Blueprint for the Ternary Goldbach Formalization

8 Reusable formal infrastructure

These are significant ordinary-Lean or conditional interfaces developed or exposed alongside the production proof. Some occur in the final theorem’s current cone; the large-sieve and squarefree-density nodes are intentionally shown as reusable side results rather than claimed production dependencies.

Theorem 8.1 Classical additive large sieve

Trust tier: BT-U. The classical additive large-sieve inequality holds for arbitrary complex coefficients in the repository’s separated-point formulation.

Source: montgomery-vaughan-hilbert-1974, Theorem 1 and Corollary 1, pp. 73–74 [adapted] Source: yangjit-2022, equations 1.2–1.4 and Lemmas 2.1–2.2 [method] Source: iwaniec-kowalski-2004, Theorem 7.7 [method]

Proof

Apply the unconditional cosecant Hilbert norm bound to the off-diagonal Gram matrix and add the diagonal contribution.

Theorem 8.2 Phi-gain large sieve

Trust tier: BT-U. Averaging over squarefree dilations yields the Montgomery \(\sum \mu ^2(r)/\varphi (r)\) amplification of the large sieve.

Source: helfgott-minor-2013, Section 4.2, Lemma 4.3 and equations 4.38–4.40, p. 45 [method] Source: iwaniec-kowalski-2004, Lemma 7.15 [adapted]

Proof

Apply Montgomery’s prime-support estimate at each squarefree dilation, sum and identify the resulting fan energy, then apply the classical large sieve once to the full well-spaced fan.

Theorem 8.3 Single-point phi gain

Trust tier: BT-U. The amplified inequality implies the single-point \((\varphi (q)/q)\log R\) large-sieve gain.

Source: helfgott-minor-2013, Section 4.2, Lemma 4.3 and equations 4.38–4.40, p. 45 [derived]

Proof

Lower-bound the squarefree amplification sum by its explicit totient/logarithm estimate and rearrange.

Theorem 8.4 Explicit squarefree density

Trust tier: BT-U. The squarefree count satisfies \(|Q(x)-(6/\pi ^2)x|\le 3\sqrt{x}+2\).

Source: cohen-dress-el-marraki-2007, Lemma 1, p. 55 [method]

Proof

Expand \(\mu ^2\) by square divisors, isolate the main zeta value, and bound the floor and tail errors explicitly.

Theorem 8.5 Ramaré m-star continuation

Trust tier: BT-C. The finite \(m^\star \) table through \(1.4\cdot 10^8\) and the explicit little-Mertens tail imply \(m^\star (N)\log (N+1)\le 4/5\).

Source: ramare-2015, Section 7, equation 7.1 and Lemma 7.1, p. 1375; Lemma 7.4 and equations 7.2–7.3, pp. 1376–1377 [adapted] Source: ramare-2013, Corollary 1.4 and the following sentence, p. 366 [adapted]

Proof

Split at successive fourth roots until the argument enters the finite table, and use the analytic tail bound on every continuation step.

Theorem 8.6 Corrected Ramaré finite adapter

Trust tier: BT-C. A first-Mertens table through \(10^8\) and the four Lemma-7.1 rows imply the readable corrected finite certificate bundle.

Source: ramare-2013, Lemmas 4.5, 6.2, and 6.3 [adapted]

Proof

Normalize the checker contracts to the paper variables and project the required endpoint and coefficient inequalities.

Theorem 8.7 Rosser–Schoenfeld Mertens assembly

Trust tier: BT-C. The displayed finite forms of equations (4.10) and (8.9) imply the Rosser–Schoenfeld Mertens-product upper bound for \(x\ge 286\).

Source: rosser-schoenfeld-1962, Theorem 8, equation 3.29; using Theorem 23, equation 4.10, and Lemma 13, equation 8.9 [derived]

Proof

Split into the seed, finite ladder, and analytic tail ranges; in each range substitute the corresponding explicit input and match at the shared endpoints.

Theorem 8.8 C.17 tail from an integer mirror

Trust tier: BT-C. The literal finite integer-mirror inequality through \(2{,}560{,}000\) implies the singular-series C.17 tail-count estimate.

Proof

Prove soundness of the scaled integer accumulator, convert its finite bound to the real count inequality, and attach the analytic tail.

Theorem 8.9 Sharp Chebyshev psi continuation

Trust tier: BT-C. The finite interval through \(10^7\) and Chirre–Helfgott Corollary 1.3 imply \(\psi (x)\le 1.03883x\) for all \(x\ge 0\).

Source: chirre-helfgott-2025, Corollary 1.3, p. 4; proof p. 36 [derived] Source: rosser-schoenfeld-1962, Theorem 12, equation 3.35 [source-shaped]

Proof

Use the checked step-function interval below the cutoff and the analytic corollary above it, verifying the overlap constants exactly.

Theorem 8.10 Sharp psi-minus-theta continuation

Trust tier: BT-C. The finite interval through \(10^6\) and the sharp psi input imply \(\psi (x)-\vartheta (x)\le 1.4262\sqrt{x}\).

Source: rosser-schoenfeld-1962, Theorem 13, equation 3.36 [source-shaped]

Proof

Split the prime powers by exponent, control the square term with the psi input, and fold the higher powers into an explicit polynomial envelope.

Theorem 8.11 Chirre–Helfgott Lemma A.6 continuation

Trust tier: BT-C. Middle-interval nonnegativity and the von-Mangoldt prefix contract imply \(t g(t)\ge -2.9702\) on the full required domain.

Source: chirre-helfgott-2025, Lemma A.6, p. 42, and Lemma 8.5, pp. 30–31 [adapted]

Proof

Cover the compact middle interval by the first contract, express the endpoint ray through the prefix sum, and use monotonicity to join the ranges.

Theorem 8.12 Backlund f3 interval certificate

Trust tier: BT-U. The Backlund remainder satisfies \(|f_3(\tau )|\le 0.06\) for \(1\le \tau \le 12\).

Source: titchmarsh-1986, Sections 9.3–9.4, Theorems 9.3–9.4 [method]

Proof

Replay a 277-cell ordinary-kernel interval transcript. Each stopped cell uses rational Taylor bounds and the cells cover the whole interval without gaps.