A group-theoretic approach to fast matrix multiplication
arXiv:math/0307321 · doi:10.1109/SFCS.2003.1238217
Abstract
We develop a new, group-theoretic approach to bounding the exponent of matrix multiplication. There are two components to this approach: (1) identifying groups G that admit a certain type of embedding of matrix multiplication into the group algebra C[G], and (2) controlling the dimensions of the irreducible representations of such groups. We present machinery and examples to support (1), including a proof that certain families of groups of order n^(2 + o(1)) support n-by-n matrix multiplication, a necessary condition for the approach to yield exponent 2. Although we cannot yet completely achieve both (1) and (2), we hope that it may be possible, and we suggest potential routes to that result using the constructions in this paper.
12 pages, 1 figure, only updates from previous version are page numbers and copyright information
Cited by in corpus (8)
- Group-theoretic algorithms for matrix multiplication
- Fast linear algebra is stable
- On cap sets and the group-theoretic approach to matrix multiplication
- Fast matrix multiplication is stable
- Faster Algorithms for Rectangular Matrix Multiplication
- The tensor rank of 5x5 matrices multiplication is bounded by 98 and its border rank by 89
- Universal points in the asymptotic spectrum of tensors
- On Matrix Multiplication and Polynomial Identity Testing