Pith. sign in

REVIEW 5 minor 7 references

A counterexample to the Kato conjecture for positive commutators

T0 review · 0 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read A single commutator, i[arctan(P), arctan(Q)], is nonnegative and trace class with trace π/2, disproving the conjectural converse to Kato's positivity criterion.

desk verdict A clean, likely correct counterexample to the Kato conjecture; the two reviewer worries I checked evaporate on reading. read the letter →

arxiv 2608.07805 v1 pith:SRH35XEW submitted 2026-08-07 math.FA math-phmath.MPmath.OAmath.SP

classification math.FAmath-phmath.MPmath.OAmath.SP MSC 47B4747A6047A63
keywords positivecommutatorHowland-KatoproblemKatoclasstraceintegraloperatorconditionallynegativedefiniteFockspaceconjecture
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

This paper disproves the conjectural converse to Kato's positivity criterion for commutators of functions of position and momentum. It proves that the operator i[arctan(P), arctan(Q)] is nonnegative, nonzero, and trace class with trace π/2. Since arctan belongs to the Kato class K_1 but not to any larger strip, and −arctan belongs to no Kato class, the positivity of this commutator shows that a positive commutator need not arise from functions lying in Kato strips. A sympathetic reader should care because this closes a long-standing structural question about which pairs of functions produce positive commutators.

What carries the argument

The central object is the integral operator T on $L^{2}$(−π/2, π/2) with kernel T(θ,φ) = cosθ cosφ F''(θ−φ), where F(t) = ln(t/ sin t). Its even and odd parts control the sign of the auxiliary operator A_s = 2s − B_s, with B_s unitarily equivalent to $s^{{-1}}$T. The key identity is F''(t) = 2∫_0^∞ r $\cosh$(rt)/($e^{{πr}}$−1) dr, which decomposes T into even and odd rank-one integrals and yields T_even ≥ 0, T_odd ≤ 0. Nonnegativity of A_s is then fed into a Fock-space factorization: writing D_s(x,y) = ½∥Φ_x − Φ_y∥² with Φ_x = $A_s^{{1/2}}$ h_x, the kernel of C_s becomes an inner product of exponential vectors, giving C_s = V^*V and trace π/2 by a Hilbert–Schmidt norm computation.

What would settle it

Compute the quadratic form ⟨v, T_odd v⟩ for a smooth odd function v on I with support concentrated near ±π/2, e.g. v(θ) = sinθ cosθ χ(θ) with a cutoff χ, using high-resolution numerical quadrature; if any odd v yields a positive value, Lemma 9's sign assertion would fail. Alternatively, evaluate the double integral of cosθ cosφ F''(θ−φ) v(θ)v(φ) by integrating in the order θ,φ first and compare with the r-integral formula to test whether the interchange in Lemma 9 is valid.

Watch

Extended reading notes

Core claim

The paper's central claim is that the operator C = i[arctan(P), arctan(Q)] is nonnegative and has trace π/2. The proof works with a scaled family C_s = i[arctan(P/s), arctan(Q/s)] and shows that C_s = V^* V, an explicit factorization in a symmetric Fock space built from a positive operator A_s. The nonnegativity of A_s follows from splitting the relevant integral operator T into even and odd parts, proving T_even ≥ 0 and T_odd ≤ 0, and then using the bound ||T_even|| ≤ 0.64 to ensure 2s − $s^{{-1}}$T_even ≥ 0 under the scaling condition. Since arctan is in K_1 but in no larger Kato class and −arctan is in no Kato class, this directly contradicts Conjecture 2, the Kato conjecture as formulated by Herbst and Kriete.

Load-bearing premise

The proof of Lemma 9, which establishes the signs of the even and odd parts of T, relies on interchanging a double integral over I×I with an integral over r; near the endpoints θ,φ → ±π/2 the r-integrand is only conditionally integrable, and the paper does not supply a limiting or truncation argument to justify this interchange.

Editorial extensions

If this is right

  • The Kato conjecture, as formulated by Herbst and Kriete, is false; a positive commutator does not force both functions into Kato strips with product of widths π/2.
  • The explicit operator i[arctan(P), arctan(Q)] provides a concrete, trace-class counterexample with known trace π/2, so the failure is not a marginal or pathological artifact.
  • The more general family i[arctan(αP), arctan(βQ)] is nonnegative whenever 0 < αβ ≤ 2||T_even||^{-1}, and each such operator is trace class with trace π/2, exhibiting a continuum of positive commutators outside the Kato-class regime.
  • The sufficient condition in Kato's theorem, involving membership in Kato classes, is strictly one-way: it describes a class of positive commutators but not all of them.
  • The Fock-space factorization method establishes nonnegativity and trace simultaneously, and it may apply to other kernels of the form e^{-D} where D is conditionally negative definite.

Reading between the lines

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

  • The counterexample suggests that the boundary of the Kato-class condition is sharp in a subtle way: arctan lies exactly on the edge of K_1, and the positive commutator emerges from the borderline behavior rather than from a Kato-class interior.
  • The proof technique, which interprets D_s as a squared Hilbert-space distance, connects the Howland–Kato problem to Schoenberg's theory of positive-definite kernels; similar positivity phenomena may be systematically generated from conditionally negative definite functions.
  • A testable extension is to examine other functions f with f' = (1+x²)^{-1}, such as arctan plus small perturbations, and ask whether the commutator remains nonnegative; the paper's numerics suggest the boundary is not isolated to the pure arctangent.
  • Froese and Herbst's earlier positive result under exponential moment assumptions remains compatible, since arctan's derivative has no exponential moment; the counterexample may be understood as the minimal violation of that moment condition.
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

0 major / 5 minor

Summary. This paper disproves the conjectural converse to Kato's positivity criterion by exhibiting a nonnegative commutator i[arctan(P), arctan(Q)] that is not representable as a Kato-class commutator with strip product pi/2. The proof is self-contained: the kernel of C_s is computed explicitly, expressed via a positive quadratic form in a Fock space, and the trace is evaluated as pi/2. The authors introduce an integral operator T on L^2(-pi/2,pi/2) and prove bounds on its norm; a simple bound ||T|| <= 3pi/5 < 2 suffices for the main theorem, while a sharper bound ||T_even|| <= 0.64 is derived as a refinement. They also determine the maximal Kato strip of the arctangent.

Significance. If correct, the result is a major contribution to operator theory, settling a conjecture that has been open for decades. The construction is explicit and elegant, using classical tools (Schoenberg's theorem, symmetric Fock space, positive-definite kernels) in a novel combination. The paper is fully self-contained: the trace computation is exact, the bounds are analytic and checkable, and no numerical fitting or target-based assumption enters. The weaker bound used for Theorem 1 is elementary, making the central counterexample robust to the more delicate parts of the paper.

minor comments (5)
  1. [Section 1, after Proposition 5] The statement that Theorem 1 is an immediate consequence of Propositions 4 and 5 is confusingly paired with the note that the proof of Theorem 1 is much simpler and does not need Lemma 9. Please state and prove explicitly the weaker version of Proposition 4 with ||T|| in place of ||T_even||, so that the proof of Theorem 1 is self-contained and does not force the reader to extract a variant of the proof of Proposition 4.
  2. [Lemma 9] The proof of Lemma 9 interchanges the r-integral in (21) with the double integral over I x I to obtain the formulas for <v,T_even v> and <v,T_odd v>. For arbitrary v in L^2(I) this interchange is not justified by the standard Fubini theorem directly, since the triple integrand is not absolutely integrable near the corners of I x I. A truncation of the r-integral at R, followed by a limit R -> infinity, would make the argument rigorous. This does not affect Theorem 1, because the introduction explains that the weaker version with ||T|| suffices, but it is needed for Proposition 4 and for Lemma 13.
  3. [Eq. (47) and (49)] Please verify the displayed identity in Eq. (47): the correct expression should contain the factor r in the sinh term, namely (-1)^n sqrt(2/pi) r sinh(pi r/2) d_n(r), in order to be consistent with Eq. (49). If the current typesetting omits this r, it should be corrected.
  4. [Proof of Proposition 4, Step 0] There is a typo in the first sentence: 'We being with' should be 'We begin with'. Similar spacing and typesetting issues appear in the paragraph after Lemma 14 ('Thefunctions', 'formanorthonormalbasis').
  5. [References] References [1] and [2] have the same title 'The Howland-Kato commutator problem II'; please clarify whether [2] is the preprint version of [1] or a distinct paper, and adjust the reference list accordingly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the proof is self-contained and derives positivity and trace from explicit kernel estimates, not from fitted or target-dependent assumptions.

full rationale

The derivation chain is self-contained. Theorem 1 follows from Proposition 4 in its weak form with ∥T_even∥ replaced by ∥T∥, and Lemma 11 supplies the parameter-free bound ∥T∥ ≤ 3π/5 < 2. Positivity of C_s is shown by Lemma 8's exact representation D_s = (1/2)⟨h_x − h_y, (2sI − B_s)(h_x − h_y)⟩, Lemma 9's sign computation for T_even and T_odd via the positive representation (21), and the Fock-space factorization in Step 3 of Proposition 4; the only input, A_s ≥ 0, is proved from the same lemmas rather than assumed. The trace Tr C_s = π/2 is computed from ∫_R s/(s² + x²) dx = π, with no fitted parameter. Lemma 11's bound is independent of the conclusion, and Proposition 15's Kato-strip determination is a separate analytic continuation argument. No load-bearing self-citation appears; citations to Kato, Herbst–Kriete, Froese–Herbst, Schoenberg, and Simon provide external or classical results but do not carry the counterexample itself. The alleged Fubini gap in Lemma 9 is not circular and does not affect Theorem 1: for v ∈ L²(I) the triple integral is absolutely convergent because the kernel T is bounded, so Tonelli/Fubini applies to the sign computation, and Theorem 1 only needs the simpler Lemma 11 bound. The Eq. (47) typo affects only an optional refinement. Hence there are no circular steps.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

No free parameters: the numerical bound in Proposition 5 is proven, not fitted to data. No invented entities. The proof leans on standard functional calculus and a cited kernel formula for commutators, neither of which is circular.

assumptions (2)
  • standard math Spectral theorem and functional calculus for unbounded self-adjoint operators P and Q
    Used without proof to define arctan(P) and arctan(Q) as bounded self-adjoint operators and to justify the integral kernel representation.
  • standard math Integral kernel formula (12) for i[f(P), g(Q)]
    Taken from Herbst-Kriete [3, Lemma 3.3]; used to compute the kernel of C_s in Lemma 6.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A counterexample to the Kato conjecture for positive commutators." pith.science (2026). https://pith.science/paper/SRH35XEW

@misc{pith2026260807805,
  author       = {Pith},
  title        = {Pith review of: A counterexample to the Kato conjecture for positive commutators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SRH35XEW}},
  note         = {Machine review of arXiv:2608.07805}
}
abstract

We disprove the conjectural converse to Kato's positivity criterion for commutators of functions of the canonical position and momentum operators $Q$ and $P$ by showing that the operator \[ i\,[\,\arctan(P),\,\arctan(Q)\,] \] is nonnegative and nonzero.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

7 extracted references · 7 canonical work pages

  1. [1]

    Froese and I

    R. Froese and I. Herbst,The Howland–Kato commutator problem. II, J. Operator Theory93 (2025), no. 2, 393–412

  2. [2]

    The Howland-Kato Commutator Problem II

    I. Herbst,The Howland–Kato commutator problem II, Preprint (2019), arXiv:1909.06514

  3. [3]

    Herbst and T

    I. Herbst and T. L. Kriete,The Howland–Kato commutator problem, inAnalysis and Operator Theory: Dedicated in Memory of Tosio Kato’s 100th Birthday, Springer Optimization and Its Applications, vol. 146, Springer, Cham, 2019, 191–223

  4. [4]

    J. S. Howland,Perturbation theory of dense point spectra, J. Funct. Anal.74(1987), no. 1, 52–80

  5. [5]

    Kato,Positive commutatorsi[f(P),g(Q)], J

    T. Kato,Positive commutatorsi[f(P),g(Q)], J. Funct. Anal.96(1991), no. 1, 117–129

  6. [6]

    I. J. Schoenberg,Metric spaces and positive definite functions, Trans. Amer. Math. Soc.44 (1938), no. 3, 522–536

  7. [7]

    Simon,Loewner’s theorem on monotone matrix functions, Grundlehren Math

    B. Simon,Loewner’s theorem on monotone matrix functions, Grundlehren Math. Wiss., 354, Springer, Cham, 2019. 15

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.