0
$\begingroup$

For which $x \in \Bbb{R}$ matrix $A$ has all positive eigenvalues, if matrix $A$ is given $$ A= \begin{bmatrix} 1 & 2 & 3 \\ 2 & x & 4 \\ 3 & 4 & 5 \\ \end{bmatrix}$$ $$-$$ We know that when matrix is symmetric and positive definite then it has positive eigenvalues. Positive definite matrices have positive determinant, trace and positive diagonal elements..so, I can conclude that det$(A)>0$ then $12-4x>0$, and $x<3$, and we have $x>0$ if matrix is positive definite. At the end conclusion is $x \in (0,3) $. My problem is when I use 1 of 2 for x I don't get all positive eignevalues, and that is contradiction (maybe I'm wrong). $$ $$ My question is, can I say that symmetric matrix can have all positive eigenvalues if it's not positive definite? And can someone help me to find $x$ in this problem..

$\endgroup$
1
$\begingroup$

Since $A$ is symmetric, if all eigenvalues of $A$ are positive, then $A$ is positive definite. We know that for $A$ to be positive definite, this submatrix of A $$ \begin{bmatrix} 1 & 2 \\ 2 & x \end{bmatrix} $$ needs to have a positive determinant, i.e. $ x > 4 $. (This is known as Sylvester's criterion)

As you already mentinoned, we also need $\det(A) > 0$, and that implies $ x < 3 $. From this we can conclude, that no matter the value of $x$, $A$ will never be positive definite.

Alternatively, you can also note that for $ v = (2, 0, -1)^T $, $$ v^T A v = v^T \begin{bmatrix} -1 \\ 0 \\ 1 \end{bmatrix} = -3 < 0, $$ which implies that $A$ is not positive definite, and this calculation doesn't actually depend on $x$.

$\endgroup$

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.