# CSM_RH Paper 02 ## Hilbert–Laplace Packet Globalizer, Exact Zero-Packet Exponential Type, and Campaign 01 Closure **Project:** `CSM_RH` **Paper:** `02` **Version:** `v0.1` **Date:** `2026-09-04` **Parent:** `CSM_RH Paper 01` **Status:** structural theorem / frontier audit; not a proof or disproof of RH **中文標題：** CSM_RH 論文 02：Hilbert–Laplace 封包全域化算子、精確零點封包指數型與 Campaign 01 閉合 **作者：** Neo.K **機構：** EveMissLab／一言諾科技有限公司 **Language:** English --- # 0. Trust boundary This paper does not prove or disprove the Riemann Hypothesis. Canonical state: ```text RH_PROVED = FALSE RH_DISPROVED = FALSE GLOBAL_RH_CERTIFICATE = FALSE ROOT_STATUS = OPEN ``` This paper proves a narrower structural result: > For the smooth $q=1$ major-arc zeta-zero packet used in the v3.18 architecture, zero-zero interference cannot lower the packet's exponential growth type below the supremal real part of the zeta zeros. Consequently, the pointwise packet-isolation lower bound proposed in Paper 01 is not needed for the fixed-exponent strength audit of EMAE. The arithmetic EMAE upper bound itself remains open. No live GLM-5.3-Flash provider run is claimed in this package. --- # 1. Canonical correction: GLM is a worker/provider, not the protocol Paper 01 introduced the names: ```text GLM_Backward Search Goal-Led Meta-Research ``` as if `GLM` were the name of the mathematical research protocol. That naming was not canonical and is withdrawn. The canonical architecture is: ```text CSM_RH = closure-space mathematics protocol / state space GLM-5.3-Flash = replaceable high-volume worker/provider implementation roles = Designer / Builder / Verifier frontier model = residual-risk judge / final promotion authority ``` Therefore: $$ \boxed{ \mathrm{GLM} \neq \mathrm{Protocol}. } $$ `Campaign 01` is henceforth a `CSM_RH worker campaign`. GLM-5.3-Flash is a preferred current worker model for that campaign, not the semantic identity of the campaign itself. --- # 2. Campaign 01 target Paper 01 created: ```text S-RH-007 LOSSLESS_WITNESS_GLOBALIZER ``` with the goal of converting an off-critical zero witness into a global invariant which cannot be erased by aggregation. The principal open instances were: ```text O-RH-005-W WEIL_AGGREGATION_ISOLATION O-RH-005-M MAJOR_ARC_AGGREGATION_ISOLATION ``` Campaign 01 generated and audited several candidate families. The principal survivor is: ```text C-RH-GLB-01 HILBERT_LAPLACE_PACKET_GLOBALIZER abbrev: HLPG ``` It applies directly to the smooth $q=1$ major-arc zero packet. --- # 3. Candidate batch The campaign-level candidates are: ```text C-RH-GLB-01 HILBERT_LAPLACE_PACKET_GLOBALIZER status: SURVIVES / PROVED_AT_PACKET_TYPE_LEVEL C-RH-GLB-02 FEJER_BOHR_LOGTIME_AVERAGING status: DEPRIORITIZED reason: amplitude-drift normalization debt; HLPG is stronger and cleaner C-RH-GLB-03 POSITIVE_RESOLVENT_REWEIGHTING status: REJECTED_AS_REPRESENTATION_ONLY reason: earlier positive-kernel audits already show no fixed-exponent discount C-RH-GLB-04 FINITE_INTERVAL_SUZUKI_SPECTRAL_LIMIT status: DEFERRED reason: relevant modern operator framework exists, but its infinite-interval spectral limit is conjectural C-RH-GLB-05 DENSITY_WEIGHTED_EXTREME_ZERO_SUPPRESSION status: REJECTED_BY_O-RH-003 reason: density alone does not exclude a single extreme zero C-RH-GLB-06 RANDOM_PHASE_PACKET_ORTHOGONALIZATION status: DEFERRED reason: creates observable-transfer and arithmetic-realizability debt ``` Only `C-RH-GLB-01` is promoted to a theorem route in this paper. --- # 4. Smooth major-arc kernel Let $$ w\in C_c^\infty((0,\infty)) $$ be nonzero, and fix $$ U>0. $$ For $$ s\in\mathbb C $$ and $$ u\in[-U,U], $$ define $$ \mathcal W_s(u) = \int_0^\infty w(v)e(uv)v^{s-1}\,dv. $$ For $$ x=e^T $$ and $$ \epsilon=\frac{u}{x}, $$ the exact scaling is $$ W_{x,u/x}(s) = e^{sT}\mathcal W_s(u). $$ --- # 5. Hilbert packet space Define the Hilbert space $$ \mathcal H_{w,U} = L^2 \left( [-U,U], |\mathcal W_1(u)|^2du \right). $$ For each nontrivial zeta zero $\rho$, define the coefficient vector $$ v_\rho(u) = \mathcal W_\rho(u). $$ Its squared Hilbert norm is $$ \|v_\rho\|_{\mathcal H_{w,U}}^2 = \int_{-U}^{U} |\mathcal W_1(u)|^2 |\mathcal W_\rho(u)|^2du. $$ This is exactly the zero cross-kernel coefficient $$ C_{w,U}(\rho). $$ The v3.18 kernel theorem gives $$ \boxed{ \|v_\rho\|_{\mathcal H_{w,U}}^2 = C_{w,U}(\rho) > 0. } $$ Thus every nontrivial zero has a nonzero Hilbert coefficient. --- # 6. Rapid vertical decay Write $$ s=\sigma+i\gamma, \qquad 0\le\sigma\le1. $$ Set $$ v=e^y. $$ Then $$ \mathcal W_{\sigma+i\gamma}(u) = \int_{\mathbb R} w(e^y)e^{\sigma y}e(ue^y)e^{i\gamma y}\,dy. $$ Because $w$ has compact support inside $(0,\infty)$, the function $$ y \mapsto w(e^y)e^{\sigma y}e(ue^y) $$ is smooth and compactly supported in a fixed interval, uniformly for $$ 0\le\sigma\le1, \qquad |u|\le U. $$ Repeated integration by parts therefore gives: ## Lemma 6.1 — Uniform rapid vertical decay For every integer $$ N\ge0, $$ there exists $$ C_N=C_N(w,U) $$ such that $$ \boxed{ \sup_{\substack{ 0\le\sigma\le1\\ |u|\le U }} | \mathcal W_{\sigma+i\gamma}(u) | \le C_N (1+|\gamma|)^{-N}. } $$ The same estimate holds for $$ \|v_{\sigma+i\gamma}\|_{\mathcal H_{w,U}}. $$ --- # 7. Absolute summability over zeta zeros Let $$ N_\zeta(Y) $$ count nontrivial zeta zeros up to height $Y$, with multiplicity. The classical zero-counting estimate gives $$ N_\zeta(Y) = O(Y\log Y). $$ Combining this with Lemma 6.1, choose $N$ sufficiently large to obtain $$ \boxed{ \sum_{\rho} \|v_\rho\|_{\mathcal H_{w,U}} < \infty. } $$ Therefore the zero packet below is absolutely convergent in $\mathcal H_{w,U}$ on every compact $T$ -interval. --- # 8. Hilbert-valued zero packet Define $$ F(T) = \sum_{\rho} e^{\rho T}v_\rho, \qquad T\ge0. $$ Define the rightmost-zero abscissa $$ \Theta_\zeta = \sup_\rho \Re\rho. $$ By symmetry, $$ \Theta_\zeta = \frac12+\Delta_\zeta. $$ The easy upper estimate is $$ \|F(T)\| \le \sum_\rho e^{(\Re\rho)T} \|v_\rho\| \le e^{\Theta_\zeta T} \sum_\rho \|v_\rho\|. $$ Hence $$ \boxed{ \|F(T)\| = O(e^{\Theta_\zeta T}). } $$ The issue is whether zero-zero interference can improve this exponent. The next theorem says no. --- # 9. Hilbert–Laplace coefficient recovery Define the packet type $$ \tau_F = \inf \left\{ a\in\mathbb R: \|F(T)\| = O(e^{aT}) \right\}. $$ ## Theorem 9.1 — Exact Hilbert packet type $$ \boxed{ \tau_F = \Theta_\zeta. } $$ ### Proof The upper bound $$ \tau_F\le\Theta_\zeta $$ was established in Section 8. Assume for contradiction that $$ \tau_F<\Theta_\zeta. $$ Choose $$ a $$ such that $$ \tau_Fa. $$ For $$ \Re s>\Theta_\zeta, $$ absolute convergence permits termwise integration: $$ \widehat F(s) = \sum_\rho \frac{v_\rho}{s-\rho}. $$ By the rapid decay from Lemma 6.1, the series $$ \sum_\rho \frac{v_\rho}{s-\rho} $$ converges locally normally on compact subsets avoiding the zeta zeros. Hence it defines an $\mathcal H_{w,U}$ -valued meromorphic function whose residue at a zero $\rho_0$ of multiplicity $m(\rho_0)$ is $$ m(\rho_0)v_{\rho_0}. $$ Because $$ a<\Theta_\zeta, $$ there exists a zero $\rho_0$ with $$ \Re\rho_0>a. $$ The coefficient non-annihilation theorem gives $$ v_{\rho_0}\neq0. $$ Choose a continuous linear functional $$ \ell \in \mathcal H_{w,U}^\ast $$ such that $$ \ell(v_{\rho_0})\neq0. $$ Then $$ \ell(\widehat F(s)) $$ is holomorphic for $$ \Re s>a, $$ but on $$ \Re s>\Theta_\zeta $$ it equals $$ \sum_\rho \frac{ \ell(v_\rho) }{ s-\rho }. $$ The latter has a nonremovable pole at $\rho_0$ with nonzero residue $$ m(\rho_0)\ell(v_{\rho_0}). $$ By uniqueness of analytic continuation this is impossible, because $\rho_0$ lies inside the holomorphy half-plane $$ \Re s>a. $$ Therefore $$ \tau_F\ge\Theta_\zeta. $$ Combining both inequalities gives $$ \boxed{ \tau_F=\Theta_\zeta. } $$ $\square$ --- # 10. Interpretation Theorem 9.1 is the major-arc analogue of the successful fixed-aperture singularity-separation mechanism. The key fact is: > Cross terms may cancel at individual scales, but they cannot erase a nonzero exponential coefficient from the analytic Laplace resolvent. Thus: ```text POINTWISE PACKET INTERFERENCE does not imply EXPONENTIAL-TYPE INTERFERENCE. ``` No literal rightmost zero is required. If the supremum is not attained, the proof chooses a zero with $$ \Re\rho>a $$ for every $$ a<\Theta_\zeta. $$ --- # 11. Positive packet energy The $q=1$ pole-zero packet energy on the core arc is $$ \mathcal E_1(e^T;U) = e^T \|F(T)\|_{\mathcal H_{w,U}}^2. $$ Define $$ \tau_E = \inf \left\{ b: \mathcal E_1(e^T;U) = O(e^{bT}) \right\}. $$ ## Theorem 11.1 — Exact $q=1$ packet-energy type $$ \boxed{ \tau_E = 1+2\Theta_\zeta. } $$ ### Proof Theorem 9.1 gives the upper estimate immediately. Conversely, if $$ \mathcal E_1(e^T;U) = O(e^{bT}), $$ then $$ \|F(T)\| = O \left( e^{(b-1)T/2} \right). $$ Theorem 9.1 therefore forces $$ \Theta_\zeta \le \frac{b-1}{2}. $$ Hence $$ b \ge 1+2\Theta_\zeta. $$ $\square$ Using $$ \Theta_\zeta = \frac12+\Delta_\zeta, $$ we also have $$ \boxed{ \tau_E = 2+2\Delta_\zeta. } $$ --- # 12. Closure of the growth-type MZI debt Paper 01 introduced ```text F-RH-004 MAJOR_ZERO_PACKET_ISOLATION ``` because the isolated pole-zero channel $$ x^{1+2\Re\rho} $$ was not known to give a pointwise lower bound for the full packet. Paper 02 splits that frontier. ## F-RH-004A ```text POINTWISE_MAJOR_ZERO_PACKET_LOWER_ENVELOPE status: OPEN priority: DEPRIORITIZED ``` This is the old pointwise MZI form. ## F-RH-004B ```text MAJOR_ZERO_PACKET_EXPONENTIAL_TYPE status: CLOSED certificate: Theorem 11.1 ``` The fixed-exponent strength audit needs `F-RH-004B`, not `F-RH-004A`. Therefore the major-arc aggregation-isolation barrier is closed at exponential-type level. Canonical transition: ```text O-RH-005-M MAJOR_ARC_AGGREGATION_ISOLATION OPEN -> CLOSED_AT_EXPONENTIAL_TYPE ``` --- # 13. EMAE strength is now certified Recall the positive character major-arc energy gate $$ \operatorname{EMAE}(\eta): \qquad \mathfrak Z_2(x;Q,U) + \mathfrak Z_4(x;Q,U) \ll x^{3-\eta+o(1)}. $$ The $q=1$ contribution is a nonnegative term of $$ \mathfrak Z_2. $$ Therefore EMAE implies $$ \mathcal E_1(x;U) \ll x^{3-\eta+o(1)}. $$ Theorem 11.1 gives $$ 1+2\Theta_\zeta \le 3-\eta. $$ Hence: ## Corollary 13.1 — Fixed-strip consequence of EMAE If $$ \operatorname{EMAE}(\eta) $$ holds for a fixed $$ \eta>0, $$ then $$ \boxed{ \Theta_\zeta \le 1-\frac{\eta}{2}. } $$ Equivalently, $$ \boxed{ \Delta_\zeta \le \frac12-\frac{\eta}{2}. } $$ Thus every fixed positive $\eta$ yields a fixed zero-strip breakthrough. If $$ \eta=1, $$ then $$ \Theta_\zeta\le\frac12. $$ By functional-equation symmetry, $$ \Theta_\zeta\ge\frac12, $$ so $$ \Theta_\zeta=\frac12, $$ which is RH. Therefore: ```text EMAE(eta>0) = at least S2 fixed-zero-strip strength EMAE(1) = S3 RH-level strength ``` This is now a certified structural implication, not a single-zero dominance heuristic. --- # 14. What has and has not been solved ## Closed ```text C1 smooth zero coefficients are rapidly summable C2 every zero has nonzero Hilbert coefficient C3 full q=1 zero packet has exact exponential type Theta_zeta C4 packet interference cannot reduce exponential type C5 q=1 pole-zero energy has exact type 1 + 2 Theta_zeta C6 EMAE fixed-power strength audit is certified ``` ## Still open ```text G1 prove EMAE(eta) for any fixed eta > 0 G2 derive a genuinely new arithmetic cancellation theorem strong enough for G1 G3 close the Weil constructive isolation frontier if that branch remains active G4 prove the RH-scale eta = 1 major-arc energy bound G5 RH ``` The campaign has removed a representation/globalization uncertainty. It has not supplied the missing arithmetic fixed-power saving. --- # 15. Why this is not another RH-equivalent representation loop Theorem 11.1 alone does not assert $$ \Theta_\zeta=\frac12. $$ It identifies the exact exponent encoded by a pre-existing positive packet energy. The new closure gain is: ```text before: full packet interference could invalidate the isolated-channel strength audit after: full packet exponential type is exactly known ``` This discharges a bridge debt. It does not discharge the arithmetic energy upper-bound debt. Therefore this result is not classified as `REPACKAGING_ONLY`. --- # 16. Candidate C-RH-GLB-02 — Fejer / Bohr log-time averaging A natural attempt is to average the packet over long log-time windows so distinct imaginary frequencies become orthogonal. For finite trigonometric sums this is standard. However the amplitudes are $$ e^{(\Re\rho)T}, $$ so a long translation window changes both phase and magnitude. Without a normalization tied to an unknown extremal real part, the averaging theorem introduces a new scale-selection debt. Because HLPG already recovers coefficients without this debt, the candidate is: ```text DEPRIORITIZED ``` not false. --- # 17. Candidate C-RH-GLB-03 — Positive resolvent reweighting Earlier positive-kernel work showed that fixed nonzero positive spectral pieces retain the same off-axis exponent, while moving spectral windows pay a sensitivity tax. Full moving-center energies are elliptically equivalent, at fixed-exponent level, to classical normalized PNT mean-square quantities. Therefore another positive reweighting without a new arithmetic theorem is classified: ```text REJECTED_AS_REPRESENTATION_ONLY ``` under the existing representation-closure obstruction. --- # 18. Candidate C-RH-GLB-04 — finite-interval Suzuki operator limit Modern Suzuki operator work provides a strong and relevant finite-interval framework for the Weil quadratic form and connects the construction with de Branges / operator theory. However the proposed limiting self-adjoint operator whose spectrum would recover zeta-zero ordinates is presently formulated as a conjectural limit construction. Therefore it cannot currently serve as a closed `CSM_RH` globalizer certificate. Classification: ```text DEFERRED external dependency: conjectural finite-interval -> infinite-limit spectral bridge ``` This remains a valuable future route, especially for the Weil branch. --- # 19. Candidate C-RH-GLB-05 — density weighted suppression Any candidate whose decisive step is only a density theorem still permits a sparse or isolated extreme zero. The packet-type theorem actually makes the obstruction sharper: a single nonzero zero coefficient already creates an analytic singularity in the Hilbert-Laplace resolvent. Therefore density-only suppression remains blocked. Classification: ```text REJECTED_BY_O-RH-003 ``` --- # 20. Candidate C-RH-GLB-06 — random phase orthogonalization Introducing an auxiliary random phase can diagonalize a finite packet in expectation. But a new phase variable must be connected back to a canonical arithmetic observable. Without that transfer theorem the route introduces: ```text OBSERVABLE_TRANSFER_DEBT ARITHMETIC_REALIZABILITY_DEBT ``` Because HLPG achieves the required noncancellation without altering the observable, this candidate is deferred. --- # 21. Updated obstruction topology The major-arc branch now becomes: ```text OFF-CRITICAL ZERO | v SMOOTH MELLIN COEFFICIENT v_rho != 0 | v HILBERT ZERO PACKET F(T) | v HILBERT-LAPLACE COEFFICIENT RECOVERY | v exact packet type = Theta_zeta | v exact q=1 energy type = 1 + 2 Theta_zeta | v EMAE fixed-power bound | v fixed zero strip ``` The globalization bridge is now closed. The only unresolved major-arc step in this chain is the arithmetic upper bound. --- # 22. Updated CSM_RH frontier ## Principal frontier ```text F-RH-003 EXCEPTIONAL_MAJOR_ARC_ENERGY status: OPEN strength: S2 for every fixed eta > 0 S3 at eta = 1 ``` ## Deprioritized auxiliary frontier ```text F-RH-004A POINTWISE_MAJOR_ZERO_PACKET_LOWER_ENVELOPE status: OPEN priority: LOW reason: no longer required for exponent-level closure ``` ## Closed bridge frontier ```text F-RH-004B MAJOR_ZERO_PACKET_EXPONENTIAL_TYPE status: CLOSED certificate: Theorem 9.1 Theorem 11.1 ``` --- # 23. GLM worker architecture for the next campaign The next worker campaign must preserve the canonical provider/protocol separation. ## CSM_RH produces ```text Frontier Task Contract Source Pack Known Obstruction Pack Strength Budget Required Output Schema ``` ## Designer worker Produces: ```text candidate mechanism minimal assumptions target estimate known theorem mapping failure modes ``` ## Builder worker Produces: ```text derivation lemmas explicit constants / exponents proof obligations computational crosschecks ``` ## Verifier worker Produces: ```text objections hidden-RH-premise audit quantifier audit uniformity audit counterexample attempt strength classification ``` ## Frontier model Receives only the typed artifacts: ```text Task Contract Candidate Artifact Objection Repair Patch Verification Report ``` It does not treat worker self-confidence as proof authority. --- # 24. Campaign 02 target Because the packet globalizer is closed, the next high-volume worker target is no longer `LOSSLESS_WITNESS_GLOBALIZER` on the major-arc branch. The new target is: ```text CSM_RH Campaign 02 target: ARITHMETIC_EMAE_FIXED_POWER question: find the weakest genuinely arithmetic theorem that yields EMAE(eta) for some fixed eta > 0 forbidden shortcuts: density-only representation-only shrinking-aperture false saving moving-band sensitivity omission isolated-channel dominance assumption finite verification -> global theorem ``` Candidate mechanism families should include only mechanisms that can change arithmetic strength. Examples of admissible search families: ```text zero repulsion with quantitative energy consequence character packet bilinear cancellation large-sieve refinement with extremal-zero sensitivity new weighted prime-correlation inequality structural cancellation tied to functional-equation pairing hybrid conductor-height-energy estimate ``` These are search families, not established theorems. --- # 25. Finite synthetic crosscheck The package includes: ```text hlpg_finite_packet_crosscheck.py hlpg_finite_packet_crosscheck.csv ``` The script verifies on a synthetic finite Hilbert-valued exponential packet: 1. direct norm-square energy equals its pair expansion; 2. the Laplace rational form equals direct numerical integration; 3. the maximal-real-part diagonal term produces a positive real-axis residue; 4. lower-real-part same-frequency terms do not cancel the maximal diagonal pole; 5. pointwise interference can be substantial while exponential-type recovery remains intact. This is a formula crosscheck only. It is not numerical evidence for RH. --- # 26. External literature boundary The proof of Theorem 9.1 is self-contained at the level needed here. Relevant surrounding literature includes: 1. M. Suzuki, *Aspects of the screw function corresponding to the Riemann zeta-function*, Journal of the London Mathematical Society 108 (2023), 1448–1487, DOI `10.1112/jlms.12785`. 2. M. Suzuki, *Weil's quadratic form via the screw function*, arXiv:`2606.09096` (2026). This develops a finite-interval operator framework and formulates a spectral-limit conjecture; the conjectural limit is not used as a theorem here. 3. D. Carando, A. Defant, F. Marceca, I. Schoolmann, *Vector-valued general Dirichlet series*, Studia Mathematica 258 (2021), 269–316. 4. A. Defant, A. Pérez, *Hardy spaces of vector-valued Dirichlet series*, Studia Mathematica 243 (2018), 53–78. These references provide context for vector-valued Dirichlet-series and operator language but do not replace the explicit proof given above. --- # 27. State transition The canonical transition is: ```text CSM_RH v0.2 -> CSM_RH v0.3 ``` with: ```text GLM semantic correction: GLM_Backward Search = WITHDRAWN NAME GLM-5.3-Flash = WORKER / PROVIDER O-RH-005-M: OPEN -> CLOSED_AT_EXPONENTIAL_TYPE F-RH-004: SPLIT INTO F-RH-004A OPEN / LOW PRIORITY F-RH-004B CLOSED B-RH-005: old pointwise lower-envelope bridge remains OPEN B-RH-007: HILBERT_PACKET_TO_EXACT_EXPONENTIAL_TYPE = CERTIFIED B-RH-008: EMAE_TO_FIXED_ZERO_STRIP = CERTIFIED S-RH-007: LOSSLESS_WITNESS_GLOBALIZER -> PARTIALLY_CLOSED major-arc instance = CLOSED Weil instance = OPEN ``` --- # 28. Final status ```text RH = OPEN WEIL GLOBALIZATION = OPEN FIXED-APERTURE GLOBALIZATION = CLOSED FIXED-APERTURE RH-COMPLETE TAIL = OPEN MAJOR-ARC PACKET GLOBALIZATION = CLOSED_AT_EXPONENTIAL_TYPE MAJOR-ARC EMAE FIXED POWER = OPEN EMAE STRENGTH AUDIT = CERTIFIED GLM = WORKER / PROVIDER CSM_RH = PROTOCOL / STATE SPACE NEXT CAMPAIGN: ARITHMETIC_EMAE_FIXED_POWER ``` The main mathematical conclusion is: $$ \boxed{ \inf \left\{ a: \left\| \sum_\rho e^{\rho T} \mathcal W_\rho \right\|_{\mathcal H_{w,U}} = O(e^{aT}) \right\} = \Theta_\zeta. } $$ Therefore: $$ \boxed{ \operatorname{type} \left( \mathcal E_1 \right) = 1+2\Theta_\zeta = 2+2\Delta_\zeta. } $$ The major-arc zero packet cannot hide an off-axis zero at exponential-type level. The remaining major-arc obstruction is arithmetic, not representational.