Answers are grounded in this paper and cite their sources.
Reveals algebraic challenges in achieving rank-22 matrix multiplications, indicating limitations in current strategies.
This document makes no claim about the rank of ⟨3,3,3⟩, which remains open (19 ≤ R ≤ 23). We investigate whether known rank-23 decompositions of the 3×3 matrix multiplication tensor ⟨3,3,3⟩ (Laderman-type, Perminov, Sedoglavic families) can be completed to exact rank-22 decompositions via local algebraic operations. We introduce an exact, rank-preserving completion order c(S,R), sharper than the conventional residual support metric ℓ₀, and calibrate it against a known exact control witness. On eight identified frontier candidates we show: (1) no candidate admits direct one- or two-term absorption of the residual; (2) an algebraically certified non-existence result for joint W-solutions under modular reduction (GF(2), GF(3)) on 20 stratified candidates; (3) this pattern extends consistently to modular computation over {2,3,5,7,11,13}. We additionally document a broadly supported but only empirically evidenced reentry invariant under ε-perturbation. All results are negative with respect to rank 22 within the investigated search space; no global impossibility claim is made.
No takes yet. Share an insight, caveat, or question.
Stefan Beuchert (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: