Prob. 6, Sec. 3.10 in Kreyszig's functional analysis book: Powers of self-adjoint operators

492 Views Asked by At

Let $H$ be a Hilbert space. If $T \colon H \to H$ is a bounded self-adjoint linear operator and $T \neq 0$, then $T^n \neq 0$ for all $n \in \mathbb{N}$.

How to show this?

I've managed to show this for $n = 2, 4, 8, 16, 32, \ldots$ as follows:

Suppose that $T^2 = 0$. Let $x \in H$. Then $T^2 x = 0$. Using the definition of the Hilbert adjoint operator $T^*$ of $T$, we obtain $$\Vert Tx \Vert^2 = \langle Tx, Tx \rangle = \langle T^* Tx, x \rangle = \langle T^2 x, x \rangle = \langle 0, x \rangle = 0,$$ showing that $Tx = 0$ for all $x \in H$, a contradiction. So $T^2 \neq 0$.

Thus, we have shown that if $T^* = T$ and $T \neq 0$, then $T^2 \neq 0$ either.

Since $(T^2)^* = T^2$ also and since $T^2 \neq 0$, we must have $(T^2)^2 = T^4 \neq 0$ either.

Now let's suppose that, for some $k \in \mathbb{N}$, we have $$T^{2^k} \neq 0.$$ Since $$\left(T^{2^k}\right)^* = T^{2^k},$$ we must have $$\left(T^{2^k}\right)^2 = T^{2^{k+1}} \neq 0 \ \mbox{ either}. $$ Hence $$T^{2^k} \neq 0 \ \mbox{ for all } \ k \in \mathbb{N}.$$

Is the above proof correct?

How to proceed from here toward proving this assertion for all $n \in \mathbb{N}$?

1

There are 1 best solutions below

0
On

Your proof looks fine.

For the general case, if $T^n=0$, then $T^m=0$ for all $m \geq n$, contradicting what you have shown.