I am multiplying two sparse matrices $A$ and $A^T$ such that I have $A^T*A$. From what I know the complexity of this operation depends on nnz(A). Yet, when I generate a random matrix with fixed nnz(A) but increasing number of rows and compute the product $A^T*A$ I see that the CPU time it takes to perform this operation increases. Am I missing something ?
MATLAB: Sparse matrix multiplication complexity and CPU time
MATLABmatrixmatrix arraymatrix manipulationmemorymultiplicationsparse
Best Answer