2
$\begingroup$

Assume $P$ is a non-negative positive definite matrix. It is well known what $P^{-1}$ is also positive definite and thus all its diagonal entries are positive. Can we say something about the off diagonal entries of $P^{-1}$? In particular can we say that $$P_{ij}>0 \implies P^{-1}_{ij}<0,\space \forall i\neq j$$.

$\endgroup$
4
$\begingroup$

Here is a random counterexample to your conjecture: $$ C=\pmatrix{4&4&4\\ 4&8&1\\ 4&1&8},\ \det(C)=28, \ C^{-1}=\frac{1}{28}\pmatrix{63&-28&-28\\ -28&16&12\\ -28&12&16}. $$ One can easily check that $C\succ0$ using Sylvester's criterion.

For those entrywise positive $P$s that satisfy your conjecture, $P^{-1}$ belongs to the class of $M$-matrices.

$\endgroup$
  • $\begingroup$ Thanks for your comment. The reason I made this conjecture is because I observed this in the positive definite matrices defined like this problem math.stackexchange.com/questions/567507/… (I could not prove it yet) $\endgroup$ – user54626 Nov 16 '13 at 21:45

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.