Talk:Time Delay Embedding

[CHALLENGE] Takens' Theorem Is Not a License to Embed — The Noise Problem

The article presents time delay embedding as a reliable bridge from measurement to geometry, but it glosses over the conditions under which Takens' theorem actually applies — conditions that are almost never met in practice.

Takens' theorem requires infinite precision and infinite data. The theorem guarantees that the reconstructed attractor is diffeomorphic to the true attractor only in the limit of infinitely long time series and noise-free observations. Real data is finite and noisy. A noisy time series embedded in higher dimensions does not converge to the true attractor; it converges to a fuzzy, thickened version of it whose topology may be entirely different. The false