3
$\begingroup$

Assume $S$ and $T$ are unital, separable operator systems, and assume further that we have an affine homeomorphism between the state space of $S^*$ and the state space of $T^*$. Does that mean that $S^*$ and $T^*$ are isometrically isomorphic?

I know that each element $\varphi\in S^*$ can be written uniquely as $\varphi_1+i\varphi_2$ where $\varphi_k$ are self adjoints. Now each $\varphi_k$ can be decomposed uniquely as $(\varphi_k)_+-(\varphi_k)_-$ such that $\|\varphi_k\|=\|(\varphi_k)_+\|+\|(\varphi_k)_-\|$ for $(\varphi_k)_{\pm}\geq 0$, and eventually we can present each $\varphi$ as a linear combination of elements in $S^*$ (same goes for $T^*$), and assuming the norm condition, even uniquely. However, I don't see a way to extend this nicely to an isometric isomorphism. Is this even true? any help would be appreciated.

$\endgroup$
9
  • 1
    $\begingroup$ Two comments. First, your condition does not imply that $S$ and $T$ are completely isometrically isomorphic. That fails for any C*-algebra that is not isomorphic to its opposite algebra. (The two will be isometric and have isometrically isomorphic duals, but not completely isometric.) $\endgroup$ Commented Nov 19 at 17:43
  • $\begingroup$ Second, the affine homeomorphism between the two state spaces should extend to an algebraic isomorphism between $S^*$ and $T^*$, and since the real part of the unit ball is $\{\phi - \psi\ : \phi$ and $\psi$ are states$\}$, this isomorphism is an isometry between the real parts of $S^*$ and $T^*$. But does that determine the norms on all of $S^*$ and $T^*$? I don't know. $\endgroup$ Commented Nov 19 at 17:45
  • $\begingroup$ @NikWeaver Isn't it true that if we let $\varphi=\varphi_1+i\varphi_2$ be the decomposition to real and imaginary part, then $\|\varphi_1\|=\|\varphi\|$ (assuming it is non-zero)? $\endgroup$ Commented Nov 19 at 19:12
  • 2
    $\begingroup$ There is a well-known example of an order isomorphic but not isometric operator systems: $\{ a, bz, c\bar{z} : a,b,c \in \mathbb{C}\}\subset C(\mathbb{T})$ and $\left[\begin{smallmatrix} a & 2b \\ 2c & a \end{smallmatrix}\right]\subset M_2(\mathbb{C})$. In both cases, the element is positive iff $c=\bar{b}$ and $a\geq 2|b|$. I cannot remember (an easy argument) why they are not isometric. $\endgroup$ Commented Nov 19 at 23:41
  • 1
    $\begingroup$ $\ell_\infty^2$ embeds isometrically into the second (spanned by $b$ and $c$), but not into the first. $\endgroup$ Commented Nov 21 at 1:53

0

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.