0
$\begingroup$

The following statement is false or true:

If $A \in M(n, \mathbb{C})$ is a matrix with complex entries of order $n$ such that $A^4=I$ then \begin{pmatrix} i & 1\\ 0 & i \end{pmatrix} Can be a Jordan Block of $A$.

I believe this statement is false, but I could not formalize the demonstration.

$\endgroup$
3
$\begingroup$

Observe that

$$A^4=I\implies (A-I)(A+I)(A^2+I)=(A-I)(A+I)(A-iI)(A+iI)$$

Thus, over $\;\Bbb C\;$ , the matrix's minimal polynomial decomposes as a product of different linear factors and is thus diagonalizable, which means it cannot have a Jordan Block as the one you wrote.

$\endgroup$
  • 3
    $\begingroup$ Yes, except you're not saying what you mean. In fact $(A-I)(A+I)(A^2+I)=(A-I)(A+I)(A-iI)(A+iI)$ regardless of whether $A^4=I$; also a polynomial cannot be diagonalizable. $\endgroup$ – David C. Ullrich Jun 23 at 23:08
  • $\begingroup$ I had thought of that, but I thought more arguments were needed. Thank you @DonAntonio $\endgroup$ – Mrcrg Jun 24 at 0:33
  • $\begingroup$ @DavidC.Ullrich 1) That decomposition you mention is due precisely to the fact that $\;A^4=I\;$ . It is not independent of it but resulting from it, (2) Yes, my grammar made it sound like the polynomial is the one which is (or is not) diagonalizable, but making a little effort, like the OP did, I think it can be understood correctly...in particular because immediately before that I wrote " the matrix's polynomial..." . The OP had to round up and finish the argument: a square matrix is diagonalizable iff its minimal polynomial can be written as a product of different linear factors. $\endgroup$ – DonAntonio Jun 24 at 7:17
  • 1
    $\begingroup$ "That decomposition you mention is due precisely to the fact that $A^4=I$. It is not independent of it but resulting from it," What??? Can you give an example of $A$ where $(A-I)(A+I)(A^2+I)\ne(A-I)(A+I)(A-iI)(A+iI)$? No. What's due to $A^4=I$ is the fact that $(A-I)(A+I)(A-iI)(A+iI)=0$. Which of course is what matters, but which you didn't mention explicitly. $\endgroup$ – David C. Ullrich Jun 24 at 13:17
  • $\begingroup$ " I think it can be understood correctly": Yes of course. I didn't say it couldn't be understood correctly, I said that you didn't say what you meant. $\endgroup$ – David C. Ullrich Jun 24 at 13:22
1
$\begingroup$

An alternative approach: we have $B = iI + N$, where $I$ denotes the identity matrix and we have $$ N = \pmatrix{0&1\\0&0}. $$ Notably, $iI$ and $N$ commute (so the binomial theorem applies) and $N^2 = 0$. Thus, we have $$ B^4 = (iI + N)^4 = (iI)^4 + 4(iI)^3N + 6(0) + 4(0) + 0 = I -4i N\\ = \pmatrix{1&-4i\\0&1}. $$ Thus, if $B$ is a Jordan block of the Jordan form $J$ of $A$, then $J^4$ will have $B^4$ as a diagonal submatrix. So, we cannot have $J^4 = I$. It follows that $A^4 \neq I$.

$\endgroup$

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.