Pith. sign in

REVIEW 3 major objections 4 minor 13 references

Sheffer sequences with zeros on a line

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

Pith's one-line read Zeros of a broad Sheffer polynomial family fall on one vertical line

desk verdict A genuine generalization of Bump et al. with a real proof gap: the final zero-counting step is outsourced to an unpublished companion, so the main theorem is not yet established as written. read the letter →

arxiv 2508.18229 v1 pith:GPXGQCEM submitted 2025-08-25 math.NT math.CV

classification math.NTmath.CV MSC 05A1505A4030C1530E15
keywords SheffersequenceszerosonalineMellintransformRiemannzetafunctionLaguerrepolynomialsRiordanmatricessaddle-pointmethodinterlacing
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

The paper's central claim is that a wide class of Sheffer sequences—polynomial families generated by (1−α_0z)^(p+s)(1+α_0z)^(p*−s)∏_{i=1}^N(1−α_i²z²)^(p_i)—have all but finitely many zeros on the vertical line Re(s)=c=(p*−p)/2. This extends a result of Bump and collaborators, which had been proved only for the single-factor generating function behind the generalized Laguerre polynomials. The extension matters because it shows the 'zeros on a line' phenomenon survives when the generating product is multiplied by further even factors, and because a particular member of the family is linked, through a scaled Mellin transform, to the Riemann zeta function: the transform of the sum ψ*_j(x)=∑_{n≥1}Φ_j^*(n√x) equals q_j(s/2)π^(−s/2)Γ(s/2)ζ(s). A sympathetic reader would care because this makes the zero location of these polynomials a concrete cousin of the Riemann hypothesis.

What carries the argument

The central object is the Sheffer sequence {h_n(s)} for the pair (g(z), f(z)) with g(z)=∏_{i=0}^N G_i(z) and f(z)=log((1−α_0z)/(1+α_0z)). The load-bearing identity is the exponential Riordan matrix factorization ⟨g(z),z⟩ = ⟨∏_{i=1}^N G_i(z),z⟩ ⟨G_0(z),z⟩, which writes h_n(s)=∑_{k=0}^n n!b_{n−k}q_k(s)/k! and isolates the two-factor core q_k(s) whose zeros are tractable. That core is studied through the contour integral p_n(t)=∫_{Γ_2}ψ(z)e^{nϕ(z,t)}dz, where ϕ(z,t)=it Log(1+z)−it Log(1−z)−Log z; the saddle-point curve z(y) with −y²=ϕ(z(y),t)−ϕ(ζ,t) and the sign of Im z′(y) are what make the central piece dominate the tails. The zeta connection rides on the Mellin-transform chain (2.13), (2.19)

What would settle it

For a concrete case with p+p*>0 (e.g. p=0, p*=1, N=0, α_0=1), compute the zeros of q_n(s) for n up to, say, 40 and compare the number of off-line zeros with 2⌈c+p⌉; additionally, recompute the constrained extrema in Lemma 18 with high precision—if the minimum t is less than 0.924256... or the maximum exceeds 0.707107..., the sign of Im z′(y) can be wrong and the tail estimate fails.

Watch

Extended reading notes

Core claim

Theorem 1 states that for real parameters α_i, p_i, p, p* with |α_0|<|α_1|<...<|α_N|, the polynomials h_n(s) defined by relation (2.4) have all their zeros on Re(s)=c=(p*−p)/2 when p*+p≤0, and all but a specified finite set of 2⌈c+p⌉ zeros there when p*+p>0. The proof splits h_n through an exponential Riordan factorization into a finite linear combination of intermediate polynomials q_k(s) generated by the two-factor core (1−z)^(p+s)(1+z)^(p*−s); the zero-location theorem is proved for q_k by an integral representation on a Hankel contour, a saddle-point deformation through ζ(t)=−it+√(1−t²), and saddle-point asymptotics, with the off-line exception count obtained by comparing the change of a

Load-bearing premise

The final zero count and the asymptotic tail estimates rely on arguments in the authors' companion manuscript [5], which is not yet available, and on numerical Lagrange-multiplier extrema (0.924256... and 0.707107...) in Lemma 18; if the analogy to [5] fails for the generalized family, the exceptional-zero count 2⌈c+p⌉ is not proved.

Editorial extensions

If this is right

  • For p*+p≤0 every q_n(s) has all of its zeros on Re(s)=c; for p*+p>0, the only off-line zeros are the 2⌈c+p⌉ exceptions counted in Theorem 1.
  • The q_n(s) interlace after the affine change s=c+it: q_n⪯q_{n+1} for n≥1, so they behave like real-rooted polynomials in the vertical coordinate.
  • The scaled Mellin identity ∫_0^∞ ψ*_j(x)x^(s/2−1)dx = q_j(s/2)π^(−s/2)Γ(s/2)ζ(s) means information about zeros of these transforms is information about zeros of ζ(s).
  • The limiting density of zeros of h_n(c−int) on t∈(0,1) is D(x)=1/(2π) ln((1+√(1−x²))/(1−√(1−x²))).
  • Bump's local Riemann hypothesis is recovered as the special case N=0, α_0=1, p=−1−α/2, p*=−α/2.

Reading between the lines

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

  • Inference: the exceptional count 2⌈c+p⌉ suggests a phase transition controlled by p*+p; a cheap test is to compute the zeros of q_n(s) for p=0, p*=1, N=0 for moderate n and count off-line zeros.
  • Inference: the Mellin identity points to a family of Dirichlet-series-like objects whose non-vanishing on Re(s)=1/2 could be equivalent to a Riemann hypothesis for that family; constructing an explicit Euler product for ∫ψ*_j would make that precise.
  • Inference: the proof's tail estimates hinge on the numerically verified extrema in Lemma 18; replacing those Lagrange-multiplier checks with an analytic inequality would remove a computational dependency from the theorem.
  • Inference: the same Riordan splitting should in principle work for other products of even factors, so a natural conjecture is that the zero-line statement persists for more general palindromic generating functions.
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.

Referee Report

3 major / 4 minor

Summary. The paper claims a substantial generalization of Bump et al.'s result on zeros of Sheffer polynomials. For real parameters satisfying the hypotheses of Theorem 1, the sequence {h_n(s)} generated by (1 - α_0 z)^{p+s}(1 + α_0 z)^{p*-s} ∏_{i=1}^N (1 - α_i^2 z^2)^{p_i} is asserted to have all zeros on the vertical line Re(s) = (p* - p)/2, except possibly for 2⌈c+p⌉ zeros when p*+p > 0. The authors also derive recurrences for the auxiliary polynomials q_n(s), prove an interlacing result for p*+p ≤ 0, establish a Mellin-transform identity connecting q_n with the Riemann zeta function, and compute a limiting zero density. The proof of Theorem 1 follows the contour-deformation/saddle-point strategy of Bump et al., but several decisive asymptotic lemmas and the final zero-counting step are explicitly delegated to the unpublished companion manuscript [5].

Significance. If completely proved, the zero-location theorem would be a meaningful extension of the known local Riemann hypothesis results, and the Mellin-transform factorization in Theorem 11 is an explicit and verifiable identity. The paper has clear strengths: the recurrences in Theorems 4 and 5 are correctly derived, the Riordan-matrix viewpoint is informative, and the calculation of the limiting density is elegant. However, as submitted, the main theorem is not proven within the paper: key Lemmas 14 and 17 depend on [5], the final comparison with Eq. (3.21) of [5] is not shown, and the Appendix relies on unsupported numerical extrema. These are load-bearing gaps in the central claim, so the significance of the paper cannot be assessed as a proof.

major comments (3)
  1. [§3.3 (after Eq. (3.17))] The proof of Theorem 1 is not contained in the paper. Lemmas 14 and 17 invoke Lemma 10 and Lemmas 10/12 of the unpublished manuscript [5], and the final jump from (3.17) to the exceptional-zero count 2⌈c+p⌉ is left to 'arguments completely analogous to those in [5]' without showing the comparison to Eq. (3.21) of [5]. Since [5] is not available to the reader, the claim that all but 2⌈c+p⌉ zeros lie on Re(s)=c is unproven as submitted. This is the central result, so the gap is load-bearing.
  2. [Appendix, Lemma 18] The proof that Im z'(y) > 0 for y > 0 rests on two 'numerical' Lagrange-multiplier extrema (0.924256... and 0.707107...) computed from algebraic constraints. No rigorous certificate, interval bound, or exact rational enclosure is provided. Floating-point output is not a proof. This inequality is used in Lemma 15 to show the central saddle-point section dominates the tails, and hence Lemma 16 also depends on it. Without a rigorous proof of these extrema, the saddle-point argument is incomplete.
  3. [§2.1, Theorem 7] The proof begins 'It can be readily shown that bq_n(s) ∈ R[s] and all zeros of bq_n(s) are real.' The real-rootedness is exactly the conclusion of Corollary 3/Theorem 1 for p*+p ≤ 0, which is not proved at that point and is the main theorem of the paper. As written, Theorem 7 is either circular or lacks the key input. The theorem should be stated after Theorem 1, or the real-rootedness should be proved independently.
minor comments (4)
  1. [Throughout] There are numerous typos and corrupted display artifacts ('poynomials', 'fucntions', 'generealized', 'Appel', 'NY' in the abstract/intro). The paper needs careful proofreading.
  2. [Lemma 13] The assertion tπ/2 > ln z* for all t ∈ (0,1) is justified only by 'perhaps with a CAS'. This is elementary but should be proved explicitly.
  3. [Theorem 11 proof] In the last line of the proof, the index switches from j to n: 'qn(s/2)' should be 'qj(s/2)'. Also, the interchange of summation and integration in the derivation of (2.23) should be justified by a convergence or dominated-convergence argument.
  4. [§3 notation] The symbol p_n(t) is used for the Hankel integral in (3.4), which conflicts with the earlier use of p_n for polynomial sequences. Consider denoting the integral by I_n(t) or similar.

Circularity Check

1 steps flagged · score 6.0 of 10

Theorem 1's decisive zero-counting step is outsourced to the same authors' unpublished companion [5], so the central claim rests on a load-bearing self-citation.

  1. self citation load bearing [Section 3.3, final paragraph after equation (3.17); cf. Lemma 14 and Lemma 17 proofs]
    "The completion of the proof of Theorem 1, i.e. the claim about the location of the zeros of the polynomials pn(t), is obtained using arguments completely analogous to those in [5], after comparing the right hand side of equation (3.17) to the absolute value of (3.21) in [5]."

    This is the step that converts the asymptotic change-of-argument in (3.17) into the claimed zero count on Re(s)=c except for 2⌈c+p⌉ zeros. The conversion is not carried out in the paper; it is asserted by analogy to [5]. The same dependency appears earlier: Lemma 14's Hankel asymptotics invoke 'Lemma 10 in [5]', and Lemma 17 invokes 'Lemma 10 in [5] once more' and 'the computations in the proof of Lemma 12 in [5]' for the Gamma argument change. Reference [5] is the authors' own manuscript 'under review', not machine-checked or independently reproduced, so the central theorem's exceptional-zero count is justified by a self-citation chain rather than by a proof contained in the paper.

full rationale

Most of the paper is not circular. The interlacing proof (Theorem 7) uses the three-term recurrence and Lemma 6 directly. The zeta connection (Theorem 11) is an identity: substituting y=n√x into the definition ψ*_j(x)=∑Φ*_j(n√x) and using (2.21) literally produces q_j(s/2)π^{-s/2}Γ(s/2)ζ(s); this is a factorization, not a derivation of one side from the other in a circular sense. The contour deformations and saddle-point estimates in Lemmas 12, 13, 15 are self-contained. However, the completion of the proof of Theorem 1—the argument-principle count that yields the finite exceptional set 2⌈c+p⌉—is not proved in this paper. The final sentence delegates it to the same authors' unpublished companion [5], and Lemmas 14 and 17, which feed the change-of-argument estimate (3.17), borrow Lemmas 10 and 12 from [5]. Because [5] is under review with overlapping authorship and is not independently certified, the central claim's decisive step is load-bearing self-citation. This is a partial circularity/completeness dependence rather than an identity-of-inputs circularity. The numerical Lagrange-multiplier extrema in Lemma 18 are an additional rigor gap, but that is a correctness risk, not circularity. Score 6 reflects that the main theorem's exceptional-zero count rests on the self-citation chain while substantial independent content remains.

Assumptions & free parameters 0 free parameters · 4 assumptions · 1 invented entities

The central theorem depends on the unpublished companion paper [5] for the final argument and two lemmas, and on numerical extrema calculations in the Appendix for a monotonicity claim. The zeta identity is derived from standard Mellin and Meixner identities.

assumptions (4)
  • domain assumption The lemmas and final argument of the unpublished companion paper [5] apply verbatim to the generalized family of Theorem 1.
    Used in Lemmas 14 and 17 and the final zero count; [5] is not publicly available.
  • ad hoc to paper The numerical Lagrange multiplier computations in Lemma 18 (minimum 0.924256..., maximum 0.707107...) are accurate enough to prove the required inequalities for all t in (0,1).
    Underpins Im z'(y)>0, essential for tail estimates; no analytic proof or code supplied.
  • domain assumption Bump et al. theorem [4]: zeros of P_n^{(α)}(s) generated by (1-z)^{s-1-α/2}(1+z)^{-s-α/2} lie on Re(s)=1/2.
    The base case being extended; cited from [4].
  • standard math Mellin transform identities (2.13) and (2.15) as stated in [2] and [11].
    Used to derive Theorems 9 and 11.
invented entities (1)
  • Polynomial families {h_n(s)} and {q_n(s)}
    purpose: Central objects of the theorem; q_n connects to zeta via Mellin transform.
    Defined by generating functions (2.4) and (2.7); existence and properties are the paper's own construction with no external falsifiable handle.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sheffer sequences with zeros on a line." pith.science (2026). https://pith.science/paper/GPXGQCEM

@misc{pith2026250818229,
  author       = {Pith},
  title        = {Pith review of: Sheffer sequences with zeros on a line},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GPXGQCEM}},
  note         = {Machine review of arXiv:2508.18229}
}
read the original abstract

We extend a result of Bump et al. to show that a large family of Sheffer sequences has their zeros - up to perhaps a finite number of exceptions - on a vertical line. We connect a particular such sequence to the Riemann zeta function via a product representation of a scaled Mellin transform, analogously to the product decomposition of a Mellin transform involving the generalized Laguerre polynomials into factors of the Gamma function and Meixner polynomials.

Figures

Figures reproduced from arXiv: 2508.18229 by the authors.

Figure 2
Figure 2. shows the location of the zeros of these polynomials. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 2.1
Figure 2.1. The zeros of q9(s) and q10(s), with p = −1 and p ∗ = 0 2.2. A connection to the Riemann zeta function ζ(s). This section establishes a connection between the Riemann zeta function, our sequence of polynomials {qn(s)} and a scaled Mellin transforms of a family of fucntions reltated to the generalized Laguerre polynomials. We first recall (see [2]) that the Mellin transform of a family of functions related to the gene… view at source ↗
Figure 3.1
Figure 3.1. Γ1 and Γ2 curves On the other hand, the substitution z → z yields ˆ Γ2 ψ(z)e nϕ(z,−t) dz = (−1)n+1 ˆ Γ2 ψ(z)e nϕ(z,t)dz. We deduce from (3.3) that πhn(c − int) is either the imaginary part, or −i times the real part of (3.4) pn(t) = ˆ Γ2 ψ(z)e nϕ(z,t) dz, depending of whether n is even or odd. We now commence the analysis of the asymptotic behavior of pn(t) as n → ∞ using the saddle point method. Observe that the tw… view at source ↗
Figures from the paper (5 more)
Figure 3.2
Figure 3.2. Figure 3.2: The new contour of integration: the z(y) curve along with a vertcial line replacing Γ2 Proof. By the definition of ζ, expanding ϕ(z, t) about z = ζ yields ϕ (z, t) − ϕ (ζ, t) = ϕz 2 (ζ, t) 2 (z − ζ) 2 (1 + b(z)), [PITH_FULL_IMAGE:figures/full_fig_p012_3_2.png]
Figure 3.3
Figure 3.3. Figure 3.3: The Hankel contours before and after the change of variable z → e z Suppose ϵ > 0 is such that nϵ = o(1). We break up the range of integration of the last integral as (3.11) ˆ ln5n/n±iϵ (0−) ψ(e z )e nϕ(e z ,t) e z dz + ˆ +∞+iϵ ln5 n/n+iϵ ψ(e z )e nϕ(e z ,t) e z dz +…
Figure 3.4
Figure 3.4. Figure 3.4: The limiting zero distribution density function Appendix The work in this Appendix is dedicated to showing that for any y ∈ (−∞, L), Im z ′ (y) > 0. This result is used in establishing asymptotic estimates of the tails of the integral representations of the pn(t). As…
Figure 3.5
Figure 3.5. Figure 3.5: The three essential curves in our proof: the ζ curve, the z(y) curve and the γ(y) curve γ(L˜) ∈ (0, 1), then on the one hand Im ϕz(γ(L˜), t) = 0, since ϕz is real along the curve γ(y). On the other hand, Im ϕz(γ(L˜), t) = Im  it 1 + γ(L˜) + it 1 − γ(L˜) − 1 γ(L˜)  …
Figure 3.6
Figure 3.6. Figure 3.6: The triangle with vertices 0, 1 − z, and 1 + z Using the identities sin(π/2 + x − y) = cos(x − y) = cos x cos y + sin x sin y we find that sin π 2 + Arg(1 + z) − Arg(1 − z)  = cos (Arg(1 + z)) cos (Arg(1 − z)) + sin (Arg(1 + z)) sin (Arg(1 − z)) = 1 + Re z |1 + z| …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

  1. [5]

    Cheon, T

    G.-S. Cheon, T. Forgács, K. Tran, On the zeros of certain Sheffer sequences and their cognate sequences,under review

  2. [1]

    G. E. Andrews, R. Askey, and R. Roy, Special functions, Cambridge University Press (1999)

  3. [2]

    M. K. Atakishiyeva and N. M. Atakishiyev, On the Mellin transform of hypergeometric polynomials, J. Phys. A: Math. Gen. 32 (1999), L33-L41

  4. [3]

    Bump and Eugene K.-S

    D. Bump and Eugene K.-S. Ng, On Riemann’s zeta funcion, Math. Z. 192 (1986), 195-204

  5. [4]

    Bump, K.-K

    D. Bump, K.-K. Choi, P. Kurlberg and J. Vaaler, A local Riemann hypothesis, I, Math. Z. 233 (2000), 1-19

  6. [6]

    Cheon, T

    G.-S. Cheon, T. Forgács, A. Mwfase, and K. Tran, On combinatorial properties and the zero distri- bution of certain exponential Sheffer sequences, J. Math. Anal. App , 2024, 128513, ISSN 0022-247X, doi.org/10.1016/j.jmaa.2024.128513

  7. [7]

    Cheon, T

    G.-S. Cheon, T. Forgács, H. Kim and K. Tran, On combinatorial properties and the zero distribution of certain Sheffer sequences, J. Math. Anal. Appl. 514 (1) (2022), DOI: 10.1016/j.jmaa.2022.126273

  8. [8]

    L. L. Liu and Y. Wang, A unified approach to polynomial sequences with only real zeros, Adv. in Appl. Math. 38 (2007), 542-560. Sons, INC., New York, 1958

Show all 13 references
  1. [9]

    B., The asymptotics of the Mittag-Leffler polynomilals, Journal of Classical Analysis,1 (1), pp

    Paris, R. B., The asymptotics of the Mittag-Leffler polynomilals, Journal of Classical Analysis,1 (1), pp. 1-16, 2012

  2. [10]

    Piotrowski, Linear Operators and the Distribution of Zeros of Entire Func- tions, Ph.D

    A. Piotrowski, Linear Operators and the Distribution of Zeros of Entire Func- tions, Ph.D. Dissertation, University of Hawai‘i, 2007. Full text available at http://scholarspace.manoa.hawaii.edu/bitstream/handle/10125/25932/ PhD_2007_Piotrowski_r.pdf?sequence=1

  3. [11]

    A. D. Polyanin and A. V. Manzhirov, Handbook of Mathematics for Engineers and Scientists, Champman& Hall/CRC; 1 edition, 2007

  4. [12]

    Roman, The Theory of the Umbral Calculus I, J

    S. Roman, The Theory of the Umbral Calculus I, J. Math. Anal. Appl., 87 (1), May 1982

  5. [13]

    Shapiro, R

    L. Shapiro, R. Sprugnoli, P. Barry, G.-S. Cheon, T.-X. He, D. Merlini and W. Wang, The Riordan Group and Applications, Springer Monographs in Mathematics, 2022. SHEFFER SEQUENCES WITH ZEROS ON A LINE 31 1Department of Mathematics/ Applied Algebra and Optimization Research Cent...

Pith tools

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