Independent analysis review

What Lean checks

Lean proves the original sampled-sequence and weighted transform-class theorems for every positive degree and weight. The proof uses all finite ordered Toeplitz minors and proves both centered disks and attained exact radii. All 17 modules passed strict sources, the complete axiom audit, and stock-kernel replay.

21 November 2026. Reviewer: analysis lane. Reviewed the complete written proof, including the corrected finite Vandermonde identity in Section 5. The all-degree, all-positive-weight proof passes this review. This is an internal mathematical review, not external peer review or a guarantee of priority.

Exact scope

The original source is Anna Vishnyakova, Polynomials interpolated totally positive sequences, arXiv:2607.06207v1, Open Problem 6.2, printed page 24. It asks whether the supremum of the largest complex-root modulus is n for every degree n in the sampled-total-positivity class.

I independently read the weighted source, Polynomially Deformed Normalized Pochhammer Sequences Having Generating Functions With Only Real Non-positive Zeros, arXiv:2608.03723v1, in its author-uploaded full text: https://www.researchgate.net/publication/411305396_Polynomially_Deformed_Normalized_Pochhammer_Sequences_Having_Generating_Functions_With_Only_Real_Non-positive_Zeros

Definition 1.4, formula (8), uses exactly the map L_beta in PROOF.md. Open Problem 7.1 defines the class U_n by the condition that L_beta(P) have all roots in [-1,0]; Open Problem 7.2, printed page 27, conjectures R_n=beta+n-1. The proof covers precisely this class for every beta>0 and every positive integer n. No extra nonnegative-coefficient condition on P, real-rootedness condition on P, or integer restriction on beta is inserted. Normalizing the leading coefficient of L_beta(P) is valid because nonzero scalar multiplication does not change either root set.

The original and weighted forms are one conjecture family and one count.

Mathematical checks

  1. Normalization and generating function. The family in (2) has degree exactly n with leading coefficient 1/(beta)_n, even for complex parameters. On the diagonal, multiplication by (beta)_n makes it monic. In the formal exponential generating series, setting n=k+j produces the binomial series with parameter beta+k, and then the second binomial expansion gives (5). Differentiating that expression gives exactly the stated recurrence. No convergence or orthogonality assumption is needed.

  2. Tridiagonal representation. The diagonal entries of J are j-(2j+beta)t. The product of its two off-diagonal entries at position j equals -t(1-t)j(j+beta-1); the minus sign in the determinant recurrence therefore gives the positive second term in (4). Since beta>0, K is real symmetric. Complex t and a nonnormal J cause no problem: an eigenvector still satisfies vJv=z, while vDv and v*Kv are real because D and K are Hermitian.

  3. Quadratic estimate. Summing the pairwise elementary inequalities gives exactly |v*Kv| <= 2d+beta-(n+beta-1)|v_(n-1)|^2 <= 2d+beta. The empty-sum n=1 case also agrees with this equality. Consequently c=d+beta/2 is strictly positive and kappa is real in [-1,1].

  4. Reciprocal-segment geometry. For Im u<0 the two endpoints have product one and opposite nonzero imaginary parts. Choosing r in the upper half-plane gives |r|<1 directly from the imaginary part of r+1/r. The real-axis intersection is 2 Re(r)/(1+|r|^2), strictly between -1 and 1. The entire upper part of the segment is therefore strictly inside the unit disk. This is valid for either square-root branch, including arbitrarily large complex t. It gives the asserted diagonal zero exclusion.

  5. Grace–Walsh–Szegő. For fixed z, the parameter polynomial is symmetric and multiaffine. The open upper half-plane is a convex circular domain. I independently read Borcea–Brändén, arXiv:0809.3087, Theorem 1.1, printed page 4: https://arxiv.org/pdf/0809.3087 Its conclusion supplies a common upper-half-plane parameter with the same value. A zero with all parameters in the upper half-plane would contradict the diagonal exclusion. Convexity makes any separate full-degree hypothesis unnecessary.

  6. Boundary specialization. A pointwise limit of nonzero numbers would be insufficient. The proof correctly uses Hurwitz on the connected open domain Omega_(n,beta) in the z variable. The polynomials converge locally uniformly there. Their common nonzero leading coefficient excludes an identically zero limiting function. This validates the real-parameter specialization, including repeated parameters and endpoints. Real coefficients give the lower-half-plane conclusion by conjugation.

  7. Real roots and reflection. For z>n-1 all falling-factorial terms in (2) are positive and the coefficients are nonnegative. The corrected finite Vandermonde identity has no alternating factor inside the sum. In applying it to Q(-w-1), the global factor (-1)^j comes from the monomial w^j, yielding exactly the reflection identity (13). This excludes roots below -beta-n+1. Thus the real roots lie in the diameter of the claimed disk.

  8. Sharpness. Parameters all equal to one give (z+beta)_n/(beta)_n. Its root -(beta+n-1) realizes the upper modulus bound from the disk. The supremum is attained. The conclusion concerns all roots, not only nonreal roots or fixed degree.

Correction found during review

The first written draft put (-1)^k inside the finite Vandermonde sum in Section 5. Already j=1 would contradict that display. The author removed that factor and explicitly placed (-1)^j in the subsequent expansion of Q(-w-1). I reread the corrected passage. This was an error in a supporting displayed identity, not an obstruction to the reflection identity or the main argument.

Additional priority checks

The following substantial earlier work was located by global GitHub searches for the exact arXiv identifiers and inspected directly.

The shared repository HISTORY.md and all-state GitHub issue search for Vishnyakova were checked. They expose no older root-radius resolution or exclusion; issue 182 is this investigation. Both search engines were also used for Meixner complex zeros, falling-factorial zero localization, Grace/polarization disk bounds, the exact source identifiers, and the author's root-radius statement. These are bounded searches; the absence of a located earlier full proof does not establish exhaustive priority.