A normal form for matrix multiplication schemes
From MaRDI portal
Abstract: Schemes for exact multiplication of small matrices have a large symmetry group. This group defines an equivalence relation on the set of multiplication schemes. There are algorithms to decide whether two schemes are equivalent. However, for a large number of schemes a pairwise equivalence check becomes cumbersome. In this paper we propose an algorithm to compute a normal form of matrix multiplication schemes. This allows us to decide pairwise equivalence of a larger number of schemes efficiently.
Cites work
- A noncommutative algorithm for multiplying 3×3 matrices using 23 multiplications
- Equivalent polyadic decompositions of matrix multiplication tensors
- Fast commutative matrix algorithms
- Gaussian elimination is not optimal
- scientific article; zbMATH DE number 976329 (Why is no real title available?)
- New ways to multiply \(3 \times 3\)-matrices
- Noncommutative Bilinear Algorithms for 3 \times 3 Matrix Multiplication
- On multiplication of 2 2 matrices
- On the complexity of the multiplication of matrices of small formats
- On the inequivalence of bilinear algorithms for \(3\times 3\) matrix multiplication
- On varieties of optimal algorithms for the computation of bilinear mappings. II. Optimal algorithms for \(2\times 2\)-matrix multiplication
- The bilinear complexity and practical algorithms for matrix multiplication
Cited in
(3)
This page was built for publication: A normal form for matrix multiplication schemes
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6108729)
