I am struggling to formally derive the expression found in the Stone's theorem for one-parameter unitary groups. I am aware that this can be done by using the spectral theorem. I am mostly interested in the discrete 'version' of the spectral theorem.

Here's a short statement of Stone's theorem:

If ${\cal H}$ is a Hilbert space and $U(t)$ is a strongly-continuous, one-parameter unitary group, then $U(t)=\exp\bigl(-itH\bigr)$, where $H$ is self-adjoint.

Can anyone help me out or point me to some reference where this is explicitly done?

  • $\begingroup$ What do you mean by "discrete version" of the spectral theorem? $\endgroup$ – MaoWao Nov 1 '19 at 10:41
  • 1
    $\begingroup$ By discrete I mean the case in which an operator in question has a purely discrete spectrum $\endgroup$ – omsorg Nov 1 '19 at 11:05

A standard reference is Reed, Simon. Methods of Modern Mathematical Physics I. Stone's theorem is treated in Section VIII.4.

The rough idea is as follows. Define a (possibly unbounded) operator $H$ by $$ D(H)=\left\{\xi\in \mathcal{H}\mid\lim_{t\to 0}\frac{\xi-U(t)\xi}{it}\phantom{x}\text{ exists }\right\},\ \ \ H\xi=\lim_{t\to 0}\frac{\xi-U(t)\xi}{it}. $$

It needs some effort, but one can show that $H$ is self-adjoint. Then one can check that if $\xi\in D(H)$, both $u_1(t)=U(t)\xi$ (by the group property of $U$) and $u_2(t)=\exp(-itH)\xi$ (by the spectral theorem) solve the initial-value problem \begin{align*} \dot u(t)&=-iHu(t),\\ u(0)&=\xi. \end{align*} In particular, $$ \|u_1(t)-u_2(t)\|^2=\int_0^t\frac{d}{ds}\|u_1(s)-u_2(s)\|^2\,ds=-2\int_0^t\operatorname{Re}i\langle u_1(s)-u_2(s),H(u_1(s)-u_2(s)\rangle\,ds=0. $$ Thus $U(t)\xi=\exp(-itH)\xi$. The equality for arbitrary $\xi\in\mathcal{H}$ follows by continuity.

I do not know how purely discrete spectrum for the generator can be detected from the unitary group.

| cite | improve this answer | |
  • $\begingroup$ thanks a lot! I think i can work it out for my specific case using this! $\endgroup$ – omsorg Nov 1 '19 at 11:49
  • $\begingroup$ @omsorg if this answers your question sufficiently, you should accept it to prevent it from being bumped up to the front page unnecessarily in the future. $\endgroup$ – Cameron Williams Nov 1 '19 at 18:21
  • $\begingroup$ thanks for the headsup, just done! $\endgroup$ – omsorg Nov 4 '19 at 8:55

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.