0
$\begingroup$

Assume that $\mathbf{A}_{M\times N}$ is a sparse complex matrix. Then, what is the complexity of computation of its pseudo inverse, i.e., $$\mathbf{A}^{\mathrm{H}}(\mathbf{A}\mathbf{A}^{\mathrm{H}})^{-1}.$$ For a usual $\mathbf{A}$ (not sparse), the complexity of computation is $$\alpha_1 N M^{\nu-1}+\alpha_2 M^{\nu}+\alpha_3 M^2 N^{\nu-2},$$ where $2.37 \leq\nu\leq 3$.

Assume that $\mathbf{A}$ is sparse in a way that $\mathbf{A}\mathbf{A}^{\mathrm{H}}$ is also sparse. Is it better for the sparse case?

$\endgroup$
2
  • $\begingroup$ What random distribution on the sparse matrices do you use? $\endgroup$ Commented Jan 31, 2020 at 10:04
  • $\begingroup$ It is not important. For example, consider the entries have bernouli distribution with $p<<1$. $\endgroup$
    – Math_Y
    Commented Jan 31, 2020 at 10:07

0

You must log in to answer this question.

Browse other questions tagged .