Here is a quotation of a book:

Let $\phi$ and $\psi$ be the faithful states on $A$ and $B$ respectively, and let $||.||_{\alpha}$ be any C*-norm on $A\odot B$ (algebraic tensor product). As we know, we can extend $\phi\odot \psi$ (on $A\odot B$) to a state $\phi \otimes_{\alpha} B$ on $A\otimes_{\alpha} B$. By uniqueness of GNS representations it follows that the representations $\pi_{\phi\otimes_{\alpha} \psi}|_{A\odot B}$ and $\pi_{\phi} \odot \pi_{\psi}$ are unitarily equivalent. (Here, the $\pi_{\phi\otimes_{\alpha} \psi}$, $\pi_{\phi}$ and $\pi_{\psi}$ denote the GNS representations of $\phi\otimes_{\alpha} \psi$, $\phi$ and $\psi$ respectively.)

I do not know how to verify the unitarily equivalent here. Could someone give me some hints?


You need to convince yourself that $\pi_\phi\otimes\pi_\psi$ is a GNS representation for $\phi\otimes_\alpha\psi$.

You have \begin{align*} \langle \pi_\phi\otimes\pi_\psi(\sum a_j\otimes b_j)\,v_\phi\otimes v_\psi,v_\phi\otimes v_\psi\rangle &=\sum_j \langle \pi_\phi(a_j)v_\phi,v_\phi\rangle\,\langle\pi_\psi(b_j)v_\psi,v_\psi\rangle\\ &=\sum_j\phi(a_j)\psi(b_j)\\ &=\phi\otimes\psi(\sum_ja_j\otimes b_j). \end{align*} And $$ \pi_\phi\otimes\pi_\psi(\sum_ja_j\otimes b_j)(v_\psi\otimes v_\psi)=\sum \pi_\phi(a_j)v_\phi\otimes \pi_\psi(b_j)v_\psi $$ shows that $H=\overline{\pi_\phi\otimes\pi_\psi(A\odot B)v_\phi\otimes v_\psi}=H_\phi\otimes H_\psi$. Then the uniqueness of the GNS representation (a particular case of the uniqueness in Stinespring) shows that existence of your unitary.

  • $\begingroup$ Oh, yes. In your answer, does the $\sum a_{j}v_{\phi}\otimes b_{j}v_{\psi}$ means $\sum \pi_{\phi}(a_{j})v_{\phi}\otimes \pi_{\psi}(b_{j})v_{\psi}$? $\endgroup$ – Yan kai May 26 '14 at 1:59
  • $\begingroup$ Yes. I have edited that part. $\endgroup$ – Martin Argerami May 26 '14 at 2:01

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.