So I have some very large matrices with nearly all elements being zero. Are matrix multiplication libraries able to avoid having to multiply the zero elements? More importantly, would they at least ...