The conjecture predicts how the mass of points with prediction error exceeding δ\delta depends on the number of delay coordinates relative to the information dimension. It was originally formulated for natural measures on attractors of smooth diffeomorphisms; the supplied text gives no evidence that it has been resolved.

Record

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.