Back to search
2006.13429

A Higher Order Unscented Transform

Deanna Easley, Tyrus Berry

correctmedium confidence
Category
math.DS
Journal tier
Specialist/Solid
Processed
Sep 28, 2025, 12:55 AM

Audit review

The paper’s Theorem 3.1 proves linear convergence of the deflation scheme with rate r = sqrt(1 − c^2/d^k), using the eigenpair identity ‖T − λ v^{⊗k}‖_F^2 = ‖T‖_F^2 − λ^2 (Lemma 2.14) and the entrywise-to-eigenvalue bound λ ≥ (c/d^{k/2})‖T‖_F derived from the assumption λ ≥ c|T_{i_1…i_k}| for all entries; see Lemma 2.14 and Theorem 3.1 in the uploaded paper . The candidate solution applies exactly the same steps: (i) sum-of-squares to get λ_ℓ^2 ≥ (c^2/d^k)‖T_ℓ‖_F^2; (ii) plug into the identity to obtain a uniform contraction; and (iii) telescope to get T = ∑_{ℓ=1}^L λ_ℓ v_ℓ^{⊗k} + E_L with ‖E_L‖_F = O(r^L). The paper also supplies the needed entrywise lower bounds on λ for orders 3 and 4 (Lemma 3.2, Appendix), confirming the assumption that underpins the contraction . Hence both arguments are correct and essentially the same.

Referee report (LaTeX)

\textbf{Recommendation:} minor revisions

\textbf{Journal Tier:} specialist/solid

\textbf{Justification:}

The theorem and its proof are concise, correct, and provide a useful linear-convergence guarantee for a practical tensor deflation method. The contribution is specific but pertinent to rank-1 tensor approximation and to the paper’s HOUT construction. Minor clarifications about the scope of the entrywise bound and its constants would enhance readability and rigor.