MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am trying to determine when a certain parametric matrix is inverse-positive (it's actually the one about which I asked in Explicit formula for Cholesky factorization in a special case, but the question might have general interest). This matrix is not a $Z$-matrix so the whole body of knowledge that had been developed for them is not suitable.

What I'd like to find is a simple sufficient condition that I can easily analyze. The best approximation to that that I've found is Peris's criterion using $B$-splittings but it still didn't help me enough.

Are you aware of any such results?

share|cite|improve this question
One example I've just found is - where just such results are obtained for small sign-changing perturbations of tridiagonal $M$-matrices. I'm looking for more general results. – Felix Goldberg Apr 25 '12 at 10:20
up vote 1 down vote accepted

Sorry for promoting my own results, but I think the condition in my old paper "A sufficient condition for the monotonicity of a positive definite matrix" (Computational Mathematics and Mathematical Physics vol. 41, No 9., pp. 1237-1238) may be of help. (Unfortunately, I lost the file years ago). I don't know whether it actually works, for you didn't say much about your $Q$, but it looks promising.

share|cite|improve this answer
Is there an English version anywhere? (I can read the Russian but I like to have an English version). – Felix Goldberg Jan 11 '13 at 14:44
This journal is translated, so there must be an English copy somewhere. – Alex Gavrilov Jan 11 '13 at 15:14
You may try this: – Alex Gavrilov Jan 11 '13 at 15:16
I accepted the answer, it looks interesting! The last link you gave didn't seem to lead to a softcopy I could obtain but I've read the Russian text and it's good enough for me! I'll see now if it helps with my original problem but it's a great result anyway. – Felix Goldberg Jan 21 '13 at 13:55
Thanks. Good luck. – Alex Gavrilov Jan 22 '13 at 11:11

$\mathbb{R}^n_+ \subseteq A(\mathbb{R}^n_+)$ is one sufficient condition for inverse positivity.

share|cite|improve this answer
That's the definition essentially... – Felix Goldberg Oct 5 '12 at 11:51
What about semipositivity ? An $m \times n$ matrix is said to be semipositive if there exists $x \in \mathbb{R}^n_+$ such that $Ax \in int(\mathbb{R}^m_+)$; $A$ is said to be minimally semipositive if in addition to being semipositive, no proper $m \times p$ submatrix is semipositive. For a square matrix, minimal semipositivity is the same as inverse positivity. I do not know if MSP is easy to verify. Two references are : Johnson, Kerr and Stanford, Semipositivity of matrices, LAMA, 37(4) (1994), 265-271 and H.J Werner, Characterizations of semipositivity, LAMA, 37(4) (1994), 273-278. – user27020 Oct 6 '12 at 4:10

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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