# CSM_RH Paper 59 ## Seeded $L^p$ Residue-Chain Amplification, $L^1$ Optimality, and the Minimal Exceptional-Set Strip-Gap Gate **Project:** CSM_RH **Paper:** 59 **Version:** v0.1 **Date:** 2026-09-08 **Campaign:** 46 — `SEEDED_ARITHMETIC_STRIP_GAP_GENERATION` **Direct frontier:** F-RH-017 **Status:** STRONGER DETERMINISTIC EXCEPTIONAL-SET AMPLIFIER CERTIFIED / SUPERCRITICAL ARITHMETIC THRESHOLD STILL OPEN **Canonical entry state:** v1.49 / Paper 58 v0.1 **RH_PROVED:** FALSE **RH_DISPROVED:** FALSE **GLOBAL_RH_CERTIFICATE:** FALSE --- # Abstract Paper 56 converted a seeded shrinking-threshold exceptional-set theorem into a lag $L^2$ estimate and then into a PESC exponent improvement. That route required $$ c>2(1-\alpha) $$ when $$ H=N^\alpha $$ and $$ |\{x:|U_H(x)|>HN^{-\nu}\}| \ll N^{1-c}. $$ The present paper proves that this is not the optimal deterministic use of the exceptional set. Let $$ A(n)=\psi(n)-n, \qquad U_H(n)=A(n+H)-A(n). $$ Assume PESC $(\kappa)$ and write $$ d=\frac{\kappa}{2}. $$ Paper 55 gives the pointwise seed envelope $$ |A(n)| \ll n^{1-d+o(1)}. $$ For every fixed $p\ge1$, a residue-chain inequality gives $$ \sum_{n\le2N}|A(n)|^p \ll_p \frac{N}{H} \sum_{r\le H}|A(r)|^p + \left( \frac{N}{H} \right)^p \sum_{n\le2N-H}|U_H(n)|^p. $$ If $$ H=N^\alpha $$ and $$ \boxed{ \#\left\{ n\in[N,2N]: |U_H(n)|>HN^{-\nu} \right\} \ll N^{1-c}, } $$ then the seed pointwise envelope on the exceptional set yields $$ \sum_{n\le2N}|A(n)|^p \ll N^{1+p(1-d_p')+o(1)} $$ for every $$ \boxed{ d_p' < \Psi_p(d;\alpha,\nu,c) := \min \left\{ \nu,\, d+\alpha-1+\frac{c}{p},\, 1-\alpha(1-d) \right\}. } $$ A dyadic $L^p$ Mellin argument then excludes all zeta zeros with $$ \Re\rho>1-d_p'. $$ Using Paper 55's fixed-exponent PESC/zero-strip equivalence, this gives PESC $(\kappa')$ for every $$ \boxed{ \kappa' < 2\Psi_p \left( \frac{\kappa}{2}; \alpha,\nu,c \right). } $$ Among all $p\ge1$, the only $p$ -dependent term is $c/p$. Therefore: $$ \boxed{ p=1 } $$ is the optimal member of the entire $L^p$ residue-chain family. The resulting $L^1$ amplification law is $$ \boxed{ \kappa' < \Phi_1(\kappa;\alpha,\nu,c) := \min \left\{ 2\nu,\, \kappa+2c+2\alpha-2,\, 2-2\alpha+\alpha\kappa \right\}. } $$ Strict amplification occurs exactly when $$ \boxed{ \nu>\frac{\kappa}{2}, \qquad c>1-\alpha, \qquad \alpha<1. } $$ Thus the exceptional-set requirement from Paper 56 is cut in half. For the near-macroscopic parametrization $$ H=N^{1-\tau}, $$ the strict gate is simply $$ \boxed{ \nu>\frac{\kappa}{2}, \qquad c>\tau. } $$ The output gain is $$ \boxed{ \eta < \min \left\{ 2\nu-\kappa,\, 2(c-\tau),\, (2-\kappa)\tau \right\}. } $$ If the threshold is not the bottleneck and $c$ is fixed, the optimal scale is $$ \boxed{ \tau_* = \frac{2c}{4-\kappa}, \qquad \alpha_* = 1-\frac{2c}{4-\kappa}, } $$ and the one-step gain is $$ \boxed{ \eta_* = \frac{ 2c(2-\kappa) }{ 4-\kappa }. } $$ This is twice the exceptional-set gain produced by the $L^2$ bridge of Paper 56 for the same $c$. The paper also audits the current Gafni–Tao exceptional-set architecture against a polynomially shrinking threshold. Their published theorem fixes the relative threshold $\delta>0$ and a sufficiently large integer $J=J(\delta,\theta,\varepsilon)$ before taking $X\to\infty$. In their explicit-formula reduction the truncation error is $O(X^\theta/J)$, so replacing $$ \delta $$ by $$ X^{-\nu} $$ requires at least $$ J\gg X^\nu. $$ Thus the published theorem cannot be invoked by direct substitution. More importantly, even after rebuilding the proof with a polynomial truncation height, the standard $L^2$ and $L^4$ zero-packet Markov exponents are supercritical at the seeded boundary whenever $$ \nu>\frac{\kappa}{2}. $$ Hence the fixed-threshold parameter issue is not the sole obstruction. The boundary mode remains the genuine wall. The direct Campaign-46 frontier is therefore strengthened: ```text F-RH-017-v2 SEEDED_SUPERCRITICAL_SHRINKING_THRESHOLD_EXCEPTIONAL_SET ``` with the minimal deterministic requirements $$ \boxed{ H=N^{1-\tau}, \qquad \nu>\frac{\kappa}{2}, \qquad c>\tau. } $$ No RH theorem is claimed. --- # 1. Seed notation Assume PESC $(\kappa)$ for fixed $$ 0<\kappa<1. $$ Set $$ \boxed{ d=\frac{\kappa}{2}. } $$ Paper 55 gives $$ \boxed{ \beta_*\le1-d } $$ and the pointwise PNT error estimate $$ \boxed{ |A(x)| = |\psi(x)-x| \ll x^{1-d+o(1)}. } $$ Let $$ H=N^\alpha, \qquad 0<\alpha<1. $$ For integers $n$ define $$ \boxed{ U_H(n) = A(n+H)-A(n). } $$ --- # 2. A general residue-chain $L^p$ inequality Fix $$ 1\le p<\infty. $$ Write every integer $n\le2N$ as $$ n=r+kH, $$ with $$ 1\le r\le H $$ and $$ 0\le k\le M, \qquad M\ll\frac{N}{H}. $$ Along each residue chain, $$ A(r+kH) = A(r) + \sum_{j=0}^{k-1} U_H(r+jH). $$ Using $$ |x+y|^p \le 2^{p-1} \left( |x|^p+|y|^p \right) $$ and $$ \left| \sum_{jHN^{-\nu} \right\}. $$ ## Good set On the complement, $$ |U_H(n)|^p \le H^pN^{-p\nu}. $$ Hence $$ \boxed{ \sum_{n\notin\mathcal E}|U_H(n)|^p \ll N H^p N^{-p\nu}. } $$ ## Bad set The seed pointwise envelope gives $$ |U_H(n)| \le |A(n+H)|+|A(n)| \ll N^{1-d+o(1)}. $$ Therefore $$ \boxed{ \sum_{n\in\mathcal E}|U_H(n)|^p \ll N^{1-c+p(1-d)+o(1)}. } $$ Thus $$ \boxed{ \sum_{n\le2N-H}|U_H(n)|^p \ll NH^pN^{-p\nu} + N^{1-c+p(1-d)+o(1)}. } $$ --- # 5. Global seeded $L^p$ error exponent Multiply the lag terms by $$ (N/H)^p. $$ The good-set term becomes $$ \boxed{ N^{p+1-p\nu}. } $$ The bad-set term becomes $$ \boxed{ N^{1-c+p(2-d-\alpha)+o(1)}. } $$ The anchor term is $$ \boxed{ N^{1+\alpha p(1-d)+o(1)}. } $$ We compare these with the target form $$ \boxed{ N^{1+p(1-d')+o(1)}. } $$ The three constraints are: $$ d'<\nu, $$ $$ d' < d+\alpha-1+\frac{c}{p}, $$ and $$ d' < 1-\alpha(1-d). $$ Therefore: ## Theorem 5.1 — Seeded exceptional-set $L^p$ exponent For every $$ d_p' < \boxed{ \Psi_p(d;\alpha,\nu,c) = \min \left\{ \nu,\, d+\alpha-1+\frac{c}{p},\, 1-\alpha(1-d) \right\}, } $$ one has $$ \boxed{ \sum_{n\le2N}|A(n)|^p \ll N^{1+p(1-d_p')+o(1)}. } $$ Create: ```text B-RH-063 SEEDED_EXCEPTIONAL_SET_TO_GLOBAL_LP_PNT_ERROR_GAIN CERTIFIED ``` --- # 6. Global $L^p$ error implies a zero-free strip The discrete / continuous difference between $$ A(n) $$ and $$ \psi(x)-x $$ inside a unit interval is bounded by $1$, so the same exponent holds for $$ \int_N^{2N} |\psi(x)-x|^pdx. $$ Suppose $$ \int_X^{2X} |\psi(x)-x|^pdx \ll X^{1+p(1-d')+\varepsilon}. $$ Let $$ s=\sigma+it. $$ For $p>1$, Hölder gives $$ \begin{aligned} \int_X^{2X} |\psi(x)-x| x^{-\sigma-1}dx &\le \left( \int_X^{2X} |\psi(x)-x|^pdx \right)^{1/p} \\ &\quad\times \left( \int_X^{2X} x^{-q(\sigma+1)}dx \right)^{1/q} \\ &\ll X^{1-d'-\sigma+\varepsilon/p}, \end{aligned} $$ where $$ 1/p+1/q=1. $$ For $p=1$ the same bound follows directly. Thus the dyadic Mellin sum converges locally uniformly for $$ \boxed{ \Re s>1-d'. } $$ By Paper 53's Mellin identity, $$ -\frac{\zeta'(s)}{\zeta(s)} - \frac{s}{s-1} $$ is analytic there. Therefore: ## Theorem 6.1 — $L^p$ Mellin strip theorem If the dyadic $L^p$ exponent $d'$ holds, then $$ \boxed{ \zeta(s)\ne0 \quad \text{for} \quad \Re s>1-d'. } $$ By Paper 55, this is equivalent at exponent scale to PESC $(\kappa')$ for every $$ \boxed{ \kappa'<2d'. } $$ Combining with Theorem 5.1: ## Theorem 6.2 — Seeded $L^p$ exceptional-set PESC amplifier $$ \boxed{ \kappa' < 2 \Psi_p \left( \frac{\kappa}{2}; \alpha,\nu,c \right). } $$ Create: ```text B-RH-064 SEEDED_LP_EXCEPTIONAL_SET_TO_ZERO_STRIP_AND_PESC_AMPLIFICATION CERTIFIED ``` --- # 7. $L^1$ is optimal in the entire $L^p$ family The function $$ \Psi_p = \min \left\{ \nu,\, d+\alpha-1+\frac{c}{p},\, 1-\alpha(1-d) \right\} $$ depends on $p$ only through $$ c/p. $$ For $$ p\ge1, $$ $$ \frac{c}{p} \le c. $$ Hence: ## Theorem 7.1 — $L^1$ optimality Among all fixed $$ p\ge1, $$ the strongest deterministic zero-strip output from the seeded residue-chain / exceptional-set architecture is attained at $$ \boxed{ p=1. } $$ Create: ```text B-RH-065 L1_IS_OPTIMAL_SEEDED_EXCEPTIONAL_SET_RESIDUE_CHAIN_EXPONENT_CONVERTER CERTIFIED ``` This is the key improvement over Paper 56. --- # 8. The optimal $L^1$ amplification law Set $$ p=1. $$ Then $$ d_1' < \min \left\{ \nu,\, d+\alpha-1+c,\, 1-\alpha(1-d) \right\}. $$ Returning to $$ \kappa=2d, $$ we obtain: ## Theorem 8.1 — Seeded $L^1$ shrinking-threshold amplification law Every $$ \boxed{ \kappa' < \Phi_1(\kappa;\alpha,\nu,c) } $$ is admissible, where $$ \boxed{ \Phi_1 = \min \left\{ 2\nu,\, \kappa+2c+2\alpha-2,\, 2-2\alpha+\alpha\kappa \right\}. } $$ Create: ```text B-RH-066 SEEDED_L1_SHRINKING_THRESHOLD_PESC_AMPLIFICATION_LAW CERTIFIED ``` --- # 9. Exact strict-amplifier gate We ask when $$ \Phi_1>\kappa. $$ The three inequalities are $$ 2\nu>\kappa, $$ $$ \kappa+2c+2\alpha-2>\kappa, $$ and $$ 2-2\alpha+\alpha\kappa>\kappa. $$ For $$ \alpha<1, $$ the last is automatic: $$ 2-2\alpha+\alpha\kappa-\kappa = (1-\alpha)(2-\kappa) >0. $$ Thus: ## Corollary 9.1 — Minimal deterministic exceptional-set gate Strict amplification occurs exactly when $$ \boxed{ \nu>\frac{\kappa}{2}, \qquad c>1-\alpha, \qquad \alpha<1. } $$ Compared with Paper 56, $$ c>2(1-\alpha) $$ has been improved to $$ \boxed{ c>1-\alpha. } $$ The boundary threshold $$ \nu>\kappa/2 $$ cannot be weakened. --- # 10. Near-macroscopic formulation Write $$ \boxed{ \alpha=1-\tau, \qquad \tau>0. } $$ Then $$ \Phi_1 = \min \left\{ 2\nu,\, \kappa+2(c-\tau),\, \kappa+(2-\kappa)\tau \right\}. $$ Thus the exponent gain $$ \eta=\kappa'-\kappa $$ may be any fixed number satisfying $$ \boxed{ \eta < \min \left\{ 2\nu-\kappa,\, 2(c-\tau),\, (2-\kappa)\tau \right\}. } $$ The minimal gate becomes $$ \boxed{ \nu>\frac{\kappa}{2}, \qquad c>\tau. } $$ This is the strengthened direct frontier. --- # 11. Optimal use of an exceptional exponent Assume $$ 2\nu $$ is not the active bottleneck. For fixed $c$, maximize $$ \min \left\{ \kappa+2c+2\alpha-2,\, 2-2\alpha+\alpha\kappa \right\}. $$ Balance the two terms: $$ \kappa+2c+2\alpha-2 = 2-2\alpha+\alpha\kappa. $$ This gives $$ \boxed{ \alpha_* = 1- \frac{2c}{4-\kappa}. } $$ Equivalently, $$ \boxed{ \tau_* = \frac{2c}{4-\kappa}. } $$ The optimized output exponent is $$ \boxed{ \kappa_* ' = \kappa + \frac{ 2c(2-\kappa) }{ 4-\kappa }. } $$ Thus the optimized one-step gain is $$ \boxed{ \eta_* = \frac{ 2c(2-\kappa) }{ 4-\kappa }. } $$ This is exactly twice the modest- $c$ exceptional-set gain produced by the $L^2$ conversion in Paper 56. To keep the threshold term non-bottlenecking, it suffices that $$ \boxed{ \nu \ge \frac{\kappa}{2} + \frac{ c(2-\kappa) }{ 4-\kappa }. } $$ If the formula attempts to produce $\kappa'>1$, the assumptions are themselves incompatible with the known critical-line zeros; in applications one stops at any fixed $\kappa'<1$. --- # 12. General $L^p$ optimization For completeness, if $p\ge1$ is fixed and the threshold is not active, balance $$ d+\alpha-1+\frac{c}{p} $$ with $$ 1-\alpha(1-d). $$ This gives $$ \boxed{ \alpha_{p,*} = 1- \frac{ 2c }{ p(4-\kappa) }. } $$ The PESC exponent gain is $$ \boxed{ \eta_{p,*} = \frac{ 2c(2-\kappa) }{ p(4-\kappa) }. } $$ This decreases exactly like $1/p$. Thus the optimality of $p=1$ is quantitative, not merely qualitative. --- # 13. Comparison with Paper 56 Paper 56 used: ```text exceptional set -> lag L2 energy -> MLEPG -> seeded residue-chain L2 -> PESC ``` The present paper uses: ```text exceptional set -> lag Lp moment -> residue-chain Lp directly -> global PNT-error Lp -> Mellin analyticity -> zero strip -> PESC ``` The direct Mellin route avoids the extra $L^2$ cost of squaring the exceptional-set pointwise envelope. Therefore Paper 56 remains correct but is not optimal as an exceptional-set exponent converter. The improved canonical direct target should use the $L^1$ bridge. --- # 14. F-RH-017 v2 Replace the old preferred direct target by the stronger version: ```text F-RH-017-v2 SEEDED_SUPERCRITICAL_SHRINKING_THRESHOLD_EXCEPTIONAL_SET ``` At scale $$ \boxed{ H=N^{1-\tau}, } $$ prove $$ \boxed{ \#\left\{ n\in[N,2N]: |\psi(n+H)-\psi(n)-H| > HN^{-\nu} \right\} \ll N^{1-c} } $$ with $$ \boxed{ \nu>\frac{\kappa}{2}, \qquad c>\tau. } $$ Then a strict fixed PESC exponent improvement follows. This is the cheapest currently certified exceptional-set amplifier. --- # 15. Audit of the Gafni-Tao fixed-threshold proof Gafni and Tao define their exceptional set using a fixed relative threshold $$ \delta>0. $$ In the proof they explicitly: 1. fix $\delta$ ; 2. choose a natural number $J$ sufficiently large depending on $\delta,\theta,\varepsilon$ ; 3. hold $\delta,J,\theta,\varepsilon$ fixed as $X\to\infty$. Their explicit-formula truncation height is $$ T = J(\log^2X)X^{1-\theta}, $$ and the truncation error is $$ O\left( \frac{X^\theta}{J} \right). $$ To compare this with a polynomial threshold $$ \delta_X = X^{-\nu}, $$ one would need $$ \boxed{ J \gg X^\nu } $$ already at the truncation step. Thus the published theorem cannot be used by the formal substitution $$ \delta=X^{-\nu}. $$ Its asymptotic notation permits constants depending on fixed $\delta$ and $J$. This is a theorem-scope statement, not a criticism of the result. --- # 16. Fixed- $H$ removes one technical polynomial cost, but not the wall The Gafni-Tao proof also subdivides $$ [X,2X] $$ into $$ O_\delta(1) $$ multiplicative intervals in order to replace the varying length $$ x^\theta $$ by a nearly fixed multiplicative increment. For the CSM_RH frontier we may instead formulate the interval length as a fixed dyadic $$ H=N^\alpha. $$ This removes the need for that subdivision and therefore avoids a polynomial covering factor when $\delta$ shrinks. This is a useful technical simplification. However it does not resolve the seeded boundary threshold. The explicit formula still requires a polynomial height $$ T \gtrsim N^{1-\alpha+\nu}. $$ More importantly, a zero on $$ \beta=1-\kappa/2 $$ contributes relative size $$ N^{-\kappa/2}. $$ No finite positive moment argument can prove a density-small exceptional set below that amplitude while the boundary zero remains admissible. --- # 17. Polynomial-threshold $L^2$ packet exponent For comparison with the current exceptional-set technology, take a narrow zero packet around real part $$ \sigma $$ and let $$ q = 1-\alpha+\nu $$ be the minimal explicit-formula height exponent. The standard $L^2$ zero-packet estimate has average-square exponent $$ 2\alpha+2\sigma-2 + qA(\sigma)(1-\sigma). $$ Markov at threshold $$ N^{\alpha-\nu} $$ gives exceptional-measure exponent $$ \boxed{ \mu_{2,\sigma}^{\rm poly} = 1 + 2 \left( \nu-(1-\sigma) \right) + qA(\sigma)(1-\sigma). } $$ At the seed boundary $$ 1-\sigma = \frac{\kappa}{2}, $$ if $$ \nu>\frac{\kappa}{2}, $$ then $$ \boxed{ \mu_{2,\sigma}^{\rm poly}>1 } $$ even before any positive zero-density cost is used. Thus $L^2$ Markov cannot cross the boundary. --- # 18. Polynomial-threshold $L^4$ packet exponent Similarly the standard fourth-moment estimate gives $$ \boxed{ \mu_{4,\sigma}^{\rm poly} = 1 + 4 \left( \nu-(1-\sigma) \right) + qA^*(\sigma)(1-\sigma). } $$ At the seed boundary and supercritical threshold, $$ \boxed{ \mu_{4,\sigma}^{\rm poly}>1. } $$ The same baseline obstruction occurs at every finite even moment: $$ \boxed{ 1 + 2r \left( \nu-\frac{\kappa}{2} \right) } $$ before nonnegative density / additive-energy costs. Therefore the failure of direct polynomial-threshold uniformization is structural, not merely caused by the fixed- $\delta$ quantifiers in the published theorem. This refines Paper 56's moment barrier using the current Gafni-Tao parameter ledger. --- # 19. Current technology verdict Current zero-density and additive-zero-energy technology is highly effective for: $$ \boxed{ \nu<\frac{\kappa}{2} } $$ after a seed strip is present, because the boundary packet then lies below the threshold. It cannot, through a standard finite positive moment + Markov argument, enter $$ \boxed{ \nu>\frac{\kappa}{2}. } $$ But the new $L^1$ deterministic bridge substantially lowers the amount of exceptional-set rarity required once a supercritical arithmetic theorem is found. This is the exact division of labor: ```text analytic seed: moves the right edge to 1-kappa/2 current density/moment technology: controls subcritical thresholds new arithmetic theorem: must cross nu=kappa/2 L1 residue-chain bridge: converts even modest c>tau into a fixed strip gain ``` --- # 20. New minimal arithmetic target Choose a small fixed $$ \tau>0 $$ and put $$ H=N^{1-\tau}. $$ The minimal direct target is now: $$ \boxed{ \#\left\{ n\in[N,2N]: |U_H(n)| > HN^{-\kappa/2-\varepsilon} \right\} \ll N^{1-c} } $$ for some fixed $$ \varepsilon>0, \qquad c>\tau. $$ Then Theorem 10.1 gives the explicit exponent gain $$ \boxed{ \eta < \min \left\{ 2\varepsilon,\, 2(c-\tau),\, (2-\kappa)\tau \right\}. } $$ This is the sharpest currently certified direct Campaign-46 target. --- # 21. State transition Advance the candidate state from $$ v1.49 $$ to $$ v1.50. $$ Add: ```text B-RH-062 SEEDED_RESIDUE_CHAIN_LP_INEQUALITY CERTIFIED ``` Add: ```text B-RH-063 SEEDED_EXCEPTIONAL_SET_TO_GLOBAL_LP_PNT_ERROR_GAIN CERTIFIED ``` Add: ```text B-RH-064 SEEDED_LP_EXCEPTIONAL_SET_TO_ZERO_STRIP_AND_PESC_AMPLIFICATION CERTIFIED ``` Add: ```text B-RH-065 L1_IS_OPTIMAL_SEEDED_EXCEPTIONAL_SET_RESIDUE_CHAIN_EXPONENT_CONVERTER CERTIFIED ``` Add: ```text B-RH-066 SEEDED_L1_SHRINKING_THRESHOLD_PESC_AMPLIFICATION_LAW CERTIFIED ``` Update preferred frontier: ```text F-RH-017-v2 nu > kappa/2 c > 1-alpha ``` No RH certificate is created. --- # 22. Conclusion The exceptional-set route has become cheaper. The earlier $L^2$ conversion paid the square of the exceptional-set envelope. The $L^1$ residue-chain / Mellin route pays it only once. For $$ H=N^{1-\tau}, $$ the strict exceptional-set gate is now $$ \boxed{ \nu>\frac{\kappa}{2}, \qquad c>\tau. } $$ The threshold wall did not move. The exception-count wall did. This matters because any future arithmetic theorem only needs a modest polynomial rarity of supercritical failures, not the stronger rarity previously required. The current Gafni-Tao machinery cannot simply be parameter-substituted into this regime, and its finite-moment architecture remains critically blocked at the boundary. Thus the next breakthrough target is as small and explicit as the present chain can make it: $$ \boxed{ \text{beat the boundary amplitude on almost all } N^{1-\tau}\text{-intervals, with only }N^{1-c} \text{ exceptions and }c>\tau. } $$