Pith. sign in

REVIEW 1 major objections 6 minor 2 cited by

Quasitubal Tensor Algebra Over Separable Hilbert Spaces

T0 review · 1 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For tubal tensors with tubes in a separable Hilbert space, truncating the quasitubal SVD at rank q gives the optimal rank-q approximation, with error tending to zero as q grows.

desk verdict Solid construction and a credible q-SVD theory, but the proof of the central finite-rank optimality theorem (Theorem 7.1) has a real gap: the image of T_Y is p·rank(Y)-dimensional, not rank(Y)-dimensional, and the combinatorial matching step is unsound as written. read the letter →

arxiv 2504.16231 v2 pith:26PXI3OZ submitted 2025-04-22 math.NA cs.NAmath.FA

Uria Mor , Haim Avron This is my paper · ORCID
classification math.NAcs.NAmath.FA MSC 15A6946L0546L08
keywords tubaltensorsquasitubalalgebraseparableHilbertspacetensorSVDEckart-Youngoptimalitylow-rankapproximationC*-algebrainfinite-dimensional
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper extends the tubal-tensor framework—treating tensors as matrices of short vectors ('tubes')—to tubes drawn from an infinite-dimensional separable Hilbert space. Because no such Hilbert space can carry a multiplicative identity, the paper embeds the tube space in a larger commutative unital C*-algebra of bounded operators, the quasitubal algebra, and develops a quasitubal singular value decomposition (q-SVD) for tensors over this algebra. The central result is an Eckart–Young theorem: for any such tensor $X$, the explicit rank-$q$ truncation $X[q]$ built from the q-SVD minimizes the Hilbert-space error $\|X-Y\|_H$ among all rank-at-most-$q$ approximations, and the error $\|X-X[q]\|_H$ converges to zero as $q\to\infty$. This gives a principled way to approximate genuinely infinite-dimensional, function-valued tensors by finitely representable tubal tensors.

What carries the argument

The key object is the quasitubal algebra $H^*$, the commutative unital C*-algebra of bounded $H$-linear operators on a separable Hilbert space $H$, isometrically *-isomorphic to $\ell^\infty$ via the $\pi$-transform. The tubal product $\star_F$ is defined by an orthonormal basis (an isometric isomorphism $F:H\to\ell^2$), and quasitubes extend tubes by allowing bounded diagonal multipliers in the transform domain; this supplies the missing multiplicative identity and gives a notion of positivity, order, and square roots in $H^*$. The proof machinery then runs through the q-SVD: applying a pointwise matrix SVD to each frontal slice in the transform domain yields unitary quasitubal tensors and ordered nonnegative singular quasitubes, and the Eckart–Young argument compares any rank-$q$ competitor against the explicit truncation $X[q]$ using the orthonormal basis induced by the q-SVD.

What would settle it

Take $H=L^2([-1,1])$ with the Chebyshev basis, define a tubal tensor $X$ with continuously parameterized entries, compute $X[q]$ by Eq. (36) for some $q$, and run a numerical search over all rank-$q$ function-valued tubal tensors $Y$ (for example by optimizing the $q$ rank-one terms directly); if any $Y$ achieves $\|X-Y\|_H<\|X-X[q]\|_H$, Theorem 7.1 is false. A cheaper necessary check is whether the ordered singular coefficients $\sigma_n$ used in Eq. (36) are always non-increasing and square-summable, since the convergence claim depends on that ordering.

Watch

Extended reading notes

Core claim

The paper's central discovery is that a matrix-mimetic Eckart–Young theory exists for tubal tensors over separable Hilbert spaces even though the tube space with the tubal product is not unital, once the tube space $H$ is embedded as a two-sided *-ideal in $H^*$, the algebra of bounded $H$-linear operators on $H$, which is isometrically *-isomorphic to $\ell^\infty$. The quasitubal SVD of any $X\in H^{m\times p}_*$ exists (Theorem 5.9), and for $X\in H^{m\times p}$ the explicit rank-$q$ truncation $X[q]$ (Eq. (36)) satisfies $\|X-X[q]\|_H\le\|X-Y\|_H$ for every $Y\in H^{m\times p}$ with $\mathrm{rank}_{F,\star_F}Y\le q$, with $\|X-X[q]\|_H\to 0$ and $\|X-X[q]\|_{\mathrm{op}}\to 0$ as $q\to\infty$ (Theorem 7.1). In other words, the best finite-rank approximation of an infinite-dimensional tubal tensor is obtained by keeping the $q$ largest ordered singular quasitubes of the q-SVD, and these finite-rank truncations converge to the original tensor.

Load-bearing premise

Everything depends on fixing a separable Hilbert space and an orthonormal basis (an isometric transform) that defines how tubes are multiplied; the optimality claims hold only relative to that choice, and the construction breaks down without separability or without the transform being an isometry.

Editorial extensions

If this is right

  • Any tubal tensor with tubes in a separable Hilbert space has a best rank-$q$ approximation in the $H$-norm, given by the explicit truncation of its q-SVD; this extends the finite-dimensional tubal Eckart–Young theorem to infinite dimensions.
  • The rank-$q$ truncation is finitely representable: it is a finite-dimensional tubal tensor embedded in the infinite-dimensional space, so it can be stored and manipulated on a computer.
  • Truncation error vanishes in both the Hilbert-space norm and the operator norm as $q\to\infty$, so infinite-dimensional tensors can be approximated to arbitrary accuracy by finite-rank objects.
  • The framework applies to any separable Hilbert space—for instance spaces of square-integrable functions—so continuous, functional data can be treated with matrix-mimetic tensor operations rather than classical CP-style decompositions.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper fixes an orthonormal basis $F$ that defines the tube multiplication, and its optimality is relative to that choice; a natural extension is an adaptive or data-dependent choice of $F$, which the paper does not address.
  • The construction requires separability and an isometric transform; extending the Eckart–Young statement to non-separable Hilbert spaces or to non-isometric transforms would require a different proof, and the paper leaves this open.
  • Because the tube space $H$ sits inside $H^*$ as an ideal, the q-SVD truncation of a function-valued tensor can be read as a spectral-type approximation of an operator; this may connect to model-order reduction for continuous-time dynamical systems, a direction the paper mentions but does not develop.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

1 major / 6 minor

Summary. This paper develops a quasitubal tensor algebra for tensors whose tubes lie in a separable infinite-dimensional Hilbert space H. Since the natural pointwise-multiplication algebra on H lacks a unit, the authors embed H into the commutative unital C*-algebra H* of bounded H-linear operators (isometrically isomorphic to ℓ∞), define quasitubal tensors as matrices over H*, establish a q-SVD (Theorem 5.9), and prove two Eckart-Young-type optimality results: Theorem 5.11 for multi-rank truncations in the operator norm, and Theorem 7.1 for finite implicit-rank truncations in the H-norm, with convergence of X[q] to X as q → ∞. The paper also discusses computational aspects and gives a numerical demonstration with Chebyshev bases.

Significance. If the results are correct, this is a valuable extension of the tubal tensor framework to infinite-dimensional tubes, with rigorous proofs of an SVD analogue and optimal low-rank approximation. The construction is principled: it follows from the axioms of a separable Hilbert space and an orthonormal basis, and the results are proven rather than assumed. The paper is mostly self-contained and carefully builds the algebraic infrastructure (Hilbert C*-modules, spectral theory) needed for the main theorems. The potential impact is in functional data analysis, operator learning, and dynamical systems, where tensors with continuous modes arise. The main caveat is that the optimality results are relative to a chosen orthonormal basis F, which the authors acknowledge as a limitation; this is not a flaw but a scope restriction.

major comments (1)
  1. [Section 7, proof of Theorem 7.1 (Eq. (39) and the paragraph beginning 'Consider the mapping TY')] The assertion that the q' rank-one tensors {B(k)} form an orthonormal basis for the image of TY(Z)=Y⋆F Z is incorrect. For a single nonzero frontal slice at frequency τ with Y_τ = u v^T, the image of the map Z_τ ↦ Y_τ Z_τ on F^{p×p} has dimension p (all m×p matrices with columns in span(u)), not 1; summing over q' nonzero slices gives image dimension p·q', while only q' basis vectors are listed. Consequently, the subsequent expansion X = Σ α_k B(k) and the claim that a rank-q'<q approximation can be improved by adding α_{q'+1}B(q'+1) are not justified as written. The later slice-wise Eckart-Young and majorization argument for the case q' = q appears sound and independent, but it does not cover the case q' < q. The theorem may still be true, but the proof needs repair or a different reduction before the central optimality claim is fully established.
minor comments (6)
  1. [Introduction, page 2] There are several typos: 'spae' should be 'space', 'seperable' should be 'separable', and 'it’s' should be 'its' in the phrase about the matrix-mimetic SVD.
  2. [Section 3.2, paragraph after Definition 3.8] The word 'quaistubes' appears once and should be 'quasitubes'.
  3. [Remark 3.10] The claim that the set of quasitubes is independent of F is asserted without a forward reference; it is only fully justified by Theorem 4.4, so the reader would benefit from a pointer to that theorem.
  4. [Definition 5.3] The H-norm in Eq. (18) is only defined for tensors in H^{m×p}, but later it is applied to the finite-rank truncations X[q], which indeed lie in H^{m×p}; stating this explicitly at the point of Definition 5.3 would remove a potential source of confusion.
  5. [Proof of Theorem 7.1, combinatorial step] The definition of N_q and the matching argument between indices in [q]\N_q and tail indices n'>q are dense; a short explanatory paragraph would improve readability and verifiability.
  6. [Section 8.1, numerical demonstration] The demonstration would be strengthened by reporting the observed decay of the Chebyshev coefficients and quantifying how the truncation error depends on the choice of the basis F, since the theory leaves this choice free.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 7.1 is proven from explicit Hilbert-space axioms, and the basis choice is an acknowledged modeling assumption rather than a hidden fit.

full rationale

The paper's central claim, Theorem 7.1, is derived from explicit definitions rather than from a quantity that is fitted to the result it predicts. The F-transform, quasitubes as bounded H-linear operators, the q-SVD obtained slice-wise from matrix SVDs, and the rank notion as the dimension of the image of the associated multiplication map are all stated assumptions or constructed objects. The singular values used to form X[q] come from the SVD of X itself, and the optimality statement is an analytic Eckart-Young-type argument, not a tautology. The only user choices, the separable Hilbert space H, its inner product, and the orthonormal basis F, are explicitly acknowledged as defining the algebra; Remark 3.9 and the conclusion note that different bases produce different tubal products and hence different optimal truncations. This is basis-dependence, not circularity. Self-citations to the finite-dimensional tubal literature are used as background context and definitions, not as a substitute for the infinite-dimensional proofs. Any concern about the correctness of the proof of Theorem 7.1, such as the dimension of the image of the map Z -> Y ⋆_F Z, is a mathematical gap or repair issue, not a circularity, and the later slice-wise majorization argument is independent of the contested basis claim. The numerical demonstration is an illustration of the theory, not a fitted prediction used to prove the theorem. Thus no circular step is present.

Assumptions & free parameters 2 free parameters · 7 assumptions · 2 invented entities

The paper's claims rest on standard functional analysis and C*-algebra results, plus the explicit modeling choice of a separable Hilbert space with a fixed orthonormal basis. No numerical constants are fitted to data. The only user-chosen objects are the Hilbert space inner product and the orthonormal basis defining the tubal product; these are structural assumptions, not hidden free parameters.

free parameters (2)
  • Orthonormal basis F (isometric isomorphism H → ℓ₂) = User-specified (Fourier, Chebyshev, etc.)
    Defines the tubal product ⋆_F via pointwise multiplication in the transform domain. All structural and optimality results (Theorems 5.9, 5.11, 7.1) are relative to this choice. The paper proves the quasitube space H* is independent of F (Remark 3.10), but the product and the singular value ordering depend on F.
  • Inner product on H = User-specified (e.g., L² inner product, weighted Chebyshev inner product in Section 8.1)
    The Hilbert norm ∥·∥_H used in the Eckart-Young result (Theorem 7.1) depends on the chosen inner product. Changing the inner product changes the orthonormal basis and hence the algebra.
assumptions (7)
  • standard math Gelfand-Naimark theorem: every C*-algebra is *-isomorphic to a norm-closed *-subalgebra of B(H) for some Hilbert space H.
    Used in Section 4 to justify calling adjoints in abstract C*-algebras, and to motivate the H* construction.
  • standard math Spectral radius formula lim ||a^n||^{1/n} = r(a) in Banach algebras, and ∥a∥ = r(a) for self-adjoint elements in C*-algebras.
    Used in Lemma 6.2 and in proving Lemma 6.3 and positivity results.
  • standard math Uniqueness of norm on a C*-algebra (Lemma 6.3).
    Used to show the H* norm is determined by the algebra structure.
  • standard math Hilbert C*-module Cauchy-Schwarz inequality (Lance Proposition 1.1).
    Used in Lemma 6.14.
  • standard math Matrix SVD and Eckart-Young theorem over F.
    Applied pointwise to frontal slices in proofs of Theorems 5.9, 5.11, and 7.1.
  • domain assumption Separability of H and existence of a countable orthonormal basis.
    The entire framework requires H separable so that F: H → ℓ₂ is an isometric isomorphism (Definition 3.1). Non-separable Hilbert spaces are explicitly left for future work.
  • domain assumption The tube space is a Hilbert space over F, and tubal multiplication is defined via a fixed orthonormal basis F.
    The product ⋆_F in Eq. (4) is the core definition. Without it, H* and all theorems do not exist. The choice of F is a modeling assumption, flagged in Remark 3.9 and the conclusion.
invented entities (2)
  • Quasitubes (H*, the dual module of H over itself) independent evidence
    purpose: Provides a commutative unital C*-algebra containing H as a *-ideal, enabling a multiplicative identity and thus quasitubal SVD and Eckart-Young results.
    The construction is shown to be isometrically *-isomorphic to ℓ∞ (Theorem 4.4), so it is not an arbitrary postulate. Its utility is demonstrated by the q-SVD existence theorem and the optimality theorems. The entity is a mathematical definition, not a physical object; the 'independent evidence' is the internal theorem structure and the numerical illustration.
  • Quasitubal tensors (H^{m×p}_*) independent evidence
    purpose: Infinite-dimensional analogue of tubal tensors, matrices of quasitubes, supporting unitary tensors and SVD.
    Theorems 5.9 and 7.1 establish the decomposition and optimality. The entity is well-defined via the π-transform to ℓ∞.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quasitubal Tensor Algebra Over Separable Hilbert Spaces." pith.science (2026). https://pith.science/paper/26PXI3OZ

@misc{pith2026250416231,
  author       = {Pith},
  title        = {Pith review of: Quasitubal Tensor Algebra Over Separable Hilbert Spaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/26PXI3OZ}},
  note         = {Machine review of arXiv:2504.16231}
}
read the original abstract

The tubal tensor framework provides a clean and effective algebraic setting for tensor computations, supporting matrix-mimetic features like Singular Value Decomposition and Eckart-Young-like optimality results. Underlying the tubal tensor framework is a view of a tensor as a matrix of finite sized tubes. In this work, we lay the mathematical and computational foundations for working with tensors with infinite size tubes: matrices whose elements are elements from a separable Hilbert space. A key challenge is that existence of important desired matrix-mimetic features of tubal tensors rely on the existence of a unit element in the ring of tubes. Such unit element cannot exist for tubes which are elements of an infinite-dimensional Hilbert space. We sidestep this issue by embedding the tubal space in a commutative unital C*-algebra of bounded operators. The resulting quasitubal algebra recovers the structural properties needed for decomposition and low-rank approximation. In addition to laying the theoretical groundwork for working with tubal tensors with infinite dimensional tubes, we discuss computational aspects of our construction, and provide a numerical illustration where we compute a finite dimensional approximation to a infinitely-sized synthetic tensor using our theory. We believe our theory opens new exciting avenues for applying matrix mimetic tensor framework in the context of inherently infinite dimensional problems.

Figures

Figures reproduced from arXiv: 2504.16231 by the authors.

Figure 1
Figure 1. Graphical illustration of tubal-tensor as a matrix of tubes. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Diagram of the mappings and embeddings between [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Description of the original families of multidimensional curves. [PITH_FULL_IMAGE:figures/full_fig_p030_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Approximation quality as a function of truncation size. [PITH_FULL_IMAGE:figures/full_fig_p031_4.png]
Figure 5
Figure 5. Figure 5: Three dimensional curves, corresponding to the first 3 components in the reconstructed signal, [PITH_FULL_IMAGE:figures/full_fig_p031_5.png]
Figure 6
Figure 6. Figure 6: A scatter plot of the first and fourth components of the domain transform, i.e., [PITH_FULL_IMAGE:figures/full_fig_p032_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Exact Symmetry as Algebra: A Machine-Verified Tensor Calculus that Enforces Physical Selection Rules

    cs.LG 2026-05 unverdicted novelty 7.0 of 10 partial

    A group-defined tensor algebra makes equivariance exact, provably optimal low-rank compression, and recovers Wigner–Eckart selection rules from molecular features.

  2. Matrices over a Hilbert space and their low-rank cross approximation

    math.NA 2025-05 conditional novelty 5.0 of 10

    Bochner matrices (matrices with Hilbert-space entries) admit cross decompositions and a new adaptive cross-dyadic approximation algorithm that numerically approximates parametric PDE solution maps.

Reference graph

Works this paper leans on

24 extracted references · 22 canonical work pages · cited by 2 Pith papers

  1. [1]

    Battles and L

    Z. Battles and L. N. Trefethen. An extension of matlab to continuous functions and operators. SIAM Journal on Scientific Computing , 25(5):1743–1770, Jan. 2004

    work page 2004

  2. [2]

    K. Braman. Third-Order Tensors as Linear Operators on a Space of Matrices. Linear Algebra and its Applications, 433(7):1241–1253, 12 2010

    work page 2010

  3. [3]

    J. B. Conway. A Course in Functional Analysis . Springer New York, 2007

    work page 2007

  4. [4]

    R. Han, P. Shi, and A. R. Zhang. Guaranteed functional tensor singular value decomposition. Journal of the American Statistical Association , pages 1–13, Feb. 2023

    work page 2023

  5. [5]

    C. R. Harris, K. J. Millman, S. J. van der Walt, R. Gommers, P. Virtanen, D. Cournapeau, E. Wieser, J. Taylor, S. Berg, N. J. Smith, R. Kern, M. Picus, S. Hoyer, M. H. van Kerkwijk, M. Brett, A. Haldane, J. F. del R´ ıo, M. Wiebe, P. Peterson, P. G´ erard-Marchant, K. Sheppard, T. Reddy, W. Weckesser, H. Abbasi, C. Gohlke, and T. E. Oliphant. Array progra...

    work page 2020

  6. [6]

    P. Jin, S. Meng, and L. Lu. Mionet: Learning multiple-input operators via tensor product. SIAM Journal on Scientific Computing , 44(6):A3490–A3514, Nov. 2022

    work page 2022

  7. [7]

    Kernfeld, M

    E. Kernfeld, M. Kilmer, and S. Aeron. Tensor–tensor products with invertible linear transforms. Linear Algebra and its Applications , 485:545–570, Nov. 2015

    work page 2015

  8. [8]

    M. E. Kilmer, K. Braman, N. Hao, and R. C. Hoover. Third-Order Tensors as Operators on Matrices: A Theoretical and Computational Framework with Applications in Imaging.SIAM Journal on Matrix Analysis and Applications, 34(1):148–172, 2 2013

    2013

Show all 24 references
  1. [9]

    M. E. Kilmer, L. Horesh, H. Avron, and E. Newman. Tensor-tensor algebra for optimal repre- sentation and compression of multiway data. Proceedings of the National Academy of Sciences , 118(28):e2015851118, July 2021

    2021

  2. [10]

    M. E. Kilmer and C. D. Martin. Factorization strategies for third-order tensors. Linear Algebra and its Applications, 435(3):641–658, 8 2011

    2011

  3. [11]

    M. E. Kilmer, C. D. Martin, and L. Perrone. A third-order generalization of the matrix svd as a product of third-order tensors, 2008

    2008

  4. [12]

    T. G. Kolda and B. W. Bader. Tensor decompositions and applications. SIAM Review, 51(3):455–500, Aug. 2009

    2009

  5. [13]

    J. N. Kutz, S. L. Brunton, B. W. Brunton, and J. L. Proctor. Dynamic Mode Decomposition: Data- Driven Modeling of Complex Systems . Society for Industrial and Applied Mathematics, Nov. 2016

    2016

  6. [14]

    E. C. Lance. Hilbert C*-Modules: A Toolkit for Operator Algebraists . Cambridge University Press, Mar. 1995

    1995

  7. [15]

    B. W. Larsen, T. G. Kolda, A. R. Zhang, and A. H. Williams. Tensor decomposition meets RKHS: Efficient algorithms for smooth and misaligned data, 2024

    2024

  8. [16]

    X. Mao, A. Dong, Z. He, Y. Mei, S. Mei, R. Wang, and C. Chen. Data-driven analysis of t-product- based dynamical systems. IEEE Control Systems Letters , 8:3356–3361, 2024

    2024

  9. [17]

    Miron, Y

    S. Miron, Y. Zniyed, R. Boyer, A. Lima Ferrer de Almeida, G. Favier, D. Brie, and P. Comon. Tensor methods for multisensor signal processing. IET Signal Processing, 14(10):693–709, Dec. 2020

    2020

  10. [18]

    U. Mor, Y. Cohen, R. Vald´ es-Mas, D. Kviatcovsky, E. Elinav, and H. Avron. Dimensionality reduc- tion of longitudinal ’omics data using modern tensor factorizations. PLOS Computational Biology , 18(7):e1010212, July 2022. 33

    2022

  11. [19]

    P. F. Shustin and H. Avron. Semi-infinite linear regression and its applications. SIAM Journal on Matrix Analysis and Applications , 43(1):479–511, Mar. 2022

    2022

  12. [20]

    Congratulations to the 2025 SIAG/CSE Best Paper Prize recipients

    SIAM-news. Congratulations to the 2025 SIAG/CSE Best Paper Prize recipients. — SIAM — siam.org. https://www.siam.org/publications/siam-news/articles/ 2025-february-prize-spotlight/#Avron-et-al , Feb. 2025. [Accessed 20-04-2025]

    2025

  13. [21]

    N. D. Sidiropoulos, L. De Lathauwer, X. Fu, K. Huang, E. E. Papalexakis, and C. Faloutsos. Tensor decomposition for signal processing and machine learning. IEEE Transactions on Signal Processing, 65(13):3551–3582, July 2017

    2017

  14. [22]

    G. W. Stewart. Afternotes Goes to Graduate School. Society for Industrial and Applied Mathematics, Jan. 1998

    1998

  15. [23]

    L. N. Trefethen. Householder triangularization of a quasimatrix. IMA Journal of Numerical Analysis, 30(4):887–897, Aug. 2009

    2009

  16. [24]

    Virtanen, R

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. May- orov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, I. Polat, Y. Fen...

    2020

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.