What was already known
Two levels of Strassen’s bilinear algorithm give rank 49. Older commutative methods such as Waksman’s use fewer scalar products in a different model and do not give this tensor decomposition.
What changed and why it matters
A fixed-dimension improvement in complex arithmetic.
The AI contribution
Gemini-generated changes guide an evolutionary search for tensor decompositions.
Limitations
This does not establish a new asymptotic exponent or universal hardware speedup. It is not the same field as AlphaTensor’s F₂ result. AIProofWatch has not checked the decomposition. The novelty comparison concerns bilinear tensor rank, not the fewest products in every commutative arithmetic model. Waksman-type algorithms cannot be substituted as tensor-rank baselines.
Verification evidence
Evidence labels show what supports the claim. Importance rates its impact if correct.
Author claim
Algorithm reported in §3.1; no independent check recorded here.
Correction and publication history
Public discussion revealed that the novelty comparison needed the bilinear tensor-rank model. Clarify the baseline; retain notable importance. Non-notifying correction.
Read this published revisionHistorical seed, researched for the initial catalog. Initial alerting disabled.
Read this published revision