# Accept

The manuscript establishes a sound, clearly stated characterization of the rightmost zeta-zero abscissa through positive excursions of a specific smoothed prime sum. Its strongest contribution is the combination of the limit in Theorem 2, valid on every sufficiently late power interval, and the square-root endpoint dichotomy. These are relevant analytic results for the selected problem. They do not prove RH: the arithmetic estimate (6) remains an assumption, and the manuscript appropriately separates it from the proved analytic implications.

The central proof withstands examination. In (3), adding the real-center derivative to the absolute value at a complex center genuinely permits control by the positive part alone. The gamma-density localization gives exponentially small tails, while choosing the hypothesis's exponent $\eta$ after fixing $b$ yields the required unbounded sequence of derivative orders. Section 3 then handles both a positive real nearest pole and a uniquely nearest nonreal pole. In the latter case the derivative at the real center has a strictly smaller exponential scale, so it cannot cancel the obstruction. The compactness and distance arguments do not require the supremum of the zero real parts to be attained. The positive character weights likewise prevent cancellation of constituent zero residues.

Theorem 2 follows correctly from this criterion and the absolutely convergent smoothed explicit formula. In particular, convergence of $\sum_\rho |\rho(1-\rho)|^{-1}$ justifies the upper bound even when $B_\zeta$ is unattained. At the RH endpoint, uniform convergence and relatively dense simultaneous phase returns justify taking maxima on the expanding logarithmic intervals. This step needs no linear-independence assumption about zero ordinates. The evaluation $2+\gamma_E-\log(4\pi)$ agrees with the completed-zeta logarithmic derivative identity. If RH fails, the exponent theorem supplies a positive excursion on every late interval, which proves divergence of the normalized maximum rather than merely an unbounded subsequence.

The auxiliary estimates also have adequate proofs under their stated hypotheses. Proposition 3 leaves a positive horizontal margin from the allowed zeros, and its conductor and cutoff losses fit within $Y^\tau$. Proposition 4 explicitly excludes the possible exceptional real zero; the lower bound on $\log y$ makes the contour decay dominate both the conductor losses and the prescribed exponential target uniformly. Neither estimate establishes the missing arithmetic premise.

Novelty should be understood narrowly. Inspection of the pinned release [2] confirms that the localized integral and pointwise criteria, the nearest-pole proof framework, and counterparts of both auxiliary estimates are already present there. They are properly credited and should not be advertised as new theorems or a new detection method. The added value is the exact window-exponent formulation, its endpoint maximum law, and the accessible synthesis connecting these conclusions to the credited criterion. Suzuki [6, Theorems 1.1–1.2 and Section 11] supplies closely related zero expansions and positivity criteria, but for a different prime transform; those statements do not themselves give the two window limits proved here. The published rStation Paper 6, version 3 [9], also recovers the rightmost zero abscissa as an exact exponent, but through shifted Nyman–Beurling approximation distances, not these one-sided prime excursions. The focused comparison did not identify a prior statement duplicating the manuscript's combined localized characterization. The classical zero-sum constant and the positive-definite-kernel observation are supporting facts, not independent novelty claims.

The exposition meets the clarity requirement: the research question, hypotheses, proof mechanism, consequences, and unresolved arithmetic input can all be followed from the main text. The proofs themselves provide the relevant reproducible evidence; numerical experiments are unnecessary for these claims. This assessment is based on mathematical examination and source comparison, not a compilation or independent certification of the external formalization.

Two small improvements would sharpen the presentation. First, state explicitly in the introduction that Theorem 2 and the endpoint corollary are the paper's added conclusions, while Sections 2–4 and the auxiliary estimates explain results credited to [2]. Second, explain that the recurrence argument supplies no effective onset or convergence rate for the endpoint maximum. This would prevent readers from interpreting the asymptotic characterization as a finite computational test of RH. Neither clarification requires changing the scientific argument.

Reference numbers [2] and [6] refer to the manuscript's bibliography. Additional comparison:

[9] span_of_dilates. *The exact rate exponent in the shifted Nyman–Beurling criterion: $d_N(a)=N^{-(a-\Theta)^+ + o(1)}$*. rStation, riemann_hypothesis_6, version 3, Section 1 and Corollary K.
