2
$\begingroup$

Given a Hilbert space $\mathcal{H}$.

Consider normal operators: $$N:\mathcal{D}N\subseteq\mathcal{H}\to\mathcal{H}:\quad N^*N=NN^*$$

Regard their algebra: $$\mathcal{A}(N):=\{\eta(N):\eta\in\mathcal{B}(\sigma(N)):\|\eta\|_\infty<\infty\}$$

They are commutative: $$\eta(N)\vartheta(N)=(\eta\vartheta)(N)=(\vartheta\eta)(N)=\vartheta(N)\eta(N)$$

They are uniformly closed: $$M:=\lim_n\eta_n(N)=(\lim_n\eta_n)(N)=:\eta(N)$$

But also weakly or strongly closed?

$\endgroup$
  • 1
    $\begingroup$ What if you look at $(Mf)(x)=xf(x)$ on $L^{2}[0,1]$ and start with Borel functions $F$ and $F(M)=M_{F}$. Do you still end up with Borel functions $F$? $\endgroup$ – Disintegrating By Parts Jul 17 '15 at 17:49
  • $\begingroup$ @TrialAndError: Hehe right just take a nonmeasurable set and all finite subsets of it. :) $\endgroup$ – C-Star-W-Star Jul 17 '15 at 18:12
1
$\begingroup$

In general, they are not weakly closed (not even strongly closed). To see this, consider the Hilbert space $\ell^2([0,1])$ and the operator $N=M_{\rm id}$. Then the (bounded) Borel functional calculus will yield all operators $M_f$ (multiplication with $f$, where $f$ is Borel measurable and bounded.

But now consider the set $A:=\{I\subset [0,1]\,\mid\,I \text{ finite}\}$, ordered by inclusion. This is a directed set. Now let $f:[0,1]\to \Bbb{R}$ be an arbitrary non-measurable (bounded) function. Let $f_I := f \cdot \chi_I$, where $\chi_I$ is the indicator function of $I$.

It is not hard to see (since each "sequence" in $\ell^2([0,1])$ has countable support) that $M_{f_I} x \to M_f x$ for all $x \in \ell^2$, but $M_f \notin A(N)$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Not sure but shouldn't it read $N=M_\mathrm{id}$ instead of $N=\mathrm{id}$? Otherwise it would be simply $\mathcal{A}(N)=\mathbb{C}$? $\endgroup$ – C-Star-W-Star Jul 17 '15 at 18:24
  • $\begingroup$ @Freeze_S: Oh, yes. Good catch! Will edit! Oh, you already did. Thanks! $\endgroup$ – PhoemueX Jul 17 '15 at 20:42

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.