1
$\begingroup$

I'm trying to show that $A^* = A^{-3}$ iff $A^* = A$ and $A^2 = I$ given $A$ is invertible and $A\in M_n(\mathbb{C})$ but I'm having trouble with manipulating $A$ and just don't know what's legal or not. $A^*$ here is the adjoint of $A$. Any help would be appreciated!

I believe it's possible to do just by algebraic manipulation on $A$, but super lost as to how to achieve it.

Just to be super clear. $A^*$, the adjoint of $A$ is the transpose of the conjugate of $A$.

$\endgroup$
4
  • $\begingroup$ Do you mean the Adjugate with $A^*$? Usually, $A=A^*$ says that $A$ is normal, and $A^*$ is the conjugate transpose. So then just $A=A^{-3}$? Please clarify the assumptions. $\endgroup$ Nov 12 '21 at 16:04
  • $\begingroup$ Yes $A^*$ here refers to the transpose of the conjugate of $A$. Sorry if the notation wasn't clear $\endgroup$ Nov 12 '21 at 16:11
  • 1
    $\begingroup$ $\Leftarrow$ part is so trivial, because if $A^*=A\ \wedge\ A^2=I$ , then $A^{-3}=IIA^{-3}$. So... the question is prooving $\Rightarrow$ part. $\endgroup$ Nov 12 '21 at 16:15
  • $\begingroup$ Yes agreed, the backwards direction I have shown. I have a solution for the forward direction, but the solution uses unclear/perhaps illegal matrix algebra, so I'm not confident in it as well. $\endgroup$ Nov 12 '21 at 16:17
2
$\begingroup$

If $A^* = A^{-3}$ then $A^*A = AA^*$ ($A$ is normal), so by the Spectral Theorem, $A$ is diagonalizable. Do a change of basis so that $A$ is diagonal. And now you can check that the only complex numbers that satisfy $\bar\lambda = \lambda^{-3}$ are $\lambda = \pm 1$.

If $A$ is diagonal and the diagonal entries are $\pm 1$ then $A^* = A$ and $A^2 = I$.

$\endgroup$
1
  • $\begingroup$ Thank you, this makes much more sense! $\endgroup$ Nov 12 '21 at 17:28
1
$\begingroup$

Since $A^\ast=A^{-3}$, we see that $A^\ast$ commutes with $A$. Hence $A^2=(A^\ast A)^{-1}$ is self-adjoint and positive definite. It follows that $A^2=(A^2)^\ast=(A^\ast)^2=(A^{-3})^2$, meaning that $A^8=I$ or $$ (A^2-I)(A^2+I)(A^4+I)=0.\tag{1} $$ Since $A^2$ is positive definite, so are $A^2+I$ and $A^4+I$. Therefore $(1)$ implies that $A^2-I=0$. But then we also have $AA^\ast-I=0$ because $A^\ast=A^{-3}=(A^2)^{-1}A^{-1}=A^{-1}$. Hence $A=A^\ast$.

$\endgroup$

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.