Let $D$ be a diagonally dominant matrix (both row and column-wise). Assume $D$ is not necessarily symmetric.

Consider the following block representation of $D^{-1}$: \begin{equation} D^{-1}=\begin{bmatrix} B_1 & B_2 \\ B_3 & B_4 \end{bmatrix}. \end{equation} Here assume that $B_i$ are all square matrices.

My question is whether $C=(B_1^{-1} + B_1^{-T})$ is also diagonally dominant. Note that if $B_1=D^{-1}$, then we have $C=(D+D^T)$, and this trivially holds.


Actually, I found the answer. I'm including it here in case others find it helpful as well. This paper (siam.org/meetings/la03/proceedings/LeiTGWLZ.pdf) establishes that if $D=[D_1,D_2;D_3,D_4]$ is a diagonally dominant matrix, then Schur complement of $D_4$ in $D$ given by $D_1−D_2D_4^{-1}D_3$ is also diagonally dominant. $B^{−1}_1$ is equal to this matrix (by matrix inversion lemma), hence the result follows.

  • $\begingroup$ This should have remained a comment; closure of diagonal dominance under Schur complements has been also previously noted in some MO answers. $\endgroup$ – Suvrit Aug 22 '17 at 16:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.