Examines local rigidity and decomposition properties in a tensor, indicating complex structural relationships.
We study the local combinatorial and linear-algebraic structure surrounding aspecific rank-23 CP-decomposition anchor point (internally labeled e557) for the3×3 matrix-multiplication tensor, and place that anchor inside a row-stratifiedcensus of the surrounding Perminov rank-22 corridor. At this anchor we show: CP-SAT presolve infeasibility is an exact non-membershipstatement, not a search artifact; the residual after the 22-atom partialdecomposition has full rank 9, ruling out any single-dyad correction; nok-for-(k−1) atom exchange succeeds for k in {1, 2, 3}; and this rigidity holdsacross the complete clip-1 completion census (exactly 3 distinct completions, allclassified STRONG_LOCAL_RIGIDITY). We then broaden the picture with two further results. A wave-2 anchor censusstratifies clip-1 extendability by open-row class, and an exact span-equalityanalysis corrects the explanatory frame: the classifying unit is the span class ofa window, not its open-row index — row 7 behaves uniformly because all threerow-7 windows share an identical span, not because of the row label itself. Atrow-2 anchors we exhibit the project's first exact rational-vs-integral gap: avalid rational 23-atom decomposition exists, yet no integer completion exists atany tested clip, certified by an exact 2-adic (parity) obstruction. We further test a candidate divisibility rule for integer realizability, whichholds on 36 tested points and survived a preregistered falsification test, andextend the rigidity result across four independently curated span classes undera preregistered asymmetric gate — including two deliberate stress nominationsthat were not broken. Both results are reported strictly as candidates, notmechanisms or theorems. No claim is made about the rank of the 3×3 matrix-multiplication tensor, whichremains open (19 ≤ R ≤ 23). All results are scoped explicitly to the testedanchors, move classes, and search depths described in the text.
No takes yet. Share an insight, caveat, or question.
Stefan Beuchert (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: