Pith. sign in

REVIEW 2 major objections 4 minor 27 references

Square Functions and the Complete Crouzeix Conjecture in Dimension Three

T0 review · 2 major / 4 minor · reviewed 2026-08-28 · deepseek-v4-flash

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

Pith's one-line read For matrices of order at most three, every matrix-valued polynomial $F$ satisfies $\|F(A)\| \le 2\max_{z\in W(A)}\|F(z)\|$, so the numerical range is a complete 2-spectral set with optimal constant.

desk verdict A substantial and mostly convincing attack on the complete Crouzeix conjecture in dimension three, but the central square-function proof as printed has repairable conjugation errors that a referee must check. read the letter →

arxiv 2608.27346 v1 pith:W4OWRZTP submitted 2026-08-27 math.CV math.FA

Per {AA}hag , Rafa{l} Czyż , Antti Perälä , Jani Virtanen This is my paper · ORCID
classification math.CVmath.FA MSC 47A2547A1247A6046L0715A60
keywords completeCrouzeixconjecturenumericalrange2-spectralsetsquare-functioninequalitydouble-layerpotentialcompletelyboundedfunctionalcalculusq-numericalmatrix-valuedpolynomial
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 proves the complete Crouzeix conjecture for matrices of order at most three: for every $A\in M_n(\mathbb{C})$ with $n\le 3$ and every matrix-valued polynomial $F$ of arbitrary size, $\|F(A)\| \le 2\max_{z\in W(A)}\|F(z)\|$, so the numerical range $W(A)$ is a complete 2-spectral set for $A$. The constant 2 is best possible, attained already by the 2×2 nilpotent shift, so the result is sharp. The complete version of the conjecture, the matrix-amplified form introduced with the scalar conjecture, had previously been open with only the universal bound $1+\sqrt{2}$, while the plain scalar version was proved in full generality elsewhere; here the complete statement is settled through dimension three, with the two-dimensional case previously known. The proof introduces a pair of tuple square-function inequalities for functions of $A$ plus a finite-level reduction showing that any counterexample would have to appear at an $s\times r$ matrix level with $r,s\ge 2$ and $r+s\le n$, which is impossible for $n\le 3$ and leaves only the 2×2 level in dimension four. A reader should care because this is the sharp, matrix-amplified form of a central conjecture about how much of a matrix is controlled by its numerical range; the same machinery yields strict bounds for diagonalizable matrices, quantitative rigidity at the sharp constant, and sharp complete spectral constants for scaled $q$-numerical ranges in dimensions two and three.

What carries the argument

The argument turns on two mechanisms. First, the tuple square functions: for functions $f_1,\dots,f_q\in A(W(A))$ with $\sum_j|f_j|^2\le 1$ on $W(A)$ and any $0<\tau<1$, the column series $P_\tau(f,A)=\sum_{k\ge 0}\tau^{2k}\sum_{\alpha\in W_k} f_\alpha(A)^* f_\alpha(A)$ (summed over all words $\alpha$ of length $k$), and its row analogue $Q_\tau$, satisfy the two-sided estimate $I\preceq P_\tau,Q_\tau\preceq (1-\tau)^{-1}I$. This is proved by sampling a matrix-valued Herglotz kernel, the positive kernel $K_a(w,z)=(c_a(w)+c_a(z))/(1-z^*w)$ with $c_a(w)=(1+w^Ta)/(1-w^Ta)$, on the unit ball of $\mathbb{C}^q$ instead of the disk, combining the positive operator-valued double-layer calculus of [11,12,25] with a multivariable form of the sampling argument of [18]; the key algebraic input is the relation $H(w)-(I-T(w))^{-1}\in\mathrm{alg}(B^*)$, which places the double-layer sample in the algebra generated by $B^*$. At $\tau=1/2$ this yields the column and row bounds of 2, which are optimal. Second, the finite-level reduction via the rectangular form of the argument associated with [26] and the variational lemma of [16]: if the complete functional-calculus norm of $A$ exceeds 2, a minimal extremal is attained at an $s\times r$ matrix level with $r+s\le d$, its norming vectors have full Schmidt ranks, and the square-function estimates eliminate the scalar, column, and row levels $r=1$ or $s=1$. The residual levels $r,s\ge 2$ with $r+s\le d$ are untenable for $d\le 3$, which is precisely why the complete conjecture is settled through dimension three.

What would settle it

Take a concrete non-diagonalizable 3×3 matrix $A$ (for instance the family $A_t$ of (11.19) at some $t\ne 1/3$, or a Jordan block $J_3(\lambda)$) and a pair of functions $f_1,f_2$ with $|f_1|^2+|f_2|^2\le 1$ on $W(A)$, then evaluate the finite partial sums of $P_\tau(f,A)$ at $\tau=1/2$ numerically: Theorem 1.2 predicts every partial sum is $\le 2I$, so any unit vector exceeding that bound would refute the square-function estimate and, with it, Theorem 1.1. Alternatively, the paper's reduction concentrates the whole open question in dimension four at the 2×2 matrix level: a search over $A\in M_4(\mathbb{C})$ and 2×2 matrix polynomials $F$ for a violation of $\|F(A)\|>2\max_{W(A)}\|F\|$ would either exhibit the first counterexample to the complete conjecture or support its truth.

Watch

Extended reading notes

Core claim

The central discovery is Theorem 1.1: the complete Crouzeix conjecture holds with optimal constant in dimension three. Concretely, for $A\in M_n(\mathbb{C})$ with $n\le 3$, every matrix-valued polynomial $F$ of arbitrary size satisfies $\|F(A)\| \le 2\max_{z\in W(A)}\|F(z)\|$, equivalently $W(A)$ is a complete 2-spectral set for $A$, and the constant 2 cannot be lowered. The paper also proves a structural reduction: if the complete inequality fails for some $A\in M_n(\mathbb{C})$, it already fails at an $s\times r$ rectangular matrix-functional level with $r,s\ge 2$ and $r+s\le n$; consequently no failure can occur below dimension four, and in dimension four the entire question collapses to the single 2×2 matrix level. Around this core the paper obtains sharp column and row square-function inequalities with optimal constant 2 (Theorem 1.2); the equivalence $\psi_{cb}(A)=2 \iff \psi(A)=2$ in dimensions at most three (Theorem 1.3); the strict bound $\psi_{cb}(A)\le \min\{\mathrm{cond}(V),\Phi(\mathrm{cond}(V))\}<2$ for diagonalizable $A$ (Theorem 1.4); rigidity results showing that near the sharp constant $f(A)$ has a two-dimensional reducing subspace close to the nilpotent block $\begin{smallmatrix}0&0\\2&0\end{smallmatrix}$; sharp complete spectral constants for scaled $q$-numerical ranges in dimensions two and three; and a complete analysis of a one-parameter family of 3×3 matrices whose complete ratio approaches 2 only in the limit at the endpoint of the parameter range.

Load-bearing premise

Everything below dimension four rests on the tuple square-function inequality (Theorem 1.2): for every tuple $f_1,\dots,f_q$ with $\sum_j|f_j|^2\le 1$ on $W(A)$, the weighted word-sums must satisfy $I\preceq P_\tau(f,A),Q_\tau(f,A)\preceq (1-\tau)^{-1}I$; if the double-layer positivity or the multivariable sampling step failed, the column and row bounds of 2 would not follow, the scalar, column, and row levels would not be excluded, and the dimension-three conclusion would collapse.

Editorial extensions

If this is right

  • For every matrix of order at most three, all matrix-valued polynomials $F$ satisfy $\|F(A)\|\le 2\max_{z\in W(A)}\|F(z)\|$ with 2 sharp, so the numerical range is a complete 2-spectral set in this range of dimensions.
  • Any counterexample to the complete Crouzeix conjecture in higher dimension must occur at an $s\times r$ matrix-functional level with $r,s\ge 2$ and $r+s\le n$; in dimension four the question reduces exactly to the 2×2 matrix level.
  • In dimensions at most three, equality $\psi_{cb}(A)=2$ forces non-diagonalizability: diagonalizable $A$ satisfy $\psi_{cb}(A)\le \min\{\mathrm{cond}(V),\Phi(\mathrm{cond}(V))\}<2$, and any sequence of diagonalizable matrices with $\psi_{cb}\to 2$ has eigenvector condition numbers tending to infinity.
  • Near the sharp constant the structure is rigid: if $\|f\|_{W(A)}=1$ and $\|f(A)\|\ge 2-\varepsilon$ for $0\le\varepsilon\le 2-\sqrt{15}/2$, then $f(A)$ is within $8\sqrt{\varepsilon}$ of an operator with reducing summand $\begin{smallmatrix}0&0\\2&0\end{smallmatrix}$, and the pushforward of the extremal representing measure converges weakly to normalized arclength measure as equality is approached
  • For scaled $q$-numerical ranges in dimensions two and three the sharp complete spectral constant is $\max\{1,2|q|/(1+\sqrt{1-|q|^2})\}$; thus $\Omega_q(A)$ is a complete spectral set in those dimensions exactly when $|q|\le 4/5$, and in all dimensions when $|q|\le 1/\sqrt{2}$.

Reading between the lines

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

  • The reduction to $r,s\ge 2$ with $r+s\le n$ turns the open higher-dimensional case into a finite, well-posed search: deciding the complete conjecture in dimension four is equivalent to checking the 2×2 matrix level over $M_4(\mathbb{C})$, a parameter space small enough for systematic numerical sweeps.
  • The tuple square functions come from sampling the double-layer kernel on the unit ball of $\mathbb{C}^q$; the same multivariable sampling should apply to commuting tuples of operators or to matrix algebras with several generators, where no complete Crouzeix-type statement is currently available.
  • The paper leaves open whether equality $\psi(A_t)=\psi_{cb}(A_t)=2$ occurs anywhere on the one-parameter family of (11.19) away from $t=1/3$; if the rigidity picture of Section 10 is complete, equality on that curve would require the explicit extremal identities (10.2) or (10.3) to hold, which are concrete computable conditions.
  • The strictness for diagonalizable matrices and the concentration of near-extremal behavior on the 2×2 nilpotent block suggest that, if the complete conjecture is true in all dimensions, all sharpness is concentrated at defective matrices; a hypothetical 4×4 counterexample, if it exists, would be non-diagonalizable and close to the same block structure.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Request a human review

A listed scientist reviews the paper for a fee and the review publishes here regardless of verdict. See the reviewers or get listed.

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 claims to settle the complete Crouzeix conjecture for matrices of order at most three, i.e. that W(A) is a complete 2-spectral set for every A in M_n(C) with n ≤ 3. The proof combines a new tuple square-function inequality (Theorem 1.2), obtained from Delyon–Delyon / Crouzeix–Palencia double-layer calculus and a multivariable version of Jin's sampling argument, with a finite-level reduction (Section 5) based on Smith's lemma and a Hartz–McCarthy variational lemma. The paper also derives quantitative similarity estimates for diagonalizable matrices, sharp complete spectral constants for scaled q-numerical ranges in dimensions two and three, rigidity and stability results near equality, and a detailed analysis of a one-parameter family A_{\mu,r} including an exact rational-arithmetic certificate ruling out equality at a triple point.

Significance. If the main theorem is established, it resolves a natural complete-operator-algebra strengthening of Crouzeix's conjecture in the low-dimensional cases and gives the first completely bounded spectral constants for scaled q-numerical ranges in dimensions two and three. The finite-level reduction is elegant: it reduces a hypothetical counterexample to a rectangular matrix-valued level with r,s ≥ 2 and r+s ≤ n, leaving only the 2×2 level in dimension four. The paper's exact rational-arithmetic Sturm-sequence certification of the containment f1(D) ⊂ W(K3) and the explicit condition-number formulas are reproducible and are genuine strengths. However, the proof of the central Theorem 1.2 contains two algebraic inconsistencies that are load-bearing; the manuscript as written does not establish its main claim.

major comments (2)
  1. [Section 3, Lemma 3.2] The kernel K_a(w,z) = (c_a(w)+c_a(z))/(1-z^*w) is not positive under the stated definition c_a(w) = (1+w^T a)/(1-w^T a) with bilinear w^T a. For q=1, a=0.5i, w_1=0, w_2=0.5, the off-diagonal entries of the 2×2 matrix equal 1.882+0.471i, so the matrix is not Hermitian and cannot be positive semidefinite. The factorization (3.2) is algebraically inconsistent with the displayed definition of c_a; it is in fact the factorization of the Herglotz kernel (c_a(w)+\overline{c_a(z)})/(1-z^*w). Since Lemma 3.4 integrates exactly this kernel and Proposition 3.5 relies on the resulting positivity, the proof of Theorem 1.2 is unsupported as written.
  2. [Section 3, Proposition 3.5] The identities w_j^*w_i = τ^2 c_{ij} and w_i^T λ_j = τ c_{ij} are false for c_{ij}=λ_i^*λ_j and w_i=τλ_i: the first equals τ^2 c_{ji} and the second equals τ λ_i^T λ_j. These identities are used to define P,Q,Y in (3.11), to compute M in (3.13), and to derive the inequalities (3.14)–(3.19); consequently, inequality (3.9) does not follow from the argument given. The gap appears repairable — for example, taking w_i=τ\overline{λ_i} makes both displayed identities correct — but that repaired argument is not what the manuscript supplies. Until this is fixed, Theorem 1.2 and the dimension-three conclusion of Theorem 1.1 are not proven.
minor comments (4)
  1. [Theorem 1.2, Eq. (1.6)] The set W_k of words of length k is used without a formal definition; define W_k = {1,...,q}^k before the theorem.
  2. [Section 5, Lemma 5.2] The ball automorphism m_E is applied to elements of an abstract operator algebra; a sentence explaining that A is represented completely isometrically and that the automorphism is applied entrywise would improve readability.
  3. [Section 11, Lemma 11.8] In the Sturm sequence table, the equal left and right variations should be explicitly interpreted as the absence of real roots in the stated intervals; currently the table alone is ambiguous.
  4. [Section 11, Proposition 11.2] The notation A_{a,b} is used where the family was defined as A_{\mu,r}; align the notation or explain the change.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the complete Crouzeix bound is derived from an independently proved tuple square-function inequality and a finite-level extremal reduction, not from the conjecture itself.

full rationale

The paper's central estimate is Theorem 1.2, whose proof is supplied in Section 3. The positivity kernel Lemma 3.2 is asserted with a direct algebraic factorization (3.2); Lemma 3.3 is imported from Delyon-Delyon and Crouzeix-Palencia material [11,12,25]; Proposition 3.5 samples the resulting positive kernel and derives the column and row square-function bounds without invoking Theorem 1.1 or the Crouzeix conjecture. Theorem 1.1 is then obtained by a finite-level rectangular extremal argument (Theorem 5.3, Lemma 5.2) applied to the quotient algebra, with the bound 2 at levels (1,1), (1,2), and (2,1) coming from Lemma 6.1, which is exactly Theorem 1.2 shifted to the quotient. There are no fitted parameters, no data subsets, and no ansatz smuggled in by citation: the double-layer calculus and Jin's sampling argument are external, and the paper does not rely on its own prior results for the load-bearing square-function step. The skeptical objections to Lemma 3.2 and Proposition 3.5 concern the validity of the positivity calculation under the paper's own stated assumptions, which is a correctness or invalidity issue, not a circularity. The final paragraph explicitly leaves an open equality question on the family At, which is a limitation statement rather than a hidden reuse of the target. Accordingly no circular step is identified.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No fitted constants or ad hoc parameters are introduced. The paper introduces no new physical or mathematical entities; the scaled q-numerical range and the family A_{mu,r} come from the prior literature. Constants such as 2, Phi(kappa), and chi(r) are derived rather than fitted.

assumptions (6)
  • standard math Positive double-layer calculus: for a convex domain Omega containing W(B), there exists a positive unital map Phi : C(dOmega) -> M_d(C) with 2Phi(h) = h(B) + C_h^* and C_h in alg(B), used as Lemma 3.3.
    Imported from [11,12,25]; this is the foundation of the square-function estimates in Theorem 1.2 and hence of the complete bound.
  • standard math Smith's finite-level lemma: for a unital homomorphism theta : A -> M_d(C), ||theta||_cb = max_{1 <= r,s <= d} ||theta^(s,r)||, used in Lemma 5.1.
    Standard result in operator algebra theory; the dimension count r+s <= d in Theorem 5.3 depends on it.
  • standard math Hartz-McCarthy variational lemma, Lemma 5.2: a finite-dimensional homomorphism attaining its cb norm at F satisfies the orthogonality relation <Rx,(D tensor I)x> = 0 for all D.
    A proof is included in the paper; the statement is attributed to [16] and is used to obtain orthogonal Schmidt vectors for minimal rectangular extremals.
  • standard math Crouzeix's finite-dimensional extremal theorem [5, Theorem 2.1]: a normalized extremal function equals a finite Blaschke product composed with a conformal bijection from the domain to the disk.
    Used in Proposition 9.1 and Section 10 to classify extremal functions at equality.
  • domain assumption Aouichaoui-O'Loughlin similarity transfer: Omega_r(A) = union over cond(S) <= chi(r) of W(S^{-1}AS), with proper interior containment for cond(S) < chi(r), equations (7.6)-(7.7).
    External theorem [1] used in Theorems 7.2 and 7.3 for the q-numerical-range spectral constants.
  • standard math Standard tools: the joint spectral theorem for commuting normal operators in Lemma 4.3, and the Routh-Hurwitz and Sturm-sequence exact arithmetic verification in Lemma 11.8 from [3].
    The moment inequality for normal tuples and the strict inequality at the triple point both rest on these standard tools.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Square Functions and the Complete Crouzeix Conjecture in Dimension Three." pith.science (2026). https://pith.science/paper/W4OWRZTP

@misc{pith2026260827346,
  author       = {Pith},
  title        = {Pith review of: Square Functions and the Complete Crouzeix Conjecture in Dimension Three},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W4OWRZTP}},
  note         = {Machine review of arXiv:2608.27346}
}
abstract

We settle the complete Crouzeix conjecture for matrices of order at most three. We also obtain sharp column and row square-function inequalities, strict scalar bounds for operators similar to normal operators, sharp complete spectral constants for scaled $q$-numerical ranges in dimensions two and three, and rigidity, stability, and representing-measure results.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 26 canonical work pages

  1. [1]

    Sharp spectral constants for scaled $q$-numerical ranges

    Mohamed Amine Aouichaoui and Ryan O’Loughlin. Sharp spectral constants for scaledq-numerical ranges, 2026. arXiv:2608.09866v1

    work page Pith review arXiv 2026

  2. [2]

    CatalinBadea, MichelCrouzeix, andBernardDelyon.ConvexdomainsandK-spectralsets.Mathematische Zeitschrift, 252(2):345–365, 2006

    work page 2006

  3. [3]

    Springer, Berlin, 2 edition, 2006

    Saugata Basu, Richard Pollack, and Marie-Françoise Roy.Algorithms in Real Algebraic Geometry, volume 10 of Algorithms and Computation in Mathematics. Springer, Berlin, 2 edition, 2006

    work page 2006

  4. [4]

    KellyBickel, PamelaGorkin, AnneGreenbaum, ThomasRansford, FelixL.Schwenninger, andEliasWegert.Crouzeix’s conjecture and related problems.Computational Methods and Function Theory, 20(3–4):701–728, 2020

    work page 2020

  5. [5]

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

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

    work page 2004

  6. [6]

    Numerical range and functional calculus in hilbert space.Journal of Functional Analysis, 244(2):668– 690, 2007

    Michel Crouzeix. Numerical range and functional calculus in hilbert space.Journal of Functional Analysis, 244(2):668– 690, 2007

    work page 2007

  7. [7]

    Spectral sets and3×3nilpotent matrices

    Michel Crouzeix. Spectral sets and3×3nilpotent matrices. In Catalin Badea, Daniel Li, and Violeta Petkova, editors, Topics in Functional and Harmonic Analysis, volume 14 ofTheta Series in Advanced Mathematics, pages 27–42. Theta, Bucharest, 2013

    work page 2013

  8. [8]

    Some constants related to numerical ranges.SIAM Journal on Matrix Analysis and Applications, 37(1):420–442, 2016

    Michel Crouzeix. Some constants related to numerical ranges.SIAM Journal on Matrix Analysis and Applications, 37(1):420–442, 2016

    work page 2016

Show all 27 references
  1. [9]

    Numerical bounds on the crouzeix ratio for a class of matrices

    Michel Crouzeix, Anne Greenbaum, and Kenan Li. Numerical bounds on the crouzeix ratio for a class of matrices. Calcolo, 61(2):32, 2024. Paper No. 32

    2024

  2. [10]

    A bivariate extension of the crouzeix–palencia result with an application to Fréchet derivatives of matrix functions.Linear and Multilinear Algebra, 73:2493–2500, 2025

    Michel Crouzeix and Daniel Kressner. A bivariate extension of the crouzeix–palencia result with an application to Fréchet derivatives of matrix functions.Linear and Multilinear Algebra, 73:2493–2500, 2025

    2025

  3. [11]

    The numerical range is a(1 + √ 2)-spectral set.SIAM Journal on Matrix Analysis and Applications, 38(2):649–655, 2017

    Michel Crouzeix and César Palencia. The numerical range is a(1 + √ 2)-spectral set.SIAM Journal on Matrix Analysis and Applications, 38(2):649–655, 2017

    2017

  4. [12]

    Generalization of von neumann’s spectral sets and integral representation of operators.Bulletin de la Société Mathématique de France, 127(1):25–41, 1999

    Bernard Delyon and François Delyon. Generalization of von neumann’s spectral sets and integral representation of operators.Bulletin de la Société Mathématique de France, 127(1):25–41, 1999

    1999

  5. [13]

    arXiv:2503.05036v1

    Dušan Denčić, Hranislav Stanković, Mihailo Krstić, and Ivan Damnjanović.q-numerical radius of rank-one operators and the generalized Buzano inequality, 2025. arXiv:2503.05036v1

    2025 arXiv

  6. [14]

    Duren.Univalent Functions, volume 259 ofGrundlehren der Mathematischen Wissenschaften

    Peter L. Duren.Univalent Functions, volume 259 ofGrundlehren der Mathematischen Wissenschaften. Springer- Verlag, New York, 1983

    1983

  7. [15]

    Anne Greenbaum and Michael L. Overton. Numerical investigation of crouzeix’s conjecture.Linear Algebra and its Applications, 542:225–245, 2018

    2018

  8. [16]

    McCarthy

    Michael Hartz and John E. McCarthy. From Clouâtre–Ostermann–Ransford to Okubo–Ando, 2026. arXiv:2606.02922v1

    2026 arXiv

  9. [17]

    Isidro and László L

    José M. Isidro and László L. Stachó.Holomorphic Automorphism Groups in Banach Spaces: An Elementary Intro- duction, volume 105 ofNorth-Holland Mathematics Studies. North-Holland, Amsterdam, 1985

    1985

  10. [18]

    The numerical range is a2-spectral set

    Shanmu Jin. The numerical range is a2-spectral set. Preprints.org, 2026. Version 4

    2026

  11. [19]

    Mehta, and Leiba Rodman

    Chi-Kwong Li, Paras P. Mehta, and Leiba Rodman. A generalized numerical range: The range of a constrained sesquilinear form.Linear and Multilinear Algebra, 37(1–3):25–49, 1994

    1994

  12. [20]

    Schwenninger

    Emiel Lorist and Felix L. Schwenninger. A solution to crouzeix’s conjecture, 2026. arXiv:2608.03841v2

    2026 arXiv

  13. [21]

    Double-layer potentials, configuration constants, and applications to numerical ranges.International Mathematics Research Notices, 2025(8):1–34, 2025

    Bartosz Malman, Javad Mashreghi, Ryan O’Loughlin, and Thomas Ransford. Double-layer potentials, configuration constants, and applications to numerical ranges.International Mathematics Research Notices, 2025(8):1–34, 2025. Article rnaf084

    2025

  14. [22]

    arXiv:2603.15536v1

    Ryan O’Loughlin and Jyoti Rani.q-numerical ranges and spectral sets, 2026. arXiv:2603.15536v1

    2026

  15. [23]

    Paulsen.Completely Bounded Maps and Operator Algebras, volume 78 ofCambridge Studies in Advanced Mathematics

    Vern I. Paulsen.Completely Bounded Maps and Operator Algebras, volume 78 ofCambridge Studies in Advanced Mathematics. Cambridge University Press, Cambridge, 2002

    2002

  16. [24]

    Schwenninger

    Thomas Ransford and Felix L. Schwenninger. Remarks on the crouzeix–palencia proof that the numerical range is a (1 + √ 2)-spectral set.SIAM Journal on Matrix Analysis and Applications, 39(1):342–345, 2018. 36 PER ÅHAG, RAF AŁ CZYŻ, ANTTI PERÄLÄ, AND JANI VIRTANEN

    2018

  17. [25]

    Schwenninger and Jens de Vries

    Felix L. Schwenninger and Jens de Vries. The double-layer potential for spectral constants revisited.Integral Equations and Operator Theory, 97:13, 2025. Paper No. 13

    2025

  18. [26]

    Roger R. Smith. Completely bounded maps betweenC ∗-algebras.Journal of the London Mathematical Society, 27(1):157–166, 1983

    1983

  19. [27]

    The constrained bilinear form and theC-numerical range.Linear Algebra and its Applications, 56:195–206, 1984

    Nam-Kiu Tsing. The constrained bilinear form and theC-numerical range.Linear Algebra and its Applications, 56:195–206, 1984. Department of Mathematics and Mathematical Statistics, Umeå University, SE-901 87 Umeå, Sweden Email address:per.ahag@math.umu.se F aculty of Mathematic...

    1984

Pith tools

Reviewed August 28, 2026 · model on record in the stance chip above.