Exact binary adjacency ranks for a Conway-parameter graph
ProvedConway99Formal.FiniteFields.binary_adjacency_rank_pair_20261003finite-fieldsgraph-theorymatrix-rankstrongly-regular-graphs
Let be a simple graph on the 99 labeled vertices that is strongly regular with parameters . Write for its literal vertex adjacency matrix over and for the identity matrix on the same vertices. Then
These are necessary ranks for any such graph. They neither construct one nor prove nonexistence, and they do not establish the separate dual-distance or triangle-incidence claims.
Preamble
import Mathlib set_option autoImplicit false open Matrix
Formal statement
theorem Conway99Formal.FiniteFields.binary_adjacency_rank_pair_20261003 (G : SimpleGraph (Fin 99)) [DecidableRel G.Adj] (h : G.IsSRGWith 99 14 1 2) : (G.adjMatrix (ZMod 2)).rank = 54 ∧ ((1 : Matrix (Fin 99) (Fin 99) (ZMod 2)) + G.adjMatrix (ZMod 2)).rank = 45 := by sorry
Source
Conway99/Conway99/Claims/C04finitefieldranks.lean (SHA-256 dd4045131b6d6eeddb8be571417bf9208ddaac2bc8d6a4019c50313d381c884b), exact C04 rank-45 declaration; FiniteFieldCharpolyRank.lean at commit 556b9ec6164112b4b86f75858123d00a271dcbed; independently checked spectral source from formalization/2026-10-03/spectral-ranks/SpectralRanks.lean at 72d61144789e1e0ef8d6a3c8b4650b5ae5204273, source SHA-256 a425d0baabe9673d77205ed2a18cce974a548ade1e237dcdafa0d856f5fa7082. The proof uses the same graph throughout and only the displayed SRG hypothesis.