A question about Schrödinger semigroups

61 Views Asked by At

The question is about Theorem 14.1 in Teschl's Partial Differential Equations which goes like that

$\textbf{Theorem 14.1.}$ The family $T_S(t)$ is a $C_0$-group in $H^{r}(\mathbb{R}^{n})$ whose generator is $i∆$, $D(i∆) = H^{r+2}(\mathbb{R}^{n})$.

Earlier on, the family $T_S(t)$ is defined as $\mathcal{F}^{-1}e^{-i\lvert p \rvert^{2}t}\mathcal{F}$ which is just a direct consequence of computing the solutions for the linear Schrödinger equation by using the Fourier transform, i.e. our solutions look like $u(x,t) = T_S(t)u_0(x)$.

Now, the motivation to get to our generator $A = i∆$ is that we can depict the solutions of the Schrödinger equation as $u(x,t) = u_0(x)e^{it∆}$, like in the link below:

Schrödinger operator: where is the generator to be defined?

However, I do not really understand how we do come up with this solution. Especially the term $e^{it∆}$ doesn't really make sense to me. Could anybody explain, how to read that expression and how we do come up with this generator for our problem?

Furthermore, I do not understand why $D(A) = H^{r+2}(\mathbb{R}^n)$ should hold (so the domain of the generator is the Sobolev space with coefficient $r+2$). Can anybody explain me that, too?

1

There are 1 best solutions below

0
On BEST ANSWER

Time evolution is an exponential operation in the time variable. It's an odd fact of nature that needs some explanation. In order to explain, suppose you have a system whose state is described as a vector $x$ in some Hilbert space $\mathscr{H}$ at a particular time. An evolution operator $E(t_2,t_1)$ acts on a known state vector $x$ at time $t_1$ and "evolves" it to the new state vector $y=E(t_2,t_1)x$ at time $t_2$. By definition $E(t_3,t_2)E(t_2,t_1)=E(t_3,t_1)$, which is an exponential type of property. For time invariant systems, the evolution depends only on the differences in time. More explicitly, $E(t_2,t_1)=\scr{E}(t_2-t_1)$. $$ \scr{E}(t_3-t_2)\scr{E}(t_2-t_1)=\scr{E}(t_3-t_1). $$ So, a time evolution operator on a time-invariant system has a true exponential property: $$ \mathscr{E}(0)=I \\ \mathscr{E}(t)\mathscr{E}(t')=\mathscr{E}(t+t'),\;\; t,t' \ge0. $$ Stability is assumed, meaning that the following holds for all state vectors $x$: $\lim_{h\downarrow 0}E(t+h,t)x=x$. It is this assumption that allows you to talk about a generator.