2
$\begingroup$

if I have a self-adjoint operator $T:D(T) \rightarrow L^2$, then I define its unitary exponential operator by $$e^{iT}(f) := \lim_{k \rightarrow \infty} e^{iT_{k}}(f),$$

where $T_k(f):=\frac{1}{2} k^2((T+i k)^{-1} + (T-ik)^{-1})(f) \rightarrow T(f),$ for $f \in D(T).$ Notice, that $T_k$ are bounded operators.

Now, assume that our operator has a purely discrete spectrum $\lambda_0,\lambda_1,\lambda_2 ,....$ and $f:=\sum_{n=0}^{\infty} a_n e_n$, where $e_n$ are eigenfunctions to our operator. How do I show that $$e^{iT}(f)= \sum_{n=0}^{\infty}e^{i \lambda_n} a_n e_n?$$

$\endgroup$
1
$\begingroup$

For $\lambda \in \mathbb{R}$ and $k > 0$, $$ \frac{k^{2}}{2}\left[\frac{1}{\lambda+ik}+\frac{1}{\lambda-ik}\right] =\lambda \frac{k^{2}}{\lambda^{2}+k^{2}} $$ So the above function converges to $\lambda$ as $k\rightarrow\infty$. You have $$ e^{iT_{k}}x = \sum_{n}\exp\left\{i\lambda_{n}\frac{k^{2}}{\lambda_{n}^{2}+k^{2}}\right\}a_{n}e_{n}. $$ Therefore, $$ \left\|e^{iT_{k}}x-\sum_{n}e^{i\lambda_{n}}a_{n}e_{n}\right\|^{2} = \sum_{n}\left|e^{i\lambda_{n}}-\exp\left(i\lambda_{n}\frac{k^{2}}{\lambda_{n}^{2}+k^{2}}\right)\right|^{2}|a_{n}|^{2}\|e_{n}\|^{2} $$ The exponential terms on the right are uniformly bounded and converge to $0$ as $k\rightarrow\infty$ for each fixed $n$. And $$ \|x\|^{2}=\sum_{n}|a_{n}|^{2}\|e_{n}\|^{2} < \infty. $$ So you can apply the Lebesgue dominated convergence theorem to see that the right side converges to $0$ as $k\rightarrow\infty$. And that's what you want.

$\endgroup$
  • $\begingroup$ sorry, where does this second line come from? aren't we using there what we want to show (although the difference is, that we are talking about bounded operators there)? - but I like your answer.:-) $\endgroup$ – user168865 Aug 11 '14 at 0:58
  • 1
    $\begingroup$ I took "where en are eigenfunctions to our operator" to mean the eigenfunctions for $T$. Then the second line follows because of how $T_{k}$ is defined in terms of $T$. The first part of the second line was a statement about functions, not about operators. $\endgroup$ – DisintegratingByParts Aug 11 '14 at 2:01
  • 1
    $\begingroup$ I forgot to ping you with @user168865 in the previous comment. $\endgroup$ – DisintegratingByParts Aug 11 '14 at 2:09
  • $\begingroup$ Maybe I am blind, but I still don't see it. $e^{iT_k}(x):=(\sum_{n=0}^{\infty} \frac{T_k^n}{n!})(x)$ and $x = \sum_{n=0}^{\infty}a_n e_n$ and probably your $x$ is my $f$. But still, I just don't see what has happened there, sorry. $\endgroup$ – user168865 Aug 11 '14 at 9:37
  • 1
    $\begingroup$ @user168865 : $T_{k}e_{m} = [\lambda_{m}k^{2}/(\lambda_{m}^{2}+k^{2})]e_{m}$ and $T_{k}^{n}e_{m}=[\lambda_{m}k^{2}/(\lambda_{m}^{2}+k^{2})]^{n}e_{m}$. Yes, I see I switched out $f$ for $x$ in notation, and added confusion. $\endgroup$ – DisintegratingByParts Aug 11 '14 at 11:24

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.