REVIEW 5 major objections 3 minor 3 references
An entire free holomorphic function which is unbounded on the row ball
T0 review · 5 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read A free holomorphic function can be entire yet unbounded on the row ball in more than one variable.
desk verdict A short, credible counterexample answering an open question, but the proof as written is a sketch with two or three fillable gaps. 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 construction is carried by three ingredients: (1) the free topology, whose basic open sets are $B_\delta=\{X:\|\delta(X)\|<1\}$ for noncommutative polynomial matrices $\delta$; (2) the $H^2$ norm on noncommutative polynomials, defined as the square root of the sum of squared coefficient norms, which is bounded above by the sup norm on the row ball; and (3) a sequence of homogeneous polynomials $p_n$ with $\|p_n\|_{H^2}=1$ that vanish identically on $d$-tuples of $n\times n$ matrices. The function is $f=\sum_{k=1}^{\infty}q_k$, where $q_k=\prod_n p_n^{\lfloor k/(2^n\deg p_n)\rfloor}$. The degree of $q_k$ is at most $k$, and the proof estimates the sup norm of $q_k$ on a carefully chosen $B_\delta$ by $r^{2m\deg p_m}/2^k$, yielding a convergent geometric bound for $f$ on that neighborhood.
What would settle it
For $d=2$, write out the promised polynomials $p_n$ explicitly, then compute the $H^2$ norm of $q_1$ and $q_2$ directly; if any $\|q_k\|_{H^2}\neq 1$ or if the series $\sum_k q_k$ has finite $H^2$ norm, the proof's unboundedness argument fails. Alternatively, evaluate the first few partial sums on a sequence of row contractions that concentrate on growing matrix sizes and check numerically whether the norms diverge.
Extended reading notes
Core claim
The central discovery is a counterexample: for $d>1$, there is an entire free holomorphic function $f\colon \mathbb{M}_d\to\mathbb{M}$ such that for every $X\in\mathbb{M}_d$ there is a basic open set $B_\delta$ in the free topology containing $X$ on which $f$ is bounded, yet $f$ is unbounded on the row ball $\mathcal{C}_d$. The function is given explicitly as an infinite sum of products of homogeneous polynomials $p_n$ that vanish on $M_d^n$; the construction forces $\|f\|_{H^2}=\infty$, and since the $H^2$ norm is bounded above by the sup norm on the row ball, $f$ must be unbounded there. The paper also notes that the same example shows the limits of the free uniform approximation theorem, since $r\mathcal{C}_d$ is not compact in the free topology and $f$ cannot be uniformly approximated by polynomials on it.
Load-bearing premise
The proof depends on the unverified assertions that homogeneous polynomials $p_n$ of $H^2$ norm 1 vanishing on $M_d^n$ exist, and that the products $q_k$ built from them each have $H^2$ norm 1; the latter is not generally true for products of unit-norm homogeneous polynomials.
Editorial extensions
If this is right
- For $d>1$, the row ball is not a set on which every entire free holomorphic function is bounded, even though it is bounded in norm.
- Local boundedness in the free topology is strictly weaker than boundedness on the row ball; the free topology does not control growth along the row ball.
- The scaling $r\mathcal{C}_d$ of the row ball is not compact in the free topology for $r\ge 1$, as shown by the failure of uniform polynomial approximation for this $f$.
- The construction yields an entire free holomorphic function with infinite $H^2$ norm, so the $H^2$-to-sup-norm inequality cannot be reversed.
Reading between the lines
- The paper's deferred combinatorial step could be made explicit for small $d$; doing so would let one test numerically whether the $H^2$ norm of each $q_k$ is exactly 1, which is the proof's main unverified assertion.
- The same product-and-sum technique may generate free holomorphic functions with arbitrarily prescribed unbounded growth on other free-semialgebraic sets, provided one can find polynomials vanishing on the appropriate finite-level sets.
- A natural open question is whether the constructed $f$ is also unbounded on the row ball when restricted to matrices of a single fixed size $n$; the paper's argument only shows unboundedness across all sizes.
- If a more quantitative proof is found, it might yield concrete sequences of row contractions on which $\|f(X)\|$ grows, giving explicit witnesses of the failure of local-to-global boundedness.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims to construct, for d > 1, an entire free holomorphic function f on the matrix universe M_d that is locally bounded in the free topology but unbounded on the row ball C_d. The construction chooses, for each n, a nonzero homogeneous free polynomial p_n of H^2 norm 1 that vanishes on M_d^n, forms products q_k of powers of the p_n, and defines f as the infinite sum of the q_k. The author argues that f is unbounded on C_d because its H^2 norm is infinite, and that f is locally bounded by estimating the q_k on specially chosen basic sets.
Significance. If the construction is made fully rigorous, the result is significant: it answers a question attributed to Shamovich and shows that local boundedness in the free topology does not imply boundedness on the row ball, even for entire free functions. It also sharpens the picture around the Agler-McCarthy Oka-Weil theorem and the paucity of compact sets in the free topology. The paper is refreshingly brief and the underlying idea is attractive, but the proof as written is a sketch with several load-bearing gaps; the good news is that these gaps appear fillable within the scope of a revision.
major comments (5)
- [Proposition 1, proof, existence of p_n] The assertion that for d > 1 there exists a nonzero homogeneous polynomial p_n with H^2 norm 1 vanishing on M_d^n is deferred to a 'delightful combinatorial exercise.' This is not a negligible detail: for d = 2 it is essentially a polynomial identity theorem for M_n (for instance, a two-variable PI or a specialization of the Amitsur-Levitzki theorem). Since the entire construction depends on the existence of these p_n, a proof or an exact reference must be supplied.
- [Proposition 1, proof, claim ||q_k||_{H^2}=1] The equality ||q_k||_{H^2}=1 is stated without justification. It is true for this construction, but only because the p_n are homogeneous: for homogeneous free polynomials, the H^2 norm of a product equals the product of the H^2 norms, since each word of the appropriate length has a unique split at the degree boundary. This fact must be stated explicitly, because for general norm-one polynomials the H^2 norm of a product is not automatically one.
- [Proposition 1, proof, line '∞ = ||f||_{H^2} ≤ ||f||_{C_d}'] This line is the decisive step for unboundedness, and as written it is unjustified. Since f is an infinite sum, one must first prove that the infinite H^2 norm is actually infinite; in particular, one must rule out cancellation among the q_k in the same degree. This can be done by noting that the exponent vectors e_n(k)=floor(k/(2^n deg p_n)) are monotone in k, so equal weighted degrees force e_n(k)=e_n(l) for all n, and the n=1 term forces infinitely many distinct degrees. The proof must also justify applying the classical H^2-to-C_d norm inequality to a formal power series with possibly infinite H^2 norm: if f were bounded on C_d, then its H^2 norm would be finite. Neither of these points appears in the manuscript.
- [Proposition 1, proof, well-definedness of f] The sentence 'The function f well-defined for all inputs as the terms in the series are eventually zero' is false as stated. For X of size N, the factors p_n(X) with n < N generally do not vanish, and they appear in q_k(X) with exponents growing in k, so the summands q_k(X) are not eventually zero. The later local-boundedness estimate, once corrected, would give uniform convergence and hence well-definedness, but the manuscript does not present the argument in that form.
- [Proposition 1, proof, local boundedness estimate] The displayed estimate for ||q_k||_{Bδ} contains sign and direction errors. With Bδ defined using B_{(2r)^{2m deg p_m}} p_m (positive exponent), the bound ||p_m||_{Bδ} ≤ (2r)^{2m deg p_m} does not yield the reciprocal (1/(2r))^{...} used in the next line; the neighborhood would need a negative exponent to make the estimate produce geometric decay. In addition, the final inequality r^k( ... ) ≤ r^{2m deg p_m}/2^k has the wrong direction for r ≥ 1, since (2r)^N ≥ r^N. These appear to be typographical, but the local boundedness proof is invalid as printed.
minor comments (3)
- [Throughout] There are several typographical slips in the displayed text, including 'TH E ROW BALL' in the title and 'functio n' in the first paragraph; these should be cleaned up in revision.
- [Notation] The notation M_d^n and C_d^m is used without definition; in particular, the phrase 'X ∈ rC_d^m' should be explained, since the row ball C_d is a union over matrix sizes.
- [Final paragraph] The concluding remarks connecting the example to the Augat-Balasubramanian-McCullough compactness theorem and the Agler-McCarthy Oka-Weil theorem are suggestive but not proved; consider marking them explicitly as observations or adding a precise statement.
Circularity Check
No circularity: the construction is self-contained apart from deferred combinatorial lemmas, which are rigor gaps, not circular reductions.
full rationale
The paper's derivation is a direct construction: it postulates homogeneous noncommutative polynomials p_n of H^2 norm 1 that vanish on M_d^n, forms products q_k, sums them into f, and proves local boundedness by estimating the factors on a free-topology basic set. None of these steps fits a parameter to the conclusion; the unboundedness on the row ball is asserted from the inequality ∞ = ||f||_{H^2} ≤ ||f||_{C_d}, which is an application of a classical norm comparison, not a renaming of the target result. The existence of the p_n is deferred to a 'delightful combinatorial exercise' rather than proved, and the claims ||q_k||_{H^2} = 1 and linear degree growth are asserted without the needed orthogonality or multiplicativity argument. These are genuine correctness and completeness gaps, but they are not circularity: no equation in the paper is equivalent by definition to the conclusion, no fitted input is relabeled as a prediction, and no load-bearing premise rests on a self-citation chain. The cited works [1], [2], and [3] are background and contextual comparisons about the free topology and compact sets; the paper does not invoke an author-supplied uniqueness theorem to force its construction. Since the central existence claim is generated by an explicit construction rather than derived from its own target, the circularity score is 0, with the caveat that the manuscript's rigor depends on unstated combinatorial and Hilbert-space facts.
Assumptions & free parameters
assumptions (3)
- domain assumption The classical inequality ||f||_{H^2} <= ||f||_{C_d} holds for free functions on the row ball.
- ad hoc to paper For each n there exists a nonzero homogeneous noncommutative polynomial p_n in d variables that vanishes on all d-tuples of n by n matrices and has H^2 norm 1.
- ad hoc to paper The product q_k of powers of the p_n has H^2 norm exactly 1.
Cite this review
Pith. "Pith review of An entire free holomorphic function which is unbounded on the row ball." pith.science (2026). https://pith.science/paper/YXY4XDR3
@misc{pith2026190806753,
author = {Pith},
title = {Pith review of: An entire free holomorphic function which is unbounded on the row ball},
year = {2026},
howpublished = {\url{https://pith.science/paper/YXY4XDR3}},
note = {Machine review of arXiv:1908.06753}
}
abstract
We give an entire free holomorphic function $f$ which is unbounded on the row ball. That is, we give a holomorphic free noncommutative function which is continuous in the free topology developed by Agler and McCarthy but is unbounded on the set of row contractions.
Reference graph
Works this paper leans on
- [1]
- [2]
-
[3]
William Helton, Igor Klep, and Scott McCu llough
Meric Augat, J. William Helton, Igor Klep, and Scott McCu llough. Bianalytic maps between free spectrahedra. Mathema- tische Annalen , 371(1):883–959, Jun 2018. 2
work page 2018
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.