Is it possible for some non-Hermitian matrix $A \in M_n({\mathbb{C}})$ to satisfy

$1)$ $x^t Ax \ge 0 $ for all $x\in \mathbb{R^n}$ ?

$2)$ $x^* Ax \ge 0 $ for all $x \in \mathbb{C^n}$ ?

i take $ A =\begin{bmatrix} 0 &1 \\ 2& 0 \end{bmatrix}$, i thinks both
$1$ and $2$ possible ...

  • $\begingroup$ any one help,,, $\endgroup$ – Messi fifa Aug 18 '18 at 14:25

Real positive semidefinite matrices

There exist non-symmetric real matrices that are positive semidefinite. For example, take

$$A=\begin{pmatrix} 1 & 1\\ -1 & 1 \end{pmatrix}.$$

Then, we get $x^T A x=x_1^2+x_2^2\geq 0\,$ for all $x\in \mathbb{R}^2$. Your example does not work, since $$x^T\begin{pmatrix} 0 &1 \\ 2& 0 \end{pmatrix}x=3x_1x_2,$$ which is not non-negative for all $x\in\mathbb{R}^2$.

It is useful to look at symmetric positive semidefinite matrices because you get nice information about the eigenvalues, spectral decomposition, and so on. But even for non-symmetric positive semidefinite matrices you can say something about the eigenvalues. We know that a symmetric positive semidefinite matrix has only non-negative eigenvalues. A non-symmetric positive semidefinite matrix can have complex eigenvalues and therefore, saying that a complex eigenvalue is positive does not make any sense. But you can show that the real parts of the eigenvalues of a positive semidefinite (not necessarily symmetric) real matrix are non-negative. For example, the matrix $A$ has the eigenvalues $1\pm i$.

Complex positive semidefinite matrices

You can show that every complex positive semidefinite matrix is Hermitian. See, e.g., this post Positive definite matrix must be Hermitian.


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.