Pith. sign in

REVIEW 2 major objections 4 minor 32 references

Almost invariant subspaces of shift operators and products of Toeplitz and Hankel operators

T0 review · 2 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read For the Hardy-space shift, almost-invariant subspaces of the forward and backward operators are the same.

desk verdict Strong, useful paper with a real technical gap in the unbounded Toeplitz case; the backward/forward equivalence is new and worth citing once the domain issues are fixed. read the letter →

arxiv 2411.13177 v1 pith:FC65TWTK submitted 2024-11-20 math.FA

Caixing Gu , In Sung Hwang , Hyoung Joon Kim , Woo Young Lee , Jaehui Park This is my paper · ORCID
classification math.FA MSC 47A1547B3547B38
keywords almostinvariantsubspacesbackwardshiftforwardToeplitzoperatorsHankelHardyspacesmodelpartialisometries
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

Almost invariant subspaces are closed subspaces that a shift maps into themselves up to a finite-dimensional error. This paper determines what they are for the shift operator and its adjoint on vector-valued Hardy spaces $H^2_E$. It proves that the two questions have the same answer: a subspace is almost invariant for the backward shift if and only if it is almost invariant for the forward shift. The proof gives an explicit catalogue of such subspaces as ranges of Toeplitz operators composed with model-space projections, expressed through products of Toeplitz and Hankel operators. It also identifies every finite-rank perturbation of the shift that makes such a subspace genuinely invariant.

What carries the argument

The engine is the algebra of Toeplitz and Hankel operators on vector-valued Hardy spaces, together with the identities (4) and (5), for example $S^*_F T_\Phi = T_\Phi S^*_{E_1} + S^*_F T_\Phi P_{E_1}$. These turn the action of a shift on a range into the same range plus a finite-rank term, which is exactly the almost-invariance condition. The model space $K_\Theta = H^2_{E_1} \ominus \Theta H^2_E$ enters through the projection $I_{E_1} - T_\Theta T^*_\Theta$; the representation $M = R(T_\Phi(I_{E_1} - T_\Theta T^*_\Theta))$ expresses $M$ as the image of a model space under a Toeplitz operator. The partial-isometry condition (9), namely $[T_\Phi(I_{E_1} - T_\Theta T^*_\Theta)]^*[T_\Phi(I_{E_1} - T_\Theta T^*_\Theta)] = I_{E_1} - T_\Theta T^*_\Theta$, encodes the requirement that this Toeplitz operator act isometrically on $K_\Theta$, making the range closed and the parametrization one-to-one.

What would settle it

For a scalar inner $\theta$ of infinite Blaschke product and $\phi \in H^2 \setminus H^\infty$, check identity (4) on the dense domain of analytic polynomials: if $S^* T_\phi p = T_\phi S^* p + S^* T_\phi P_{\mathbb C} p$ fails for some polynomial $p$, the unbounded-symbol step behind Theorem 3.8 breaks. Alternatively, test whether $T_\phi(I - T_\theta T^*_\theta)$ satisfies the algebraic isometry condition (9) on polynomials yet fails to be a bounded partial isometry on $H^2$; that failure would refute the sufficiency direction as stated.

Watch

Extended reading notes

Core claim

The central result is Theorem 3.8 together with Corollary 3.11. A closed subspace $M$ of $H^2_F$ is $S^*_F$-almost invariant exactly when either $M = R(T_\Theta)$ for an inner function $\Theta \in H^\infty_{\mathcal B(E,F)}$, or $M = R(T_\Phi(I_{E_1} - T_\Theta T^*_\Theta))$ for an inner pure $\Theta \in H^\infty_{\mathcal B(E,E_1)}$, $\Phi \in H^2_{\mathcal B(E_1,F)}$, $\dim E_1 < \infty$, and $T_\Phi(I_{E_1} - T_\Theta T^*_\Theta)$ a partial isometry. The same two forms characterize the $S_F$-almost invariant subspaces, so almost invariance for the forward shift and for the backward shift coincide. This answers the paper's motivating question about the almost invariant subspaces of the shift operator completely, and it implies that every such subspace is almost reducing. The proof obtains these forms by passing the known characterizations of nearly invariant subspaces with finite defect through range identities for Toeplitz and Hankel operators.

Load-bearing premise

The main theorems depend on extending algebraic identities for Toeplitz operators with bounded symbols to Toeplitz operators whose symbols are only square-integrable analytic functions, and the paper does not provide the needed domain and continuity argument, remarking only that 'the justification should be easy'.

Editorial extensions

If this is right

  • Question 1.3 is settled: the almost invariant subspaces of the forward shift $S_F$ are exactly the list in Theorem 3.8, and each such subspace is also almost invariant for $S^*_F$.
  • Every $S_F$-almost invariant subspace is $S_F$-almost reducing, so its orthogonal complement shares the property with the same defect.
  • For $M = R(T_\Theta)$ the minimal defect is $\dim E - \operatorname{rank}(U)$, where $U$ is the unitary part of $\Theta$; for the model-space form the minimal defect is the orthogonal complement of $M$ inside $R(S^*_F T_\Phi P_{E_1})$.
  • Theorem 5.10 lists all finite-rank $T_0$ for which $M$ is $(S_F+T_0)$-, $(S^*_F+T_0)$-, or reducibly invariant, making Lemma 5.1 explicit.
  • When $\Phi$ is inner, $M = R(T_\Phi(I - T_\Theta T^*_\Theta))$ is a half-space exactly when $\Theta$ is not a finite Blaschke-Potapov product.

Reading between the lines

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

  • If the unbounded-symbol extension is supplied, the same range-identity method should characterize almost invariant subspaces for other operators built from Toeplitz or Hankel algebras, such as shifts on weighted Hardy spaces.
  • Corollary 4.5 realizes these subspaces as reproducing kernel Hilbert spaces with kernels built from $\Phi$ and $\Theta$, suggesting that defect spaces could be read off from kernel asymptotics instead of ranges.
  • The forward/backward coincidence is a genuinely almost-invariant phenomenon: with exact invariance the two families are essentially disjoint, so the finite defect is what buys the symmetry.
  • Conjecture 4.3 could be tested numerically on finite Blaschke-Potapov products of increasing degree to see whether 'half-space' is equivalent to 'infinite-dimensional model space' for non-inner $\Phi$.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The paper studies almost invariant subspaces of the forward and backward shift operators on vector-valued Hardy spaces H^2_F. The main results, stated as Theorem 3.8, Theorem 3.12 and Corollary 3.11, characterize the S*_F-almost invariant subspaces as ranges of Toeplitz-type operators, namely M = R(T_Θ) for an inner Θ, or M = R(T_Φ(I - T_ΘT*_Θ)) with Θ inner and pure, Φ ∈ H^2, and T_Φ(I - T_ΘT*_Θ) a partial isometry. The authors then prove the surprising equivalence that a closed subspace is S*_F-almost invariant if and only if it is S_F-almost invariant. The paper also discusses examples, reproducing kernels, and finite-rank perturbations of the shift that leave such subspaces invariant. The approach is a reformulation of known classification theorems due to Chalendar-Chevrot-Partington, Chalendar-Gallardo-Partington, Chattopadhyay-Das-Pradhan, and O'Loughlin, phrased in terms of ranges of Toeplitz and Hankel operators.

Significance. If the results are established, the paper provides a clean operator-range reformulation of almost invariant subspaces, avoids the machinery of nearly invariant subspaces, and gives explicit defect spaces. The equivalence between S*_F- and S_F-almost invariance is a striking structural fact that goes beyond the scalar examples previously in the literature. The exposition is generally clear and the paper makes good use of existing Beurling-Lax-Halmos and Hitt-Sarason theory. The central classification is not circular: it relies on external classification theorems, and the self-cited Lemma 3.7 is used only in the auxiliary Corollary 3.9, not in the main theorem. However, the main theorems currently rest on an insufficiently justified use of unbounded Toeplitz operators, and one assertion in the proof of Theorem 3.8 is false as stated. These issues are technical rather than conceptual, and appear fixable, but they are load-bearing for the stated H^2 generality.

major comments (2)
  1. [§3, Theorem 3.8 and Theorem 3.12] The proofs of Theorem 3.8 and Theorem 3.12 apply the algebraic operator identities (4), (9)–(11) and (13) to Toeplitz operators T_Φ with Φ ∈ H^2, which are generally unbounded. The sentence immediately before Theorem 3.8, "We have to use unbounded Toeplitz operators, but the justification should be easy," is not a proof. In the scalar Hitt case (Theorem 3.2(ii)), Φ = g need not belong to H^∞, so T_Φ is not a bounded operator on H^2, and the composition T_Φ(I − T_ΘT*_Θ) is not automatically defined. Equations (10) and (13), as well as the isometricity condition (9), are used as operator identities on all of H^2_{E1}. Without a dense-domain or continuity argument establishing these identities for the specific unbounded symbols arising from Theorem 3.5, the 'if' and 'only if' directions of Theorem 3.8(2), Corollary 3.11, and Theorem 3.12(2) are not established in the stated H^2 generality. A rigorous treatment—for example, defining T_Φ(I − T_ΘT*_Θ) on K_Θ, proving it is a bounded partial isometry, and verifying the commutation relations on a dense set—is required.
  2. [§3, proof of Theorem 3.8] The proof asserts that the matrix function Φ = [G_0 z g_1 ⋯ z g_p] is inner. This assertion is false in general. In the scalar Hitt representation (Theorem 3.2(ii)), M = T_g K_θ with g of unit norm and |g|^2 + |y|^2 = 1 for a nonzero admissible function y; such a g is not inner. Consequently T_Φ need not be bounded, which reinforces the unbounded-operator issue raised above. The partial-isometry conclusion should be derived from the norm identity in Theorem 3.5, which gives that T_Φ acts isometrically on K_Θ, not from the false claim that Φ is inner.
minor comments (4)
  1. [§3, Theorem 3.8 and Corollary 3.10] Several displayed formulas for the backward shift of a Toeplitz range are incorrect by a factor of z. For example, S*_F T_Θ P_E applied to a constant e equals (Θ(z) − Θ(0))e / z, not z[Θ(z) − Θ(0)]e. The same factor error appears in Corollary 3.10 in the formula for S*_F h(z). The intended arguments still go through with the corrected expression, but the formulas as written are mathematically false.
  2. [§2, Theorem 2.6] The proof of Theorem 2.6 treats separately the cases where all summands A_i contain an even number of Hankel factors and where all contain an odd number. The theorem statement allows mixed parity in the sum, and the proof does not justify that case. The statement should be restricted to uniform parity, or a separate argument for mixed sums should be supplied.
  3. [§2, Remark 2.3] Remark 2.3 claims that Lemma 2.2 also holds when E and F are infinite dimensional. However, the defect bound in Lemma 2.2 is ≤ dim E_1; if E_1 is infinite dimensional, the finite-rank correction G need not have finite rank, so R(T_ΦH_Ψ)^- need not be almost invariant in the sense of Definition 1.1. The remark is misleading as stated.
  4. [§4, Corollary 4.5] The reproducing kernel formulas (17) and (18) use the denominator 1 − zw. With the standard Hardy-space reproducing kernel k_w(z) = (1 − \bar{w}z)^{-1}, the denominator should be 1 − \bar{w}z or, equivalently, 1 − z\bar{w}. Please correct or clarify the convention.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the main operator-range classifications are derived from credited external nearly-invariant subspace theorems, and the only self-citation is peripheral.

full rationale

The central derivation is not circular. Theorem 3.8 and Corollary 3.11 are obtained by taking the external classification of nearly S*-invariant subspaces with finite defect (Theorem 3.5 and Corollary 3.6, credited to [9], [10], [8], and [23]), rewriting representations of the form M = {G0 k0 + z sum gi ki} as operator ranges M = R(TPhi(I - TTheta T*Theta)), and then applying algebraic identities (4), (5), and (13) together with the orthogonal-complement symmetry in Lemma 2.9. No parameter is fitted to data, and no target conclusion is used as a hypothesis. The 'surprising' equivalence that S* almost invariance equals S almost invariance (Corollary 3.11) follows by applying the already-proved forward direction to M-perp and using Lemma 2.9; it is a genuine derivation, not a definitional restatement. The only self-citation is Lemma 3.7 from [12] (Curto-Hwang-Lee, with two of the present authors), used in Corollary 3.9 to rewrite KTheta as R(H~Phi). This is not load-bearing for the main theorems: Theorem 3.8 uses the projection I - TTheta T*Theta directly, and Corollary 3.11 does not use Lemma 3.7. The paper's flagged technical limitation, 'We have to use unbounded Toeplitz operators, but the justification should be easy' (before Theorem 3.8), is an omitted domain and continuity argument for identities such as (9)-(13) with Phi in H^2; that is a correctness risk, not a circular reduction. Likewise, the proof's assertion that the constructed matrix Phi is inner is not justified in the scalar Hitt case, but even if that assertion fails, the issue is technical validity, not circularity. Accordingly, the circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the prior nearly S*-invariant subspace classification by other authors, standard model space theory, a standard Toeplitz-Hankel identity, and an unproved extension of algebraic identities to unbounded Toeplitz operators. There are no fitted parameters or newly invented mathematical objects.

assumptions (5)
  • domain assumption Theorem 3.5: classification of nearly S*-invariant subspaces with finite defect in H^2_{C^m}
    Used in the only-if direction of Theorem 3.8 to obtain the representation M = T_Phi K_Theta; the present paper does not reprove it.
  • standard math Beurling-Lax-Halmos theorem: S*-invariant subspaces are model spaces K_Theta = H^2_{E1} minus Theta H^2_E
    Used to go from a S*-invariant subspace K in Theorem 3.5 to K = K_Theta.
  • standard math Toeplitz-Hankel identity T_{Omega Psi} - T_Omega T_Psi = H*_{Omega*} H_Psi (equation (1))
    Basis for representing K_Theta and defect spaces via Hankel operators.
  • standard math Decomposition of contractive operator-valued functions into pure and unitary parts (Sz.-Nagy-Foias)
    Allows assuming Theta is pure in Theorem 3.8.
  • ad hoc to paper Algebraic identities (4), (5), (13) extend to unbounded T_Phi with Phi in H^2
    The main theorems require these identities for unbounded Toeplitz operators; the paper states 'the justification should be easy' but gives no domain details.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Almost invariant subspaces of shift operators and products of Toeplitz and Hankel operators." pith.science (2026). https://pith.science/paper/FC65TWTK

@misc{pith2026241113177,
  author       = {Pith},
  title        = {Pith review of: Almost invariant subspaces of shift operators and products of Toeplitz and Hankel operators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FC65TWTK}},
  note         = {Machine review of arXiv:2411.13177}
}
read the original abstract

In this paper we formulate the almost invariant subspaces theorems of backward shift operators in terms of the ranges or kernels of product of Toeplitz and Hankel operators. This approach simplifies and gives more explicit forms of these almost invariant subspaces which are derived from related nearly backward shift invariant subspaces with finite defect. Furthermore, this approach also leads to the surprising result that the almost invariant subspaces of backward shift operators are the same as the almost invariant subspaces of forward shift operators which were treated only briefly in literature.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [9]

    Chalendar, N

    I. Chalendar, N. Chevrot and J.R. Partington, Nearly invariant s ubspaces for backwards shifts on vector-valued Hardy spaces, J. Operator Theory 63 (2010), no. 2, 403-415

    work page 2010

  2. [10]

    Chalendar, E.A

    I. Chalendar, E.A. Gallardo-Gutirrez, and J.R. Partington, A Be urling Theorem for almost-invariant subspaces of the shift operator, J. Operator Theory 83 (2020), no. 2, 321-331

    work page 2020

  3. [8]

    Chattopadhyay, S

    A. Chattopadhyay, S. Das, and C. Pradhan, Almost invariant su bspaces of the shift operator on vector-valued Hardy spaces, Integral Equations Operator Theory 92 (2020), no. 6, Paper No. 49

    work page 2020

  4. [23]

    O’Loughlin, Nearly invariant subspaces with applications to tru ncated Toeplitz operators, Com- plex Analysis and Operator Theory 14 (2020), no

    R. O’Loughlin, Nearly invariant subspaces with applications to tru ncated Toeplitz operators, Com- plex Analysis and Operator Theory 14 (2020), no. 8, Paper No. 86, 24 pages

    work page 2020

  5. [1]

    Androulakis, A.I

    G. Androulakis, A.I. Popov, A. Tcaciuc, and V.G. Troitsky, Almost invariant half-spaces of operators on Banach spaces, Integral Equations Operator Theory 65 (2009), no. 4, 473–484

    work page 2009

  6. [2]

    Benhida and D

    C. Benhida and D. Timotin, Finite rank perturbations of contract ions, Integral Equations Operator Theory 36 (2000), no. 3, 253-268

    work page 2000

  7. [3]

    Beurling, On two problems concerning linear transformations in Hilbert space

    A. Beurling, On two problems concerning linear transformations in Hilbert space. Acta Math. 81 (1949), 239–255

    work page 1949

  8. [4]

    B¨ ottcher and B

    A. B¨ ottcher and B. Silbermann, Analysis of Toeplitz Operators , 2nd edition, Springer-Verlag, 2006

    work page 2006

Show all 32 references
  1. [5]

    Brown and R

    A. Brown and R. G. Douglas, Partially isometric Toeplitz operators , Proc. Amer. Math. Soc. 16 (1965) 681-682

    1965

  2. [6]

    Brown and C

    A. Brown and C. Pearcy, Compact restrictions of operators, Acta Sci. Math. (Szeged) 32 (1971) 271–282

    1971

  3. [7]

    Cˆ amara and T

    M.C. Cˆ amara and T. Ross, The dual of the compressed shift. Canad. Math. Bull. 64 (2021), no. 1, 98-111

    2021

  4. [11]

    Chevrot, Kernel of vector-valued Toeplitz operators, Integral Equations Operator Theory 67 (2010), no

    N. Chevrot, Kernel of vector-valued Toeplitz operators, Integral Equations Operator Theory 67 (2010), no. 1, 57-78

    2010

  5. [12]

    Curto, I

    R.E. Curto, I. S. Hwang and W. Y. Lee, The Beurling–Lax–Halmos theorem for infinite multiplicity, J. Funct. Anal. 280 (2021), no. 6, Paper No. 108884, 101 pp

    2021

  6. [13]

    Fricain and J

    E. Fricain and J. Mashreghi, The theory of H(b) spaces. Vol. 1 and Vol. 2, New Mathematical Monographs, 20 and 21 (2016), Cambridge University Press, Camb ridge

    2016

  7. [14]

    Gu, Multiplcative properties of infinite block Toeplitz and Hanke l matrices, Toeplitz Operators and Random Matrices (In Memory of Harold Widom), 371-399, Oper

    C. Gu, Multiplcative properties of infinite block Toeplitz and Hanke l matrices, Toeplitz Operators and Random Matrices (In Memory of Harold Widom), 371-399, Oper. Theory Adv. Appl. 289, Birkhauser, 2022

    2022

  8. [15]

    C. Gu, D. Kang and J. Park, An inverse problem for kernels of blo ck Hankel operators, Bull. Sci. Math. 171 (2021) Paper No. 103021, 23 pp. 19

    2021

  9. [16]

    Gu and S

    C. Gu and S. Luo, Invariant subspaces of the direct sum of for ward and backward shifts on vector- valued Hardy spaces. J. Funct. Anal. 282 (2022), no. 9, Art. No. 109419, 31 pp

    2022

  10. [17]

    I. Jung, E. Ko, C. Pearcy, Almost invariant half-spaces for op erators on Hilbert space, Bull. Aust. Math. Soc. 97 (2018), no.1, 133-140

    2018

  11. [18]

    Halmos, Shifts on Hilbert spaces

    P. Halmos, Shifts on Hilbert spaces. J. Reine Angew. Math. 208 ( 1961), 102–112

    1961

  12. [19]

    Hitt, Invariant subspaces of H 2 of an annulus, Pacific J

    D. Hitt, Invariant subspaces of H 2 of an annulus, Pacific J. Math. 134 (1988), no. 1, 101-120

    1988

  13. [20]

    Hoffman, Spans and intersections of essentially reducing su bspaces, Proc

    M.J. Hoffman, Spans and intersections of essentially reducing su bspaces, Proc. Amer. Math. Soc. 72 (1978), no.2, 333–340

    1978

  14. [21]

    Lax, Shift invariant spaces

    P. Lax, Shift invariant spaces. Acta Math. 101 (1959), 163–178

    1959

  15. [22]

    Luo and S

    S. Luo and S. Richter, Hankel operators and invariant subspa ces of the Dirichlet space. J. Lond. Math. Soc. (2) 91 (2015), no. 2, 423–438

    2015

  16. [24]

    Peller, Hankel Operators and Their Applications , Springer-Verlag, 2003

    V.V. Peller, Hankel Operators and Their Applications , Springer-Verlag, 2003

    2003

  17. [25]

    Popov and A

    A. Popov and A. Tcaciuc, Every operator has almost-invariant subspaces, J. Funct. Anal. 265 (2013), no. 2, 257–265

    2013

  18. [26]

    Sarason, Nearly invariant subspaces of the backward shift , in Contributions to Operator Theory and its Applications (Mesa, AZ, 1987) , 481-493, Oper

    D. Sarason, Nearly invariant subspaces of the backward shift , in Contributions to Operator Theory and its Applications (Mesa, AZ, 1987) , 481-493, Oper. Theory Adv. Appl., 35, Birkh¨ auser, Basel, 1988, pp. 481–493

    1987

  19. [27]

    Sarason, Kernels of Toeplitz operators, in Toeplitz operators and related topics (Santa Cruz, CA, 1992), Oper

    D. Sarason, Kernels of Toeplitz operators, in Toeplitz operators and related topics (Santa Cruz, CA, 1992), Oper. Theory Adv. Appl., 71, Birkh¨ auser, Basel, 1994, pp. 153– 164

    1992

  20. [28]

    Sirotkin and B

    G. Sirotkin and B. Wallis, The structure of almost-invariant half- spaces for some operators, J. Funct. Anal. 267 (2014), no. 7, 2298–2312

    2014

  21. [29]

    Sirotkin and B

    G. Sirotkin and B. Wallis, Almost-invariant and essentially-invarian t halfspaces, Linear Algebra Appl. 507 (2016), 399-413

    2016

  22. [30]

    Sz.-Nagy, C

    B. Sz.-Nagy, C. Foias, H. Bercovici and L. K´ erchy, Harmonic Analysis of Operators on Hilbert Space , Springer, 2010

    2010

  23. [31]

    Tcaciuc, The invariant subspace problem for rank one pertu rbations, Duke Math

    A. Tcaciuc, The invariant subspace problem for rank one pertu rbations, Duke Math. J. 168 (2019), no. 8, 1539–1550

    2019

  24. [32]

    Timotin, The invariant subspaces of S ⊕ S∗ , Concrete

    D. Timotin, The invariant subspaces of S ⊕ S∗ , Concrete. Operator 7 (2020), 116–123. Caixing Gu Department of Mathematics, California Polytechnic State Universit y, San Luis Obispo, CA 93407, USA E-mail: cgu@calpoly.edu In Sung Hwang Department of Mathematics, Sungkyunkwan Un...

    2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.