Skip to content

[Completed bounded delivery] Exact width-two torus/cylinder bridge and the finite-length correction #681

Description

@LightChainr

Completed and merged — 2026-09-12

The bounded primary-reading job and its smallest proposed calculation are delivered in #705, merged at eb89e94. This closes the assigned delivery, NOT every all-width spectral or critical-exponent problem.

The initial retrieval identified useful sources but incorrectly treated equality of two leading coefficients as sufficient for an exact finite-length two-eigenvalue identity. The reviewed delivery now gives an actual local square-site counterexample and complete solution: on every honest 2-by-m axis torus, m>=2,

M_{2,m}=tr(T^m)-(1-p^2)^m,
T=[[x,0,y],[0,x,y],[x,x,y]], x=p(1-p), y=p^2.

The full expansion is lambda_+^m+lambda_-^m+x^m-lambda_c^m. Both leading coefficients are ONE; the subleading remainder does not vanish. The cylinder crossing is the interior root of 2p^3+2p^2-1=0, matching Jacobsen Table 2 n=2. Every finite-m torus root is strictly below it; the leading displacement is explicitly exponentially small divided by m. No infinite-square p_c estimate or all-width pTL intertwiner is claimed.

The proof is row-event classification, not extrapolation from counts. Independent lifted-homology enumeration checked all 5456 configurations at 2x2,...,2x6 and every coefficient. The 2x2 polynomial and an integer matrix-power control also pass. Current code head fc19cc7 passed all four CI jobs before merge. Original retrieval history is preserved.

Read notes/p681-cylinder-sector-bridge-20260912.md, scripts/width2_cylinder_exact.py, and results/research-control-20260912/width2-cylinder-exact.json on main. Jacobsen's observed n^-4 law remains observational; the q=1 literal spin-twist specialization is not a usable bridge. Do not rerun a source survey or commission the width-two calculation again.

The genuinely different remaining question is the all-width observable closure and the uniformity needed to compare fixed-width cylinders with fixed-aspect tori. That is re-scoped in #636 as a bounded proof/CPU task, not the old width-10/12 engine purchase. #275's physical operator identification is separate.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    priority:P1Bounded parallel analysis or a concrete reserve direction; not all run at once.

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions