AlphaEvolve helps refresh the upper bound of matrix multiplication exponents.
The new paper uses modern optimization methods and AlphaEvolve to improve the upper bound of the matrix multiplication exponent from 2.371339 to less than 2.371177; this is an asymptotic advancement in computational complexity, but does not mean that existing matrix operations will immediately achieve the same speedup.
A new paper uses modern optimization methods and AlphaEvolve to improve the known upper bound of the matrix multiplication exponent to less than 2.371177.
The paper gives a previous upper bound of 2.371339. This achievement represents a theoretical advance in asymptotic computational complexity and cannot be directly interpreted as an immediate performance improvement of the same proportion for existing hardware or conventional matrix operations.
refer to:arXiv paper