Free Bertini's theorem and applications

Jurij Vol\v{c}i\v{c}

REVIEW 3 major objections 6 minor 22 references

Free Bertini's theorem and applications

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A noncommutative polynomial whose shifts $f-\lambda$ factor for infinitely many $\lambda$ must be a composition $p(h)$ of a univariate polynomial $p$ and a noncommutative polynomial $h$.

desk verdict Genuine free-algebra Bertini theorem with a clean proof and solid applications; only minor exposition gaps. read the letter →

arxiv 1908.08948 v1 pith:NXOEDBK7 submitted 2019-08-23 math.RA

Jurij Volčič This is my paper · ORCID

classification math.RA MSC 16U3013P0547A5652A05

keywords freeirreducibilitytheoremnoncommutativepolynomialfactorizationcompositioneigenlevelsetquasiconvexityalgebracentralizer

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 a free-algebra analog of the classical irreducibility theorem of algebraic geometry: for a nonconstant noncommutative polynomial $f$, the shifted polynomials $f-\lambda$ can factor nontrivially for infinitely many scalars $\lambda$ only when $f$ is itself a composition $f=p\circ h$ of an ordinary univariate polynomial $p$ and a noncommutative polynomial $h$. Equivalently, a non-composite $f$ has $f-\lambda$ irreducible for all but finitely many $\lambda$. The proof rests on the centralizer theorem for free associative algebras: when $f$ is not a composition, the elements commuting with $f$ in the universal skew field are exactly the rational functions of $f$. This theorem is then used to give algebraic certificates for when eigenlevel sets of matrix evaluations are nested or equal, and to classify locally quasiconvex noncommutative polynomials as either negative sums of hermitian squares or univariate compositions of a convex quadratic form.

What carries the argument

The central object is the centralizer theorem for free associative algebras: for a non-composite $f$, the subalgebra of elements commuting with $f$ consists exactly of the polynomials in $f$, and in the universal skew field of fractions the centralizer is $\mathbb{F}(f)$. Around this, the proof uses stable association, the notion that two polynomials occupy the same two-sided ideal up to invertible matrix factors in the free ideal ring, together with the finiteness of stable-associated classes and a degree-reduction lemma. Those ingredients force right factors of $f-\lambda$ across different $\lambda$ to be proportional. For the geometric applications, irreducibility results for free loci of noncommutative polynomials convert the algebraic irreducibility of $f-\lambda$ into reducedness and irreducibility of eigenlevel hypersurfaces.

What would settle it

Exhibit a nonconstant $f$ over an algebraically closed field and an infinite set of scalars $\lambda$ such that $f-\lambda$ factors in the free algebra for every $\lambda$ in the set, yet $f$ is not a composition $p\circ h$ with $\deg p>1$; equivalently, find a non-composite $f$ whose centralizer in the universal skew field contains an element outside $\mathbb{F}(f)$.

Watch

Extended reading notes

Core claim

Theorem 3.2 asserts that for an algebraically closed field and a nonconstant $f$ in the free algebra, four conditions coincide: (i) $f-\lambda$ factors for infinitely many scalars $\lambda$; (ii) $f-\lambda$ factors for every $\lambda$; (iii) the centralizer of $f$ is strictly larger than $\mathbb{F}[f]$; and (iv) $f$ is composite over $\mathbb{F}$, i.e. $f=p\circ h$ with $\deg p>1$ and $h$ noncommutative. The nontrivial direction uses a lemma showing that if $f b_1=b_1 g$ and $f b_2=\alpha b_2 g$ with degrees below $\deg f$, then $\alpha=1$ and $b_2$ is a scalar multiple of $b_1$; the centralizer theorem provides the proportionality. Applying this to the finitely many stable-associated candidates for the right factors of $f-\lambda$ gives a contradiction unless $f$ is composite. Consequently, non-composite $f$ have $f-\lambda$ irreducible for all but finitely many $\lambda$; with existing irreducibility results for free loci, their eigenlevel sets are reduced irreducible hypersurfaces for large matrix sizes. The same machinery yields the eigenlevel inclusion certificate and the locally quasiconvex classification.

Load-bearing premise

The proof leans on the centralizer theorem: for a non-composite $f$, everything in the universal skew field that commutes with $f$ is a rational function of $f$; if that statement failed, the proportionality argument would break and Theorem 3.2 would lose its proof.

Editorial extensions

If this is right

For every non-composite $f$, the family $\{f-\lambda\}$ is irreducible for a cofinite set of $\lambda$; factorization of shifts is a compositional phenomenon.
Eigenlevel inclusion is certified algebraically: if every eigenlevel set of $f$ lies in an eigenlevel set of $g$, then $g=p(h)$ and $f a = a h$ for some nonzero $a$; equality of eigenlevel sets holds exactly when $f a = a g$.
A homogeneous $f$ has $f-1$ factor in the free algebra if and only if $f=f_0^n$ for some $n>1$ and homogeneous $f_0$.
Locally quasiconvex symmetric polynomials are classified: either $-f$ is a sum of hermitian squares, or $f=p(\ell_0+\ell_1^2+\cdots+\ell_m^2)$ with explicit conditions on $p$; the convexity then holds for every $\lambda>0$.

Reading between the lines

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

Because stable association reduces to a finite linear system, the proof suggests a direct computational test for compositeness: search for degree-reducing witnesses by solving linear equations in the free algebra. The paper does not present such an algorithm.
The eigenlevel certificate $f a = a g$ is an intertwining relation; a natural extension is to ask whether a similar criterion governs inclusion of spectra of noncommutative rational or analytic free functions.
The locally quasiconvex classification may imply that every locally quasiconvex $f$ whose positivity domain is proper is LMI-representable; checking small examples in low degrees could reveal whether the classification survives beyond polynomials.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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

The paper proves a true free-algebra analogue of Bertini: if f is a noncommutative polynomial and f-λ factors for infinitely many λ, then f is composite (a univariate composed with a noncommutative h). That equivalence - actually four equivalent conditions in Theorem 3.2 - is new and is exactly the kind of result that makes the free algebra factorization theory feel less piecemeal. The proof is also better than one might expect: Lemma 3.1 reduces the infinite family of factorizations to a stable-association finiteness statement, and the centralizer argument via Bergman and Cohn is careful. I checked the step where b2 b1^{-1} is shown to be a rational function of f; the determinant argument plus degree comparison does the job. I did not find a hole.

The applications are meaningful. The eigenlevel-set characterization (Theorem 4.3) gives an algebraic certificate for inclusion of eigenlevel sets, and the quasiconvexity classification (Theorem 5.4) is a clean dichotomy: either -f is a sum of hermitian squares, or f is a univariate composed with a convex quadratic. The paper also contains a nice example (Example 4.5) showing the degree bound in Lemma 2.2 is optimal, which I appreciate.

Soft spots are minor and mostly expository. In Theorem 5.4, the claim that "since f is symmetric, h is also symmetric because it is unique up to a scalar multiple" is too terse. The uniqueness comes from Lemma 3.1, but the passage from C to R needs a sentence or two; as written it reads like an assertion rather than a proof. Similarly, several applications lean on [HKV18, Theorem 4.3] and [HKMV, Theorem 1.5]; those are legitimate external tools, but the paper doesn't spell out exactly how the constant-term perturbation is handled in Corollary 4.1. A referee should ask for these two points to be expanded, not for any mathematical change.

The citation pattern is fine. The author uses his own prior work with Helton, Klep, and McCullough, but as input to the applications, not to define the target conclusions. The core theorem stands on Bergman and Cohn.

Who is this for: anyone working in free analysis, noncommutative factorization, or free real algebraic geometry. It deserves a serious referee. My own verdict would be accept after minor revision.

Referee Report

3 major / 6 minor

Summary. The paper proves a free-algebra analogue of Bertini's irreducibility theorem (Theorem 3.2): for a nonconstant noncommutative polynomial f, if f - λ factors for infinitely many scalars λ, then f = p ∘ h for some noncommutative h and univariate p of degree greater than 1; equivalently, a non-composite f has irreducible fibers f - λ for all but finitely many λ. The proof builds on Bergman's centralizer theorem and Cohn's stable-association theory, with Lemma 3.1 as the key technical step. Two applications are developed: Theorem 4.3 characterizes inclusion of eigenlevel sets of matrix evaluations by the right-multiple condition f a = a h with g = p(h), and Corollary 4.4 characterizes equality of eigenlevel sets by f a = a g. Theorem 5.4 classifies locally quasiconvex symmetric free polynomials as either polynomials -f that are sums of hermitian squares or univariate compositions of a convex quadratic form of linear terms.

Significance. Assuming the cited external theorems, Theorem 3.2 is a valuable structural result: it gives a clean centralizer criterion for compositeness and connects classical Bertini irreducibility with noncommutative factorization. The paper is generally careful and self-contained modulo standard results of Bergman and Cohn, and it introduces useful auxiliary facts such as the degree bound in Lemma 2.2. The eigenlevel and quasiconvexity applications are natural and likely to be useful in free real algebraic geometry. The main weakness is not in the central theorem, which appears sound, but in several application-side arguments that are too compressed and need explicit justification before the paper can be accepted.

major comments (3)

[Proposition 5.1, proof] The assertion that \tilde f := f(y_1 + y_1^*, ..., y_d + y_d^*) is irreducible in C<y,y*> because f is irreducible over C is not justified. A factorization of \tilde f could in principle become a trivial factorization of f under the substitution y_j = y_j^* = x_j/2, since the substituted factors might become constants or zero. Because Proposition 5.1 is load-bearing for Theorem 5.4, please provide a proof of this irreducibility statement or cite a result covering the complexification step.
[Theorem 5.4, proof (i)⇒(iii)] After applying Theorem 3.2, the proof asserts f = p(h) with p ∈ R[t] and h ∈ R<x>, and then says that since f is symmetric, h is also symmetric because it is unique up to a scalar multiple. Two load-bearing points are passed over. First, Theorem 3.2 is stated over an algebraically closed field, so one must justify Galois descent from C to R for p and h. Second, uniqueness of h only yields h^* = c h with |c| = 1, and the antisymmetric possibility h^* = -h must be ruled out using the standing assumption that -f is not a sum of hermitian squares. Please replace this sentence with a lemma or detailed argument establishing the descent and the symmetrization.
[Theorem 4.3, proof (i)⇒(ii)] The normalization step 'By comparing det(h_1(Ω^{(n)}) - λ_1 I), det(h_2(Ω^{(n)}) - λ_2 I) one can replace h_2 with αh_2 + β' is too compressed. Equality of eigenlevel hypersurfaces gives equality of the zero sets of the two determinants, not equality of the determinants themselves, and the affine normalization is then used to reach equation (4.2), which supplies the input for stable association via [HKV18, Theorem 4.3]. Please expand this reduction, including the justification of the determinant comparison for large n.

minor comments (6)

[Theorem 3.2, statement] The sentence 'Let /CZ be the algebraic closure of a field /CZ' uses the same symbol for the base field and its algebraic closure; please denote the base field and closure by different letters.
[Introduction, page 1] There is a typo in 'over a an algebraically closed field'; it should read 'over an algebraically closed field'.
[Theorem 5.4, statement] The statement begins 'Le f' and should read 'Let f'.
[Theorem 4.3, proof (ii)⇒(i)] The equality h(Ω^{(n)}) = a(Ω^{(n)})^{-1} f(Ω^{(n)}) a(Ω^{(n)}) requires that a(Ω^{(n)}) be invertible for large n; this should be justified via Lemma 2.4 applied to the nonzero polynomial a rather than asserted.
[Corollary 4.1 and equation (4.1)] The formula refers to a hypersurface in M_n(/CZ)^g but the dimension parameter should be d (the number of variables); please correct the notation.
[References] The reference [HKMV] is cited without a year or arXiv number; please add complete publication data or mark it clearly as a preprint.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the free Bertini theorem is anchored in external centralizer and factorization theorems, and the applications' use of the author's prior work is as a tool, not as a definitional substitute.

full rationale

The main theorem (Theorem 3.2) is not derived from its own conclusion. Lemma 3.1 uses Bergman's centralizer theorem [Ber69, Theorem 5.3] and Cohn's factorization theory [Coh06, Theorem 7.9.8 and Proposition 3.2.9] to prove that two solutions of fb = bg are proportional; these are external results, not equivalent to the free Bertini statement. The proof of (iii) implies (iv) is explicitly a restatement of Bergman's theorem, but the core implication (i) implies (iii) is a new argument showing that infinitely many factorizations force a larger centralizer. Applications invoke the author's prior work ([HKV18], [HKV], [HKMV]) as black-box tools; while these citations are load-bearing for the applications, they are not being re-derived by construction from the target conclusions, and they do not introduce fitted parameters or definitions equivalent to the results. There is no self-definitional reduction, no fitted input renamed as prediction, and no uniqueness claim that is both imported from the authors and used to forbid alternatives. Therefore no circular step is exhibited.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

No free parameters are fitted; the paper is a proof-based mathematics paper. The central claim rests on established theorems in free ideal rings and free loci, listed above. No new entities are postulated.

assumptions (8)

standard math Bergman's centralizer theorem: the centralizer of a non-composite f in F<x> equals F[f].
Used in Lemma 3.1 and Theorem 3.2; external theorem [Ber69, Theorem 5.3].
standard math The centralizer of a non-composite f in the universal skew field of fractions F(<x>) equals F(f).
Used in Lemma 3.1; cited to [Coh06, Theorem 7.9.8 and Proposition 3.2.9].
standard math Only finitely many polynomials are stably associated to a given f up to scalar multiple.
Proposition 2.1, credited to Bergman in [Coh06, Exercise 2.8.8]; used in Theorem 3.2 to select infinitely many lambda with proportional stable-association witnesses.
standard math For nonconstant f, det f(Omega(n)) is nonconstant for large enough n.
Lemma 2.4, taken from [HKV, Lemma 2.2]; used in Lemma 3.1 to show alpha^n = 1 and hence alpha = 1.
standard math Irreducible free loci of f - lambda are reduced and irreducible hypersurfaces for large n.
Used in Corollary 4.1 and Theorem 4.3; cited to [HKV18, Theorem 4.3].
standard math Irreducible symmetric polynomials with proper convex positivity domain are linear-plus-squares.
Used as Proposition 5.1's main input in Theorem 5.4; taken from [HKMV, Theorem 1.5].
standard math Noncommutative polynomials nonnegative on all symmetric matrix tuples are sums of hermitian squares.
Used in Theorem 5.4 to conclude -f is a sum of hermitian squares when D(lambda - f) = S^d; cited to [Hel02, McC01].
domain assumption Field assumptions: algebraically closed for Theorem 3.2 and Theorem 4.3, characteristic 0 for Theorem 4.3, real closed for Section 5.
Explicit hypotheses of the main theorems.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Free Bertini's theorem and applications." pith.science (2026). https://pith.science/paper/NXOEDBK7

@misc{pith2026190808948,
  author       = {Pith},
  title        = {Pith review of: Free Bertini's theorem and applications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NXOEDBK7}},
  note         = {Machine review of arXiv:1908.08948}
}

abstract

The simplest version of Bertini's irreducibility theorem states that the generic fiber of a non-composite polynomial function is an irreducible hypersurface. The main result of this paper is its analog for a free algebra: if $f$ is a noncommutative polynomial such that $f-\lambda$ factors for infinitely many scalars $\lambda$, then there exist a noncommutative polynomial $h$ and a nonconstant univariate polynomial $p$ such that $f=p\circ h$. Two applications of free Bertini's theorem for matrix evaluations of noncommutative polynomials are given. An eigenlevel set of $f$ is the set of all matrix tuples $X$ where $f(X)$ attains some given eigenvalue. It is shown that eigenlevel sets of $f$ and $g$ coincide if and only if $fa=ag$ for some nonzero noncommutative polynomial $a$. The second application pertains quasiconvexity and describes polynomials $f$ such that the connected component of $\{X \text{ tuple of symmetric $n\times n$ matrices}: \lambda I\succ f(X) \}$ about the origin is convex for all natural $n$ and $\lambda>0$. It is shown that such a polynomial is either everywhere negative semidefinite or the composition of a univariate and a convex quadratic polynomial.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

[1]

Balasubramanian, S

S. Balasubramanian, S. McCullough: Quasi-convex free polynomials , Proc. Amer. Math. Soc. 142 (2014) 2581--2591

work page 2014
[2]

G. M. Bergman: Centralizers in free associative algebras , Trans. Amer. Math. Soc. 137 (1969) 327--344

work page 1969
[3]

Bodin, P

A. Bodin, P. D\`ebes, S. Najib: Irreducibility of hypersurfaces , Comm. Algebra 37 (2009) 1884--1900

work page 2009
[4]

Blekherman, P

G. Blekherman, P. A. Parrilo, R. R. Thomas (eds.): Semidefinite optimization and convex algebraic geometry , MOS-SIAM Ser. Optim. 13, SIAM, Philadelphia, PA, 2013

work page 2013
[5]

P. M. Cohn: Free Ideal Rings and Localization in General Rings , New Mathematical Monographs 3, Cambridge University Press, Cambridge, 2006

work page 2006
[6]

K. R. Davidson, A. Dor-On, O. M. Shalit, B. Solel: Dilations, Inclusions of Matrix Convex Sets, and Completely Positive Maps , Int. Math. Res. Not. IMRN (2017) 4069--4130

work page 2017
[7]

J. W. Helton: ``Positive noncommutative polynomials are sums of squares'', Ann. Math. 156 (2002) 675--694

work page 2002
[8]

J. W. Helton, I. Klep, S. McCullough: The matricial relaxation of a linear matrix inequality , Math. Program. 138 (2013) 401--445

work page 2013

Show all 22 references

[9]

J. W. Helton, I. Klep, and S. McCullough: Free analysis, convexity and LMI domains , Mathematical methods in systems, optimization, and control, 195--219, Oper. Theory Adv. Appl. 222, Birkh\"auser/Springer Basel AG, Basel, 2012

2012
[10]

J. W. Helton, I. Klep, S. McCullough, J. Vol c i c : Noncommutative polynomials describing convex sets , preprint arXiv:1808.06669

arXiv
[11]

Helton, I

J.W. Helton, I. Klep, J. Vol ci c: Geometry of free loci and factorization of noncommutative polynomials , Adv. Math. 331 (2018) 589--626

2018
[12]

Helton, I

J.W. Helton, I. Klep, J. Vol ci c: Factorization of noncommutative polynomials and Nullstellens\"atze for the free algebra , preprint arXiv:1907.04328

1907 arXiv
[13]

Helton, S

J.W. Helton, S. McCullough: Every convex free basic semi-algebraic set has an LMI representation, Ann. of Math. (2) 176 (2012) 979--1013

2012
[14]

D. S. Kalyuzhnyi-Verbovetskyi, V. Vinnikov: Singularities of rational functions and minimal factorizations: the noncommutative and the commutative setting , Linear Algebra Appl. 430 (2009) 869--889

2009
[15]

D. S. Kalyuzhnyi-Verbovetskyi, V. Vinnikov: Foundations of free noncommutative function theory , Mathematical Surveys and Monographs 199, American Mathematical Society, Providence RI, 2014

2014
[16]

S. L. Kleiman: Bertini and his two fundamental theorems , Studies in the history of modern mathematics III, Rend. Circ. Mat. Palermo (2) Suppl. No. 55 (1998) 9--37

1998
[17]

I. Klep, J. Vol c i c : Free loci of matrix pencils and domains of noncommutative rational functions , Comment. Math. Helv. 92 (2017) 105--130

2017
[18]

Marshall: Positive polynomials and sums of squares , Mathematical Surveys and Monographs 146, American Mathematical Society, 2008

M. Marshall: Positive polynomials and sums of squares , Mathematical Surveys and Monographs 146, American Mathematical Society, 2008

2008
[19]

McCullough: Factorization of operator-valued polynomials in several non-commuting variables , Linear Algebra Appl

S. McCullough: Factorization of operator-valued polynomials in several non-commuting variables , Linear Algebra Appl. 326 (2001) 193--203

2001
[20]

I. R. Shafarevich: Basic algebraic geometry 1. Varieties in projective space , 2nd edition, Springer-Verlag, Berlin, 1994

1994
[21]

Schinzel: Polynomials with special regard to reducibility , Encyclopedia of Mathematics and its Applications 77, Cambridge University Press, Cambridge, 2000

A. Schinzel: Polynomials with special regard to reducibility , Encyclopedia of Mathematics and its Applications 77, Cambridge University Press, Cambridge, 2000

2000
[22]

Stein: The total reducibility order of a polynomial in two variables , Israel J

Y. Stein: The total reducibility order of a polynomial in two variables , Israel J. Math. 68 (1989) 109--122

1989

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Balasubramanian, S

S. Balasubramanian, S. McCullough: Quasi-convex free polynomials , Proc. Amer. Math. Soc. 142 (2014) 2581--2591

work page 2014

[2] [2]

G. M. Bergman: Centralizers in free associative algebras , Trans. Amer. Math. Soc. 137 (1969) 327--344

work page 1969

[3] [3]

Bodin, P

A. Bodin, P. D\`ebes, S. Najib: Irreducibility of hypersurfaces , Comm. Algebra 37 (2009) 1884--1900

work page 2009

[4] [4]

Blekherman, P

G. Blekherman, P. A. Parrilo, R. R. Thomas (eds.): Semidefinite optimization and convex algebraic geometry , MOS-SIAM Ser. Optim. 13, SIAM, Philadelphia, PA, 2013

work page 2013

[5] [5]

P. M. Cohn: Free Ideal Rings and Localization in General Rings , New Mathematical Monographs 3, Cambridge University Press, Cambridge, 2006

work page 2006

[6] [6]

K. R. Davidson, A. Dor-On, O. M. Shalit, B. Solel: Dilations, Inclusions of Matrix Convex Sets, and Completely Positive Maps , Int. Math. Res. Not. IMRN (2017) 4069--4130

work page 2017

[7] [7]

J. W. Helton: ``Positive noncommutative polynomials are sums of squares'', Ann. Math. 156 (2002) 675--694

work page 2002

[8] [8]

J. W. Helton, I. Klep, S. McCullough: The matricial relaxation of a linear matrix inequality , Math. Program. 138 (2013) 401--445

work page 2013