1
$\begingroup$

in general, if a matrix is symmetric and positive definite over the real field, it is positive definite over the complex field. That if we drop the symmetric condition, this is no longer true is shown here https://en.wikipedia.org/wiki/Positive-definite_matrix#Consistency_between_real_and_complex_definitions .

Now, I was asking myself, if we instead of symmetry, assume the matrix having no negative entries, does it hold that it is positive definite over the real field?

$\endgroup$
1
$\begingroup$

This does not hold: Consider $M = \begin{pmatrix}1 & 3\\ 1 & 5\end{pmatrix}$. Then for every $x = (a,b)^T\in \mathbb R^n$ we have $$ x^TMx = (a,b)\cdot \begin{pmatrix}a+3b\\ a+5b\end{pmatrix} = a^2 + 3ab + ab + 5b^2 = a^2 + 4ab + 5b^2 = (a+2b)^2 + b^2 \ge0. $$ But for $x = (1,i)^T\in \mathbb C^2$ we have $$ x^*Mx = (1,-i)\cdot \begin{pmatrix}1 + 3i\\ 1 + 5i\end{pmatrix} = 1+3i -i + 5 = 6+2i, $$ which is not a real number.

$\endgroup$

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.