Correction and Research Update: Ferrers Constructions and the KR1 Bijection Program

Erratum: 26 October 2026. This notice corrects the earlier draft of From a 2D Corner Obstruction to 3D and d-Dimensional Ferrers Diagram Bijections: Tableau Positivity and the Kanade–Russell I₁ Identity and its accompanying descriptions.

The earlier “complete”, “universal” and “lookup-free” descriptions of the general KR1 construction are withdrawn. The current work does not establish a complete bijective proof of KR1. The revised manuscript retains valid local results and finite computations, and records the limitations explicitly.

Two mathematical corrections

Tableau routing and inversion weight. The previous displayed definition of Φ₂₂₁ could select a valid target with the wrong graph-inversion weight. For h=(2,2,4,5,5), the source T=((1,4),(2,5),(3,)) has weight 1, but the old second branch gives ((2,3,5),(1,4)), of weight 2. The corrected definition selects the first of the ordered candidates that is valid and has the source weight. It selects ((1,3,5),(2,4)), of weight 1, in this example. This agrees with the Python acceptance rule; the old displayed formula did not.

Dual Jacobi–Trudi permutation recovery. With β indexed by determinant columns and σ mapping rows to columns, the decreasing order τ of the shifted keys recovers σ=τ, rather than τ⁻¹. The written proof now establishes the recovered partition size, both inverse identities and the determinant sign. Tests exercise recovery of the actual columns and permutation, beyond comparing sets of beta vectors.

The KR1 obstruction

At weight 41, two distinct cluster configurations have the same column image under the current rules:

The proposed inverse returns the lower-sector cluster. This is a genuine failure of injectivity for the current map, and refutes its extension to all weights. It does not refute the KR1 identity. Nine further cluster states at weight 41 lack apex assignments; extending the finite table alone cannot fix the exhibited overlap.

Results preserved and evidence extended

The corrected finite tableau-routing checks pass at n=5 and n=6. Full KR1 roundtrips pass for 5,461 states at weights 0–40. Separately, the restricted a=3,b=0 sector passes 4,491 states through weight 60. These are exact finite checks, not proofs for arbitrary weight. The general 6D cone membership and image-exclusion statement remains a conjecture.

The supplementary Lean certificate proves specified conditional local lemmas and arithmetic identities. It does not prove the complete routing procedures or a general KR1 bijection. The corrected beta-rearrangement theorem has a written proof; it is not formalized in that certificate. Independent review of the corrections and passing checks improve the assurance of the revised record without establishing novelty or completing the KR1 program.

The next construction target is an invariant preventing higher-sector operators from entering lower-sector images, followed by a proved inverse. This is continuing research, not an announced general solution.

Corrected research update · Revised article · PDF · Verification package · Correction details and evidence

Author: Christopher Yamba. Original collaborator: Antigravity (Google DeepMind). Original draft record: 2026-10-24T22:42:00Z. Correction and revision 2: 26 October 2026. The original timestamp is retained as provenance, not as the publication date of these corrections.