5
$\begingroup$

$M$ is square and $$M(i,j)=0, i=j$$ $$M(i,j)>0, i\ne j$$ Is $M$ full-rank or invertible?

Actually the $M$ I am studying has much stronger properties but I guess the simple conditions above might be enough to make $M$ non-singular. The stronger properties of $M$ are:

  1. All elements in $M$ are non-negative integers between $0$ and $N$;

  2. The sum of each row is equal to $N$.


Edit

$N$ is not the dimension of $M$. It's just a constant positive integer.

I did search for this problem, but it seemed there was no much work on such matrices. It looks easy, but I don't know how to prove it and I couldn't find a counter-example either.

$\endgroup$
7
$\begingroup$

The matrix $M$ doesn't have to invertible. I give a simple counterexample here. $$M=\begin{pmatrix} 0&2&1&1\\ 2&0&1&1\\ 1&1&0&2\\ 1&1&2&0 \end{pmatrix}$$ You will see $r_1+r_2-r_3-r_4=0$ which indicates they are linear dependent.

$\endgroup$
  • $\begingroup$ Thanks a lot Shuchang. What if I have even stronger condition on $M$ -- the subdiagonal entries, which are directly below and to the left of the main diagonal, are also all zeros. Other properties of $M$ remain the same. Must $M$ be nonsingular now? ($Dim(M) \ge 3$) $\endgroup$ – D. Chen Oct 6 '13 at 2:44
  • $\begingroup$ @TomSmith Triangular matrix is invertible if and only if it doesn't have zero diagonal element. $\endgroup$ – Shuchang Oct 6 '13 at 2:49
  • $\begingroup$ Thanks. I meant only the entries on the diagnoal and subdiagonal of $M$ were zeros so it was not a triangular matrix. An example could be changing your matrix above to: $$M= \begin{pmatrix} 0 & 1 & 1 & 1 \\ 0 & 0 & 2 & 1 \\ 1 & 0 & 0 & 2 \\ 1 & 2 & 0 & 0 \\ \end{pmatrix}$$ Do you think $M$ with such constraints must be nonsingular? $\endgroup$ – D. Chen Oct 6 '13 at 2:55
  • $\begingroup$ Sorry for any confusion and thanks for your help. $\endgroup$ – D. Chen Oct 6 '13 at 3:01
  • $\begingroup$ @TomSmith Oh, sorry. But still, invertiability still cannot be guaranteed. A counterexample is given using 6 by 6 matrix by considering the first 3 rows add up to the last 3 rows. $\endgroup$ – Shuchang Oct 6 '13 at 3:05

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.