REVIEW 4 major objections 3 minor
An application of Titchmarsh's theorem and the Salem equivalence
T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash
Pith's one-line read The Riemann hypothesis is tied to a Fourier-support condition on a modulated convolution built from a kernel whose Mellin transform is Gamma(s) zeta(s) zeta(s-1/2).
desk verdict A short note that recycles Salem's equivalence through a new kernel; the central theorem is conditional, circular, and contains an arithmetic slip, so it should not be accepted as is. 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 kernel k(x) in (1.4) is the sum over d_{1/2}(n)e^{-nx} minus the singular terms x^{-3/2} sqrt(pi/2) zeta(3/2) and x^{-1} zeta(1/2), chosen so its Mellin transform is Gamma(s) zeta(s) zeta(s-1/2) in the critical strip. The function e^{sigma x}k(e^x) is square-integrable, which brings Titchmarsh's Theorem 95/96 about boundary values of L2 analytic functions into play; the e^{-imz} factor shifts the Fourier support so the symmetry pairs of off-critical zeros are covered.
What would settle it
Check whether e^{sigma x} k(e^x) is square-integrable by evaluating the integral of its square numerically for sigma=0.75 using the series in (1.4); if it diverges with the printed constant, the Plancherel step collapses. Alternatively, compute the Mellin transform of the explicit kernel and see if it equals Gamma(s) zeta(s) zeta(s-1/2) in the strip.
Extended reading notes
Core claim
The paper constructs a kernel k(x) from the Dirichlet series sum_{n>=1} d_{1/2}(n)e^{-nx} minus two singular terms, so that its Mellin transform is Gamma(s) zeta(s) zeta(s-1/2). For f in L1, the convolution I_sigma(z) = (f * e^{sigma x} k(e^x))(z) lies in L2, and e^{-imz} I_sigma(z) satisfies Titchmarsh's boundary-value conditions. Theorem 1.2 states that the Fourier-support condition for this modulated convolution (null for t<-m) is equivalent, under RH, to Ff(t)=0 for t<0, and that RH is true iff no bounded measurable f solves the convolution equation identically except f=0.
Load-bearing premise
The load-bearing premise is that the kernel in (1.4) exactly equals the contour integral after residue subtraction, with a correct constant in the x^{-3/2} term, so that e^{sigma x} k(e^x) is square-integrable and Titchmarsh's Hilbert transform machinery applies; the printed sqrt(pi/2) may be a typo for (sqrt(pi)/2) zeta(3/2).
Editorial extensions
If this is right
- If the paper's equivalence is correct, RH is equivalent to the statement that the convolution equation (f * e^{sigma x} k(e^x))(z)=0 has no bounded measurable solution other than zero.
- The Fourier-support condition in Theorem 1.2 provides a concrete test: for RH, any f whose modulated convolution is analytic in the upper half-plane with Fourier support in t<-m must have Ff(t)=0 for t<0.
- The asymptotic relation (3.1) shows the support edge m can be read off from the exponential growth rate of the L2 norm of the analytic extension, offering a quantitative handle.
- Using Wiener's theorem, the paper concludes that under RH the translates of e^{sigma x}k(e^x) are dense in L1.
- The Paley–Wiener integral identity (3.2) together with the logarithmic integrability condition gives a sufficient criterion for RH in terms of decay of the modulated convolution.
Reading between the lines
- The apparent typo in the residue constant, if corrected, likely yields a family of kernels indexed by r in (0,1) from the divisors d_r(n), each giving a distinct Salem-type equivalence; testing with a simple f (e.g., a Gaussian) could numerically probe the sharpness of the support condition.
- The paper leaves open whether the condition F bar-I_{sigma,m}(t)=0 for t<-m alone, without assuming RH, forces Ff(t)=0 for t<0; proving that unconditional implication would give an independent RH criterion not explicitly demonstrated.
- The Hilbert-transform formulation suggests RH might be rephrased as a statement about conjugate functions: the real and imaginary parts of the modulated convolution are Hilbert transforms of each other exactly when the zero symmetry of the critical line holds, potentially connecting to other Hilbert-space reformulations of RH.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims to extend Salem's equivalence for the Riemann hypothesis by applying Titchmarsh's theorem on Fourier transforms and Hilbert transforms. It introduces a kernel k(x) in (1.4), defines a convolution integral I_sigma(f), and states Theorem 1.2 as a set of alternative necessary and sufficient conditions for a complex-valued function to be a boundary value of an analytic function in the upper half-plane with L2 growth. Theorem 1.2 lists three conditions: (i) Hilbert-transform pairing, (ii) vanishing of the Fourier transform of the modified convolution for t < -m, and (iii) vanishing of Ff(t) for t < 0 'if the Riemann hypothesis is true.' Section 3 then asserts a Salem-type converse and gives a related sufficient condition involving logarithmic integrals. The central claim is that these Fourier-analytic conditions are equivalent to RH.
Significance. If the claimed Salem-type equivalence were rigorously established, it would provide a new Fourier-analytic reformulation of the Riemann hypothesis, connecting RH to support properties of Fourier transforms and Hilbert transforms. This would be a noteworthy contribution. The paper also aims to use Titchmarsh's classical theorem in a new way. However, as written the central equivalence is not proved: the converse direction is merely asserted, and parts of the proof are circular or rely on an incorrect residue computation. The paper contains no machine-checked proofs, reproducible code, or fully parameter-free derivations that would independently support the claims. Its value is therefore contingent on substantial additional work rather than being established in this manuscript.
major comments (4)
- [Section 3] The claimed Salem-type equivalence is asserted, not proved. The sentence 'the Riemann hypothesis would be true should no bounded measurable function f satisfy ... = 0, other than the trivial case f ≡ 0' is stated without a proof of the converse direction. Theorem 1.2's condition (iii) is explicitly conditional on RH, and the proof in §2 only shows that, assuming RH, the multiplier is nonzero, so that (ii) and (iii) coincide. No construction from a hypothetical off-critical zero of ζ is given, and no Tauberian, distributional, or approximation argument establishes that such a zero would force a nonzero bounded f annihilating the convolution. This missing converse is the core of the paper's claimed equivalence and is not a minor omission.
- [§2, proof of Theorem 1.2, condition (ii)] The derivation of condition (ii) from condition (i) is circular. The text reads: 'To see how condition (ii) follows from (i), observe that by (ii) if x < -m, F Ī_σ,m(x) = ... = 0'. This assumes the very condition that is being proved. The subsequent Parseval/L2-growth argument is a separate attempt, but it is not developed rigorously and contains variable confusion between x and t. A correct proof must derive the vanishing of the Fourier transform for t < -m from the analyticity and growth condition alone, without invoking (ii).
- [§1.4 and §2.2] The residue subtraction in the definition of k(x) is numerically incorrect. The residue of Γ(s)ζ(s)ζ(s−1/2) at s = 3/2 is (√π/2) ζ(3/2), not √(π/2) ζ(3/2) as printed in (1.4). If this is not a typo, then the singular term is not cancelled and e^{σx}k(e^x) is not in L²(−∞,∞), invalidating the Plancherel argument and the applicability of Titchmarsh's theorem. If it is a typo, it must be corrected, but as written it is a load-bearing numerical error in the kernel construction.
- [Theorem 1.2, condition (iii)] Condition (iii) is not an 'alternative necessary and sufficient condition' in the sense of the theorem. It reads 'Ff(t) = 0 for t < 0, if the Riemann hypothesis is true.' This makes the theorem conditional rather than an equivalence, and it does not establish a Salem-type reformulation. Moreover, no definition is given that relates the vanishing of Ff(t) for t<0 to the absence of bounded solutions of the convolution equation. The theorem as stated is therefore not a coherent equivalence statement.
minor comments (3)
- [Throughout] There are frequent typos and notational inconsistencies, e.g., 'Riemman' in the Introduction, and the repeated use of 'sinc(x) := e^{-imx} sin(x)/x = e^{-imx} sinc(x)' which is a tautology. These should be corrected in revision.
- [§2, after Eq. (2.2)] In the Plancherel computation, the variable of integration is written inconsistently (sometimes x, sometimes t). For example, the equality involving |Γ(σ−ix)ζ(σ−ix)ζ(σ−ix−1/2)|² should use a consistent Fourier variable. This makes the argument harder to follow and should be fixed.
- [§3, Eq. (3.4)] The integral formula is stated imprecisely. The standard evaluation is ∫_0^∞ x^{ε−1}/(1+x²) dx = π/(2 sin(πε/2)) for 0<ε<2, hence the full integral over R is π/sin(πε/2). The text 'is twice π/ sin(επ/2)' appears to contain a factor error or a typo.
Circularity Check
No significant circularity; the core claims are conditional restatements via Mellin transforms, with an omitted converse proof rather than a circular reduction.
full rationale
The paper's derivation chain is not circular. The kernel k is constructed so that its Mellin transform is Γ(s)ζ(s)ζ(s−1/2), a legitimate use of the standard Mellin representation (2.1)–(2.2); the Fourier-support condition on F\bar I_{σ,m} is then obtained by exact transform identities, not by assuming the conclusion. Condition (iii) of Theorem 1.2 explicitly contains 'if the Riemann hypothesis is true,' so the proof of (ii)⇔(iii) is a conditional algebraic consequence of the nonvanishing of the multiplier, not a derivation of RH from (ii). The paper itself concedes in Section 3 that 'the one-one property was not applied in Theorem 1.2 to relate f directly to the Riemann hypothesis,' and the subsequent 'However, the Riemann hypothesis would be true should no bounded measurable function f satisfy...' is asserted without proof. That is an omitted proof / correctness gap, not a circular step: no equation is shown to equal its own input by construction, no fitted parameter is relabeled as a prediction, and there is no load-bearing self-citation (all references are to Titchmarsh, Salem, Wiener, and standard texts). The possible residue-coefficient typo in (1.4)/(2.2) would affect L2 membership, but is a technical error independent of circularity. Therefore the circularity score is 0.
Assumptions & free parameters
free parameters (2)
- m =
arbitrary positive real
- sigma =
1/2 < sigma < 1
assumptions (4)
- domain assumption Mellin transform identity (2.1) for the divisor sum kernel
- standard math Titchmarsh Theorems 95 and 96
- domain assumption Plancherel, Stirling, and polynomial growth of zeta imply e^{sigma x}k(e^x) is in L2
- standard math Wiener's Tauberian theorem [8, Theorem II]
invented entities (1)
-
kernel k(x) = sum_n d_{1/2}(n)e^{-nx} - x^{-3/2} sqrt(pi/2) zeta(3/2) - x^{-1} zeta(1/2)
Cite this review
Pith. "Pith review of An application of Titchmarsh's theorem and the Salem equivalence." pith.science (2026). https://pith.science/paper/VNZMRRF5
@misc{pith2026250821789,
author = {Pith},
title = {Pith review of: An application of Titchmarsh's theorem and the Salem equivalence},
year = {2026},
howpublished = {\url{https://pith.science/paper/VNZMRRF5}},
note = {Machine review of arXiv:2508.21789}
}
read the original abstract
We extend the equivalence of the Salem type for the Riemann hypothesis by application of Titchmarsh's theorem. Other equivalences to the Riemann hypothesis and notes on related Fourier integrals are provided.
Reviewed August 5, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.