Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

A is a Hermitian $n\times n$ matrix over ${\Bbb C}$ . $$ A^m = I $$ for $m$ a natural number.

  1. Prove $$A^2=I $$

  2. Suppose $m$ is odd. Prove $$ A=I$$

Well, for the first question I did this: Since $A$ is Hermitian then $A$ is diagonalizable. $A$'s eigenvalues are $\{-1,1\}$. The minimal polynomial that reset $A$ and $$M(x) = (x+1)^z (x-1)^k $$ for $k,z$ natural numbers. So the options for the minimal polynomial are:

  1. $$(x-1)$$
  2. $$(x+1)$$
  3. $$(x-1)(x+1)$$
  4. $$(x-1)^2 $$

Eventually I got $A$ can be: $$ A=I \quad \mathrm{or}\quad A=-I $$

Is that even right?????????

For the second part of the question I have no clue.

share|cite|improve this question
What does hermetic mean? – Will Jagy Jun 3 '12 at 1:33
yes. sorry edited my question – Mary Jun 3 '12 at 1:37
Also one question mark is surely enough. – Peter Sheldrick Jun 3 '12 at 1:38
yes typing mistake – Mary Jun 3 '12 at 1:45
up vote 6 down vote accepted

Your approach is right but in the middle it is not clear what you are doing. First since $A$ is hermitian all the eigenvalues of $A$ are real. Further, $A^m = I$ implies that any eigenvalue $\lambda$ satisfies the relation $\lambda^m - 1 =0$. Since $\lambda \in \mathbb{R}$, if $m$ is even, we get that $\lambda = \pm 1$. If $m$ is odd, we get that $\lambda = 1$.

Also, any hermitian matrix permits a decomposition of the form $$A = U \Lambda U^{\dagger}$$ where $U$ is a unitary matrix and $\Lambda$ is a diagonal matrix with the eigenvalues of $A$ as its diagonal entries. This is so since it is possible to find an orthonormal basis of $\mathbb{C}^n$ consisting of $n$ eigenvectors of $A$.

If $m$ is odd, we showed that all the eigenvalues have to be $1$. Hence, $\Lambda$, the diagonal matrix with the eigenvalues of $A$ as its diagonal entries, is the identity matrix i.e. $\Lambda = I$. This gives us that $$A = U I U^{\dagger} = I$$

If $m$ is even, we showed that all the eigen values have to be $\pm 1$. Hence, $\Lambda^2 = I$. This gives us that $$A^2 = (U \Lambda U^{\dagger})(U \Lambda U^{\dagger}) = U \Lambda (U^{\dagger} U) \Lambda U^{\dagger} = U \Lambda I \Lambda U^{\dagger} = U \Lambda^2 U^{\dagger} = UIU^{\dagger} = U U^{\dagger} = I$$

share|cite|improve this answer
We didnt learned this form for hermitian matrix and we should answer the question with minimal polynomial...But thank you! – Mary Jun 3 '12 at 2:41
@Marvis What is $\,\Lambda\,$ ? The diagonal form of $\,A\,$? – DonAntonio Jun 3 '12 at 10:52
@DonAntonio Yes. I have added it. – user17762 Jun 3 '12 at 14:12

As $\,A\,$ is diagonalizable the only possible forms for its minimal polynomial are $\,(1)\,$ , if $\,z=0\,$, $\,(2)\,$ , if $\,k=0\,$, or $\,(3)\,$ , if $\, k,z\geq 1\,$ . Case $\,(4)\,$is impossible as a square matrix over some field is diagonalizable iff its minimal pol. is the product of different linear polynomials over that field, and from these possibilities either $$A=I\,\,,\,\,A=-I\,\,,\,\,A^2-I=0$$ from which you get what you wanted.

share|cite|improve this answer
Note that the conditions are not exclusive: either $A=I$ or $A=-I$ implies $A^2-I=0$. But the converse is not true: one can have $A^2-I=0$ but neither $A=I$ nor $A=-I$. – Marc van Leeuwen Jun 3 '12 at 16:58

Since you can compute powers in diagonal form, there is almost nothing to this once you know it is diagonalisable with eigenvalues (diagonal entries) in $\{1,-1\}$: one has $A^2=I$ because $1^2=1=(-1)^2$, and if $m$ is odd then $A^2=1$ and $A^m=I$ easily imply $A=I$. For the latter step you could use the minimal polynomail: by $A^2=I$ it is a divisor of $X^2-1$, and reducing $X^m-1$ modulo $X^2-1$ leaves $X-1$, which annihilates $A$ so $A=I$. Apart from this I don't really see how much the minimal polynomial does for you here. But in any case diagonlisable implies the minimal polynomial has no repeated factors, so its only possible values are $1$, $X-1$, $X+1$ and $X^2-1=(X-1)(X+1)$. And yes, $1$ is a possibility, but only if $n=0$.

But note that it is not true that $A=I$ or $A=-1$; the diagnal form of $A$ could have any mix of entries $1$ and $-1$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.