MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This post is an appendix of this one.

Let $H$ be an infinite dimensional separable Hilbert space and $B(H)$ the algebra of bounded operators.

Invariant subspace problem: Let $T \in B(H)$. Is there a non-trivial closed $T$-invariant subspace?

Hypothesis : The ISP admits a negative answer, i.e., there are ISP counter-examples.

Definition : A category $\mathcal{S}$ of operator algebras see the ISP if $ \forall T, T' \in B(H)$ with $\mathcal{S}(T) \simeq \mathcal{S}(T')$: $$ T \text{ is an ISP counter-example} \Leftrightarrow T' \text{ is an ISP counter-example } $$

Proposition: The category $W^{*}$ of von Neumann algebras, doesn't see the ISP.
proof: Under the previous hypothesis, let $T \in B(H)$ be an ISP counter-example. Then $T$ is irreducible, i.e., $W^{*}(T) = B(H)$. But there are many irreducible operators checking the ISP, for example, the unilateral shift $S$. So $W^{*}(T) \simeq W^{*}(S)$, $S$ checks the ISP and $T$ not. $\square$

This post asks about an equivalent result for the category of $C^{*}$-algebras :

Is there a proof that the category of $C^{*}$-algebras doesn't see the ISP ?

share|cite|improve this question
Is $\mathcal{S}(T)$ the operator algebra generated by $T$? – Ulrich Pennig Aug 4 '13 at 12:57
Yes, $\mathcal{S}(T)$ is the operator algebra (of category $\mathcal{S}$) generated by $T \in B(H)$. Just a precision, the $C^{∗}$-algebras and von Neumann algebras are here separable (the categories $C^{∗}$ and $W^{∗}$). If we can prove that $C^{∗}(T)$ is a Cuntz algebra (with $T\in B(H)$ an ISP counter-example), the result should follow. – Sébastien Palcoux Aug 4 '13 at 14:00
up vote 14 down vote accepted

C*-algebras don't see the ISP. The operators $T\in B(H)$ and $T\oplus T\in B(H\oplus H)$ generate isomorphic C*-algebras, but the latter clearly has non-trivial invariant subspaces. To have both operators in the same Hilbert space, pick isometries $v_1,v_2\in B(H)$ with orthogonal ranges that add up to $H$. Then $$ T\mapsto v_1Tv_1^*+v_2Tv_2^* $$ is an injective *-endomorphism of $B(H)$ that maps $T$ to an operator with non-trivial invariant subspaces.

share|cite|improve this answer
Thank you Leonel ! This map is continuous for the main topologies of operators algebras (norm-topology, weak-topology, strong-topology...), and this argument runs also without a $\star$-structure (on the algebra), so that it shows that no category of operator algebras see the ISP. Is it right ? – Sébastien Palcoux Sep 26 '13 at 13:09
@SébastienPalcoux: I guess so. Though I'm not sure that the question cannot be made more interesting by tweaking it a little. – Leonel Robert Sep 26 '13 at 20:11
A generic way for tweaking a question, is to improve it by excluding the counter-examples. Here we can improve the definition of "see the ISP" by : $\forall T, T' \in B(H)$ with $\mathcal{S}(T) \simeq \mathcal{S}(T') $ and $T' \ne v_{1} T v^{*}_{1} + v_{2} T v^{*}_{2}$ (with $v_{1}$, $v_{2}$ as in your answer), then : "$T$ is an ISP counter-example" $\Leftrightarrow$ "$T'$ is an ISP counter-example". Is this what you thought? Else what do you suggest ? – Sébastien Palcoux Sep 27 '13 at 8:09

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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