2
$\begingroup$

Let $A$ be a $C^{*}$-algebra, represented on a Hilbert space $H$, and $D$ a selfadjoint unbounded operator on $H$ (note that we do not impose that $D$ have compact resolvent). Let

$\mathcal{A}:=${$a\in A : [D,a]\in B(H) $}

and topologize by the spectral invariant norm $\|a\|_{1}:=\|a\|+\|[D,a] \|$. Let $\mathcal{I}$ be a two sided ideal in $\mathcal{A}$, and denote by $I$ its closure in $A$. Let $0\leq h\leq k\in\mathcal{A}$ be such that $h\mathcal{I}$ is dense in $\mathcal{I}$ (in its Banach algebra topology).

Question: Is $k\mathcal{I}$ dense in $\mathcal{I}$?

Note that the hypotheses imply that $h$ and hence $k$ are strictly positive $\mod I$ in $A$. Therefore both $hI$ and $kI$ are dense in $I$. Also, in case $h$ and $k$ commute, or when $h,k\in\mathcal{I}$, the answer to this question is yes.

$\endgroup$
3
  • $\begingroup$ This question keeps getting bumped. Do you still have a counter-example, as claimed in your comment below $\endgroup$
    – Yemon Choi
    Aug 3, 2012 at 20:39
  • $\begingroup$ Why does this question keep getting bumped? $\endgroup$
    – Jon Bannon
    Sep 18, 2012 at 12:31
  • $\begingroup$ I think it's because no answer has received an upvote. I am going to try a downvote since the OP appears to have swanned off $\endgroup$
    – Yemon Choi
    Sep 18, 2012 at 16:34

1 Answer 1

-1
$\begingroup$

${\cal I}$ is dense in $I$, since $I$ is the closure of ${\cal I}$.

Therefore $k{\cal I}$ is dense in $kI$, using continuity of multiplication.

You note $kI$ is dense in $I$.

Since density is transitive, $k{\cal I}$ is dense in $I$. Since ${\cal I} \subseteq I$, $k{\cal I}$ is also dense in ${\cal I}$. In the topology of $A$. Do you require density in the stronger topology of ${\cal A}$?

$\endgroup$
1
  • $\begingroup$ The question is about density in the stronger topology of $\mathcal{A}$. Actually, the statement is false. I have a counterexample but it takes some time to write down properly. I will do that later. $\endgroup$ Dec 9, 2011 at 12:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.