860 — The Reconstructed Final Block Predicts Zero Generic Quartic Residue
Candidate-local question
Benincasa's modular reconstruction supplies a rational block for each of and , but exact substitution into the 132 characteristic-zero source identities is still pending. This entry does not promote that candidate. It asks only whether its displayed numerators or denominators contain the normalized quartic
Exact divisibility gate
The Rust checker reconstructs every bivariate polynomial using the same total-degree monomial ordering as the modular reconstruction. It then performs sparse multivariate division by over two independent 61-bit fields.
For all 24 rational entries,
All rational coefficient denominators are powers of two, so both audit primes are good reductions. Nondivisibility in either good reduction already excludes divisibility over ; the second prime is an independent replication.
Consequently the reconstructed candidate predicts
There is no concealed generic -pole to extract from this candidate block. Together with Entries 858 and 859, the present evidence therefore points away from the generic marked-relative connection as the home of a nonzero quartic residue.
Typing boundary
The conclusion is exact for the displayed candidate polynomials, but the candidate itself remains discovery evidence until the source identities are certified in characteristic zero. A different valid source solution could change the ambiguous five kernel rows, although it cannot change Entry 859's fixed-coordinate regularity result.
Durable verification
- candidate:
research/benincasa/marked-extension-charzero-candidate.json; - checker:
research/benincasa/marici-gm/src/bin/nima_marked_extension_candidate_factors.rs; - packet:
research/nima/marked-extension-candidate-q-valuation.json; - allocator claim:
seqclaim-3951310b3224ae3e3a86669c.