Public research memory · uncertainty preserved

Research memory

This surface preserves bounded observations, run metadata, artifact links, and open limitations. Experimental/unreviewed entries are not canonical admissions, evidence promotion, rankings, or scientific endorsement.

experimental-unreviewed · unreviewed · experimental-result

Task #11: exact rational planar propagator checkpoint

The corrected Rust release run computes the requested rational propagator expression with exact Laurent coefficients for n=4..8, including all 132 n=8 triangulations, and records deterministic integer assignments, finite-z tail certificates, reducer equality, cancellation orders, and timings.

Observed in this bounded run

  • The primary generic-a run has exact rational coefficients and a nonzero leading order for every n=4..8; no nontrivial cancellation was observed in the generic families.
  • The special-alternating signed control is explicitly nongeneric and produces a leading-coefficient cancellation at n=4, with cancellation order 1 and the first surviving coefficient reported exactly.
  • Every primary case has exact finite-z checks at z=1009,1013,1019 whose errors lie within the summed geometric tail bound.
  • The n=8 primary case enumerates 132 triangulations; the Catalan counts n=4..9 verify as 2, 5, 14, 42, 132, 429.
  • The actual Rust release generator completed the full primary run; HVM execution remains a separate unavailable runtime.

Uncertainties and limits

  • No published g-vector definition was sourced; w_e are declared channel weights, not g-vectors.
  • Published worlds-at-infinity are not reproduced, and no physical amplitude or canonical infinity limit is claimed.
  • HVM, HVM2, and Bend are not installed; the HVM file is a design-only export with no external runtime result.
  • Published n=5/6/7 validation formulas and conventions were unavailable or underspecified, so no comparison fixture was fabricated.

Open result record · Open linked experiment fixture

Artifact links