A solution to Crouzeix's conjecture

Emiel Lorist; Felix Schwenninger

REVIEW 2 major objections 4 minor 1 cited by

A solution to Crouzeix's conjecture

T0 review · 2 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper claims a full proof of Crouzeix's conjecture: every bounded operator on a Hilbert space has a numerical range that is a 2-spectral set.

desk verdict A short, credible proof of Crouzeix's conjecture whose real contribution is Lemma 1; the application leans on cited double-layer machinery, and there's a fixable typo in the α formula. read the letter →

arxiv 2608.03841 v1 pith:EAXKK7XO submitted 2026-08-04 math.CA math.CVmath.FAmath.OAmath.SP

Emiel Lorist , Felix Schwenninger This is my paper · ORCID

classification math.CAmath.CVmath.FAmath.OAmath.SP MSC 47A2547A1215A60

keywords Crouzeix'sconjecturenumericalrange2-spectralsetdouble-layerpotentialfunctionalcalculusoperatortheoryHilbertspace

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 claims a proof of Crouzeix's conjecture: for every bounded operator A on a Hilbert space, the numerical range W(A) is a 2-spectral set, meaning ||f(A)|| ≤ 2 sup_{z∈W(A)}|f(z)| for every rational function f with poles off the closure of W(A). The conjecture dates to 2004 and previously the best universal constant was 1+√2, with the optimal 2 known for special classes. The proof combines a simple perturbation lemma for 2-dilations with the double-layer potential representation of f(A). If correct, it settles the conjecture with the sharp constant and shows the earlier abstract operator-theoretic program reaches the claimed bound.

What carries the argument

The double-layer potential P_Ω(σ) = (1/π) Re(n_Ω(σ)(σI - A)^{-1}) on the boundary of a smooth convex region Ω containing W(A), together with the induced map Φ(f) = ½∫_∂Ω f P_Ω |dσ|. The proof uses the identity 2Φ(f) - f(A) = α(f)(A)*, where α is a bounded antilinear Cauchy transform, to pair the functional calculus with Lemma 1. Lemma 1 is the other central mechanism: a finite-dimensional perturbation lemma showing that uniform boundedness of the E_n forces ||T|| ≤ 2.

What would settle it

A direct numerical check of the identity 2Φ(f)−f(A)=α(f)(A)^* for a non-normal 2×2 or 3×3 matrix and a rational f with poles outside W(A); any mismatch would break the proof. Independently, an exhaustive search for a matrix with ||f(A)|| > 2 sup_{W(A)}|f| would refute the conjecture itself.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central result is Theorem 3: the numerical range of any bounded Hilbert-space operator is a 2-spectral set, confirming Crouzeix's 2004 conjecture with the optimal constant 2. The proof works by first establishing Lemma 1, a perturbation statement: if an operator T has a contractive dilation Q so that the operators E_n = 2V*Q*^n V - T*^n are uniformly bounded and commute with T, then ||T|| ≤ 2. For T = f(A), the double-layer potential gives a natural dilation Q (multiplication by f on a boundary L^2-space), an isometry V, and the commutativity follows from the identity 2Φ(f) - f(A) = α(f)(A)*. Uniform boundedness of E_n follows from the boundedness of the holomor

Load-bearing premise

The proof's load-bearing premises are two cited results it does not re-prove: that the general Hilbert-space case reduces to finite-dimensional matrices with W(A) replaced by a smoothly bounded convex set, and that the double-layer potential yields a bounded antilinear map α satisfying 2Φ(f)−f(A)=α(f)(A)^*.

Editorial extensions

If this is right

Crouzeix's conjecture holds for all bounded operators on Hilbert space, making 2 the sharp universal bound for rational functions of such operators.
By the generalization cited in the paper, the result extends to closed unbounded operators whose numerical range contains their spectrum.
The proof yields an abstract theorem: any unital bounded homomorphism θ on a commutative uniform algebra with a unital antilinear α satisfying the stated positivity condition has ||θ|| ≤ 2.
The proof does not use the contractivity of α, so the earlier (1+√2) route is not a necessary ingredient; it also avoids extremal functions and measures.

Reading between the lines

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

If the abstract theorem in Remark 4(4) extends cleanly, similar 2-spectral-set bounds might hold in other uniform algebras, and the commutativity assumption in Lemma 1 becomes the natural bottleneck to attack.
The same perturbation lemma could be tried on the completely bounded version of Crouzeix's conjecture; the paper notes its proof does not directly apply, but relaxing the commutativity assumption is a concrete starting point.
The proof suggests that any sequence of matrices approaching the constant 2 must make the E_n's nearly non-uniformly bounded or the commutator condition barely satisfied, which could guide a search for extremal cases.
Remark 2's inequality hints that if the sign of ℜ⟨E_1Tx,x⟩ could be controlled, an even simpler proof of the same bound might exist; this thread is left implicit in the paper.

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.

Desk Editor's Note

read the letter

Colleague,

The thing to know: this note proves Crouzeix's conjecture with a new perturbation lemma for 2-dilations. The conjecture was already settled independently by Jin [12], so the theorem statement isn't new, but the proof strategy is. Lemma 1 is the genuine novelty, and I checked its algebra carefully—it's correct. The contraction argument from the top singular vector, the completing-the-square step, and the κ>2 contradiction all work. The uniform-boundedness assumption on E_n is used to get power-boundedness of T, which is the key step.

What the paper does well: it isolates a clean, abstract condition under which a dilation gives a 2-bound, and then shows the double-layer machinery supplies exactly that condition. That's a nice reduction. The writing is compact and honest; they acknowledge the independent proof and the limitations (no completely bounded case).

The soft spots are the imported machinery. The proof of Theorem 3 depends on two cited facts from [19,20]: the reduction to finite dimensions and the existence of a bounded antilinear α with 2Φ(f)−f(A)=α(f)(A)^*. Those aren't proved in the note. That's fine for a research announcement, but it means the result isn't self-contained.

There is also a real typo: the displayed formula for α says α(f)= (1/2πi)∫ f(σ)(σ−·)^{-1}dσ, which is the Cauchy transform of f, not of \bar f. It contradicts the statement that α is antilinear. The intended formula obviously should have \overline{f(σ)}. The stress-test worry that the corrected α might be unbounded on A(Ω) doesn't hold up—this is a known bounded operator in the double-layer theory, and the note cites [20] for it. The test's counterargument about the Cauchy transform on C(∂Ω) is beside the point, because α acts on A(Ω), not on all continuous functions.

So the proof as written is conditional on cited facts and has a fixable misstatement, but I don't see a load-bearing flaw. The central lemma is correct and the application is plausible. Given the independent proof, the marginal impact is smaller than it would have been a year ago, but this is still a serious piece of work that deserves a referee.

Recommendation: send it to peer review. The typo should be corrected and the α facts stated more precisely, but the core argument is worthy of scrutiny.

Referee Report

2 major / 4 minor

Summary. The paper claims a proof of Crouzeix's conjecture: for every bounded operator A on a Hilbert space, the numerical range W(A) is a 2-spectral set. The argument is built around Lemma 1, a finite-dimensional perturbation lemma: if T admits a contraction Q and an isometry V such that E_n = 2 V* Q^{*n} V - T^{*n} are uniformly bounded and commute with T, then ||T|| ≤ 2. The proof of Lemma 1 is algebraically explicit and appears sound. The authors then apply Lemma 1 to T = f(A) in the operator-valued double-layer potential framework. For a smoothly bounded convex Ω ⊃ W(A), they use a bounded antilinear map α with 2Φ(f) - f(A) = α(f)(A)^*, so that E_n = α(f^n)(A), yielding uniform boundedness and commutativity. The note concludes ||f(A)|| ≤ 2 for all f with ||f||∞ ≤ 1.

Significance. If the proof is correct, this settles a major open problem in operator theory with the sharp constant 2. The main conceptual contribution is Lemma 1, which is a clean and essentially self-contained perturbation argument; its proof is a strength of the paper. The application to the double-layer potential is elegant and short. However, the note is not self-contained: it relies on substantial machinery from [19,20], and the displayed formula for the key antilinear map α contains an apparent misstatement. These caveats do not necessarily invalidate the claim, but they make verification difficult and are load-bearing for the main theorem.

major comments (2)

[Theorem 3 proof, formula for α] The displayed formula α(f) = 1/(2πi) ∫_{∂Ω} f(σ)(σ-·)^{-1} dσ is not antilinear: by the Cauchy integral formula this expression equals f for f ∈ A(Ω), and it is not the Cauchy transform of \bar f. The correct formula should presumably involve \overline{f(σ)} in the integrand. This is not merely cosmetic: the uniform boundedness of E_n in the proof is obtained from ∥E_n∥ ≤ ∥θ∥ ∥α∥ ∥f^n∥∞, and the commutativity of E_n with T uses the representation E_n = α(f^n)(A). As printed, the identity 2Φ(f)-f(A) = α(f)(A)^* is not verifiable from the stated definition. The authors should correct the formula and give a precise statement (or reference) for the boundedness of the intended α.
[Reduction to finite dimensions and convex Ω] The proof of Theorem 3 begins with a reduction to H = C^d and to a smoothly bounded open convex Ω containing W(A), citing [19,20]. This is a substantial step, and the rest of the proof depends on the double-layer machinery from [20]. The note also implicitly uses a bounded homomorphism θ:A(Ω)→L(H) in the estimate ∥E_n∥ ≤ ∥θ∥ ∥α∥; its boundedness is not stated. For a result of this importance, the authors should either state the precise reduction proposition from [19,20] or indicate exactly which theorem in those references supplies it. I am not asking for a reproduction of [20], but the current dependence is too opaque.

minor comments (4)

[Theorem 3 proof] The map θ is introduced only in the sentence about commutativity; its boundedness and norm should be stated before the inequality ∥E_n∥ ≤ ∥θ∥ ∥α∥ is used.
[Displayed formula] If the intended formula is indeed the Cauchy transform of \bar f, please add the missing overline and clarify the orientation of dσ.
[Remark 4(4)] There is a minor typographical issue: 'α:A → Aa unital' should read 'α:A → A, a unital'.
[References] Reference [12] is to a preprint server; if a stable published version exists, it would be helpful to cite it.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the 2-bound is derived from a new perturbation lemma; cited double-layer facts are independent and do not assume the conjecture.

full rationale

The proof chain is: Lemma 1 gives ||T|| ≤ 2 from uniform boundedness and commutation of the operators E_n; Theorem 3 verifies these hypotheses for T = f(A) using the double-layer representation. The load-bearing external facts are (i) reduction to finite dimensions with a smoothly bounded convex Ω containing W(A) and (ii) the identity 2Φ(f) − f(A) = α(f)(A)^* with α bounded antilinear. These are cited to [19,20], one co-authored by the present author Schwenninger, so self-citation is present. However, the cited facts are parameter-free theorems about the double-layer potential and the holomorphic functional calculus; they do not assume ||f(A)|| ≤ 2 and are not equivalent to it by construction. The uniform boundedness of E_n follows from the cited boundedness of α, and Remark 4(1) even notes that Crouzeix's earlier estimate already gives sup_n ||E_n|| < ∞. Commutativity follows from the homomorphism property of the functional calculus, not from the conjecture. No fitted parameter is renamed as a prediction, and no known result is repackaged under new names. The apparent typo in the recalled formula for α (missing conjugation on f(σ)) is a correctness concern, not a circularity, because the identity used in the proof is a separate cited theorem. Overall, the central derivation does not reduce to its own inputs.

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

The central claim rests on standard double-layer potential machinery and functional calculus, plus two cited structural reductions. No free parameters are fitted and no new objects are postulated. The main novelty, Lemma 1, supplies the missing operator-theoretic step.

assumptions (5)

standard math Cauchy integral theorem and positivity of the double-layer potential give Φ(1)=I and P_Ω(σ)≥0.
Used to define the isometry V and the contraction Q in the proof of Theorem 3.
domain assumption There exists a bounded antilinear α:A(Ω)→A(Ω) with 2Φ(f)-f(A)=α(f)(A)^*.
Cited to [20]; gives E_n=α(f^n)(A) and its uniform boundedness.
domain assumption The general Hilbert space case reduces to H=C^d with W(A) replaced by a smoothly bounded open convex Ω.
Cited to [19,20]; the proof is only carried out in this setting.
standard math Holomorphic functional calculus θ:A(Ω)→L(H), g↦g(A), is a unital bounded homomorphism.
Used to show E_n and T commute in Theorem 3.
standard math Arveson extension theorem and Stinespring dilation theorem.
Used in Remark 4(4) for the abstract variant, not needed for the main proof.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A solution to Crouzeix's conjecture." pith.science (2026). https://pith.science/paper/EAXKK7XO

@misc{pith2026260803841,
  author       = {Pith},
  title        = {Pith review of: A solution to Crouzeix's conjecture},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EAXKK7XO}},
  note         = {Machine review of arXiv:2608.03841}
}

abstract

We provide a proof of Crouzeix's conjecture, which combines the tools developed previously for weaker estimates with a simple perturbation lemma for $2$-dilations. Applying the lemma to the iterates $f^{n}$ in the double-layer potential representation yields the conjectured bound.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

Sharp spectral constants for scaled $q$-numerical ranges
math.FA 2026-08 accept novelty 8.0 of 10

For every nonzero q with |q|<=1, the sharp spectral constant of the scaled q-numerical range is max{1, 2|q|/(1+sqrt(1-|q|^2))}.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages · cited by 1 Pith paper

[12]

S. Jin. The numerical range is a 2-spectral set. Preprint onhttps://www.preprints.org/ manuscript/202607.1919, 2026

work page arXiv 1919
[20]

F. L. Schwenninger and J. de Vries. The double-layer potential for spectral constants revisited. Integral Equations Operator Theory, 97(2):Paper No. 13, 22, 2025. Emiel Lorist Delft Institute of Applied Mathematics Delft University of Technology P.O. Box 5031 2600 GA Delft, The Netherlands Email address:e.lorist@tudelft.nl Felix L. Schwenninger Department...

work page 2025
[1]

Bickel and P

K. Bickel and P. Gorkin. Blaschke products, level sets, and Crouzeix’s conjecture.J. Anal. Math., 152(1):217–254, 2024

work page 2024
[2]

Bickel, P

K. Bickel, P. Gorkin, A. Greenbaum, T. Ransford, F. L. Schwenninger, and E. Wegert. Crouzeix’s conjecture and related problems.Comput. Methods Funct. Theory, 20(3-4):701– 728, 2020

work page 2020
[3]

Caldwell, A

T. Caldwell, A. Greenbaum, and K. Li. Some extensions of the Crouzeix-Palencia result. SIAM J. Matrix Anal. Appl., 39(2):769–780, 2018

work page 2018
[4]

Clouˆ atre, M

R. Clouˆ atre, M. Ostermann, and T. Ransford. An abstract approach to the Crouzeix conjec- ture.J. Operator Theory, 90(1):209–221, 2023. A SOLUTION TO CROUZEIX’S CONJECTURE 5

work page 2023
[5]

Crouzeix

M. Crouzeix. Bounds for analytical functions of matrices.Integral Equations Operator Theory, 48(4):461–477, 2004

work page 2004
[6]

Crouzeix

M. Crouzeix. Numerical range and functional calculus in Hilbert space.J. Funct. Anal., 244(2):668–690, 2007

work page 2007

Show all 20 references

[7]

Crouzeix, A

M. Crouzeix, A. Greenbaum, and K. Li. Numerical bounds on the Crouzeix ratio for a class of matrices.Calcolo, 61(2):Paper No. 32, 18, 2024

2024
[8]

Crouzeix and C

M. Crouzeix and C. Palencia. The numerical range is a (1 + √ 2)-spectral set.SIAM J. Matrix Anal. Appl., 38(2):649–655, 2017

2017
[9]

Delyon and F

B. Delyon and F. Delyon. Generalization of von Neumann’s spectral sets and integral repre- sentation of operators.Bull. Soc. Math. France, 127(1):25–41, 1999

1999
[10]

Glader, M

C. Glader, M. Kurula, and M. Lindstr¨ om. Crouzeix’s conjecture holds for tridiagonal 3×3 matrices with elliptic numerical range centered at an eigenvalue.SIAM J. Matrix Anal. Appl., 39(1):346–364, 2018

2018
[11]

Hartz and J

M. Hartz and J. E. McCarthy. From Clouatre-Ostermann-Ransford to Okubo-Ando. Preprint, arXiv:2606.02922 [math.F A] (2026), 2026

2026 arXiv
[13]

Malman, J

B. Malman, J. Mashreghi, R. O’Loughlin, and T. Ransford. Double-layer potentials, configu- ration constants, and applications to numerical ranges.Int. Math. Res. Not. IMRN, (8):Paper No. rnaf084, 34, 2025

2025
[14]

Mashreghi, A

J. Mashreghi, A. Moucha, R. O’Loughlin, T. Ransford, and O. Roth. A Schwarz-Jack lemma, circularly symmetric domains and numerical ranges.J. Math. Anal. Appl., 563(1):Paper No. 130786, 10, 2026

2026
[15]

Okubo and T

K. Okubo and T. Ando. Constants related to operators of classC ρ.Manuscripta Math., 16(4):385–394, 1975

1975
[16]

O’Loughlin and J

R. O’Loughlin and J. Virtanen. Crouzeix’s conjecture for classes of matrices.Linear Algebra Appl., 697:277–292, 2024

2024
[17]

Paulsen.Completely bounded maps and operator algebras, volume 78 ofCambridge Studies in Advanced Mathematics

V. Paulsen.Completely bounded maps and operator algebras, volume 78 ofCambridge Studies in Advanced Mathematics. Cambridge University Press, Cambridge, 2002

2002
[18]

Ransford and F

T. Ransford and F. L. Schwenninger. Remarks on the Crouzeix-Palencia proof that the nu- merical range is a (1 + √ 2)-spectral set.SIAM J. Matrix Anal. Appl., 39(1):342–345, 2018

2018
[19]

F. L. Schwenninger and J. de Vries. On abstract spectral constants. InOperator and matrix theory, function spaces, and applications, volume 295 ofOper. Theory Adv. Appl., pages 335–350. Birkh¨ auser/Springer, Cham, 2024

2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [12]

S. Jin. The numerical range is a 2-spectral set. Preprint onhttps://www.preprints.org/ manuscript/202607.1919, 2026

work page arXiv 1919

[2] [20]

F. L. Schwenninger and J. de Vries. The double-layer potential for spectral constants revisited. Integral Equations Operator Theory, 97(2):Paper No. 13, 22, 2025. Emiel Lorist Delft Institute of Applied Mathematics Delft University of Technology P.O. Box 5031 2600 GA Delft, The Netherlands Email address:e.lorist@tudelft.nl Felix L. Schwenninger Department...

work page 2025

[3] [1]

Bickel and P

K. Bickel and P. Gorkin. Blaschke products, level sets, and Crouzeix’s conjecture.J. Anal. Math., 152(1):217–254, 2024

work page 2024

[4] [2]

Bickel, P

K. Bickel, P. Gorkin, A. Greenbaum, T. Ransford, F. L. Schwenninger, and E. Wegert. Crouzeix’s conjecture and related problems.Comput. Methods Funct. Theory, 20(3-4):701– 728, 2020

work page 2020

[5] [3]

Caldwell, A

T. Caldwell, A. Greenbaum, and K. Li. Some extensions of the Crouzeix-Palencia result. SIAM J. Matrix Anal. Appl., 39(2):769–780, 2018

work page 2018

[6] [4]

Clouˆ atre, M

R. Clouˆ atre, M. Ostermann, and T. Ransford. An abstract approach to the Crouzeix conjec- ture.J. Operator Theory, 90(1):209–221, 2023. A SOLUTION TO CROUZEIX’S CONJECTURE 5

work page 2023

[7] [5]

Crouzeix

M. Crouzeix. Bounds for analytical functions of matrices.Integral Equations Operator Theory, 48(4):461–477, 2004

work page 2004

[8] [6]

Crouzeix

M. Crouzeix. Numerical range and functional calculus in Hilbert space.J. Funct. Anal., 244(2):668–690, 2007

work page 2007