Trilinear aggregating with implicit canceling for a new acceleration of matrix multiplication

AbstractA new improvement of author's techniques of trilinear aggregating, uniting and canceling, is presented and applied to accelerate multiplication of matrices of moderate sizes. This results in the Exact Computing algorithms for n×n matrix multiplication in only (n+2)[1.75(n+2)+(n2+ 4n+3)/3] essential multiplication steps for arbitrary even n. Also the new techniques allow us to simplify the design of the fastest known APA-algorithms for matrix multiplication