Example for spectral measure and spectral decomposition

417 Views Asked by At

In Werner's "Funktionalanaylsis" (which is one of the German standard functional analysis books) one finds the following example after the statement and proof of the spectral theorem for bounded self-adjoint operators:

Let $H := L^2([0,1])$, $$ T: H \to H, x(t) \mapsto t \cdot x(t). $$ Then $T$ is self-adjoint and it holds that $\sigma(T) = \sigma_c(T)= [0,1]$. The spectral decomposition looks as follows: For $A \in \Sigma$ (the set of Borel subsets of $\sigma(T)$) set $$ E_A x := \mathbb{1}_{A \cap [0,1]}(x). $$ It is easy to see that $E: \Sigma \to L(H)$, $A \mapsto E_A$ is a spectral measure with compact support.

We can see that $E$ represents $T$: $$ \int \lambda d\langle E_{\lambda} x, y \rangle = \int \lambda x(\lambda) \overline{y(\lambda)} d \lambda = \langle T x, y \rangle, $$ as $$ \langle E_A x, y \rangle = \int_{A \cap [0,1]} x(\lambda) \overline{y(\lambda)} d \lambda = \int_{0}^{1} \mathbb{1}_{A}(\lambda) x(\lambda) \overline{y(\lambda)} d \lambda. $$ Therefore, $d\langle E_{\lambda} x, y \rangle$ has the density $x \overline{y} \in L^1$ with respect to the Lebesgue measure. This shows $T = \int \lambda d E_{\lambda}$.

Questions This example helped me understand spectral measures better and the choice of $E_A$ seems very natural.

Can somebody please provide an example of an operator and its spectral decomposition which also has an uncountable spectrum (so we really need an integral instead of a sum) but a different choice of spectral measure?