Casts binary Goldbach as a self-correlation of the prime field on the 6N skeleton: the Goldbach complex tensor integral I (2E) =sumₓ H (x) H (2E-x) is exactly the count r (2E) of prime pairs summing to 2E, and its Hardy-Littlewood normalization is the Goldbach singular series SG (2E) =2*Pi2*prodₐ|₄, ₐ>₂ (q-1) / (q-2), with Pi2 the twin-prime constant. Main statement (a FLOOR on the heuristic main term): the product defining Pi2 converges absolutely under the O (1/q²) decay of the local factors, and SG (2E) >= 2*Pi2 ~ 1. 3203 > 0 for every even 2E, with equality iff E is a power of two. Numerical confirmation to 2E=10⁸ (the Goldbach comet) shows the split the floor predicts: the RAW count r (2E) grows without bound (minimum 22 at 2E=1024 rising to 1. 5e5 at 2²6), while the NORMALIZED count r (2E) (ln 2E) ²/ (2E) is held above a positive floor by the powers of two (measured ~0. 74 at 10⁸, drifting slowly toward Pi2 with the finite-size 1+2/ln correction; equivalently 2*Pi2 in the representation-counting convention), reproducing the Part XIII singular-series collapse at 10⁸ scale. SCOPE, stated without hedging: the floor bounds the SINGULAR SERIES (the expected main term), NOT the actual count; the Hardy-Littlewood asymptotic is itself conjectural and proving r (2E) >0 for all even 2E IS binary Goldbach, which is OPEN. Positivity of the main term is not positivity of the count. What is rigorously known (ternary Goldbach, Helfgott; Chen's p+P2 theorem) is not reproved here. This is the HL heuristic main term plus a numerical check, not a step toward a proof, and claims nothing about the completeness of Goldbach representations.
Ruqing Chen (Mon,) studied this question.