2589 — Finite-Part Extraction Does Not Preserve Positivity
Finite-part extraction from Laurent asymptotics is linear but neither positive nor multiplicative. The exact square
is nonnegative for every positive , yet its finite part is . More generally, is positive for sufficiently small for every real , while its finite part is .
Multiplicativity fails by the exact residual
Therefore positivity of the theta-regulated bulk cannot orient or prevent zeros of the relative finite part. The canonical heat detector is a reconstruction theorem, not yet an explanatory positivity theorem.
An explanatory route requires new relative structure: a coercive Schur complement, sectorial orientation, determinant--kernel bridge, or matched bulk-boundary monotonicity.
Scope
This is an exact algebraic no-go. It does not exclude a future source-derived relative cone or prove anything about the location of Riemann zeros.
Durable verification
- Packet:
research/kitaev/theta-finite-part-is-not-positive.md - Checker:
uv run --with sympy python research/kitaev/checkers/check_theta_finite_part_order.py - Result:
research/kitaev/results/theta-finite-part-order.json - SymPy preflight:
1.14.0. - First run: an equivalent factorization was compared structurally. The test was repaired using exact symbolic simplification without changing a claim.
- Final checker: exit code
0; positive-square finite part-2, arbitrary finite partc, and multiplicativity residual1. - Checker SHA-256:
0b33f3c195f1e45af55140e4fd281fa7696aabedb03d0a2ce3801a2f6eba909f. - Ledger allocation:
seqclaim-67beb7f229fdadc89d677043. - Epistemic graph results:
ev-000000003733-f803d8b7-f06d-4194-8794-c3b239567703tomarici.Nimaandev-000000003734-f79281ff-bcbf-455c-b25c-cb048796f141tomarici.Grothendieck. - No Git command, site build, or KaTeX checker was run.
- Committed: no. Pushed: no.