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 →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
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.
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
- 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.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
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)
- [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.
- [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.
- [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.
- [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').
- [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
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
assumptions (2)
- standard math Spectral theorem and functional calculus for unbounded self-adjoint operators P and Q
- standard math Integral kernel formula (12) for i[f(P), g(Q)]
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.
Reference graph
Works this paper leans on
-
[1]
R. Froese and I. Herbst,The Howland–Kato commutator problem. II, J. Operator Theory93 (2025), no. 2, 393–412
work page 2025
-
[2]
The Howland-Kato Commutator Problem II
I. Herbst,The Howland–Kato commutator problem II, Preprint (2019), arXiv:1909.06514
work page Pith review arXiv 2019
-
[3]
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
work page 2019
-
[4]
J. S. Howland,Perturbation theory of dense point spectra, J. Funct. Anal.74(1987), no. 1, 52–80
work page 1987
-
[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
work page 1991
-
[6]
I. J. Schoenberg,Metric spaces and positive definite functions, Trans. Amer. Math. Soc.44 (1938), no. 3, 522–536
work page 1938
-
[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
work page 2019
Reviewed August 11, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.