Exploiting the Structure in Tensor Decompositions for Matrix Multiplication
Manuel Kauers ∗, Jakob Moosbauer †, Isaac Wood ‡ · 2026
Read Exploiting the Structure in Tensor Decompositions for Matrix Multiplication by Manuel Kauers ∗, Jakob Moosbauer †, Isaac Wood ‡ (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-11041.txt.
Description
Fast matrix multiplication leans on tensor decompositions, but rank alone can mislead. Manuel Kauers and two co-authors exploit special structure in such decompositions to beat the exponent their rank would suggest. Their new recipe uses traits of the tensors to shave the exponent down.
For six-by-six multiplication, the exponent falls from about two point eight zero seven five to about two point eight zero one nine. The leading coefficient stays sane. The gain is small in digits but real in theory, and it shows that structure-aware methods can still squeeze speed from a problem studied for decades.
Catalog record
| CREATOR | Manuel Kauers ∗, Jakob Moosbauer †, Isaac Wood ‡ |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-11041