Let $A$ and $B$ be $C^{\ast}$-algebras with centers $Z$ and $Z'$ respectively. Let $\phi:A \to B$ be surjective $C^{\ast}$-morphism. Then

$\phi(Z)=Z'$ if and only if the map $\Phi: \operatorname{Prim}(Z') \to \operatorname{Prim}(Z)$ defined as $\Phi(J) = \phi^{-1}(J)$ is injective.

Can someone please give me reference for the above result?

The above result is mentioned without proof in the paper titled On the homomorphic image of Center of $C^{\ast}$-algebras by Vesterstrom.

  • $\begingroup$ Which parts of this have you been able to prove yourself? What have you tried so far? $\endgroup$
    – Yemon Choi
    Jul 27, 2020 at 4:40
  • $\begingroup$ Also, something looks wrong with your notation. What is $\tilde{\phi\vert_Z}$ supposed to be? $\endgroup$
    – Yemon Choi
    Jul 27, 2020 at 4:53
  • $\begingroup$ @YemonChoi: I got confused with the proof. Could not get much idea of the proof. I'm trying it. Regarding your second comment: fixed notation $\endgroup$
    – Math Lover
    Jul 27, 2020 at 5:06
  • 2
    $\begingroup$ The linked paper has an important condition which is missing from the question: $\phi$ is supposed to be a surjective $*$-homomorphism! This takes care of Jamie Gabe's objection (without $\phi$ being surjective, indeed $\Phi$ makes no sense). I am following Blackadar's book, II.6.5.4, if $J=\ker\phi$ then $B\cong A/J$ and $\newcommand{\prim}{\operatorname{Prim}}\prim(A/J) \cong \{K\in\prim(A) : J\subseteq K \}$. I think to understand Proposition 1 you'll need to know about the Dauns-Hofman Theorem. $\endgroup$ Jul 27, 2020 at 13:09
  • 1
    $\begingroup$ The restated version of the question finally makes sense (it would have saved everyone time if a link to the relevant paper had been posted in the original question). Without claiming to immediately see a full proof, it should be noted that fo a commutative ${\rm C}^*$-algebra primitive ideals are the same as maximal ideals, and indeed the primitive ideal space corresponds naturally to the Gelfand spectrum. So the map $\Phi$ admits a very concrete description $\endgroup$
    – Yemon Choi
    Jul 27, 2020 at 19:47

1 Answer 1


Basically it is a theorem about commutative unital C$^*$-algebras (Vesterstrom also has a blanket assumption that $A$ and $B$ are unital).

We have a map $\phi: Z\to Z'$. So $\phi(Z)$ is a C$^*$-subalgebra of $Z'$, and $\phi(Z)=Z'$ if and only if $\phi(Z)$ separates the points of ${\rm Prim}(Z')$. For $J\in{\rm Prim}(Z')$, $\Phi(J)=\phi^{-1}(J)=\{ z\in Z: \phi(z)\in J\}$.

Hence for $J_1, J_2\in{\rm Prim}(Z')$, $\Phi(J_1)=\Phi(J_2)$ if and only if for all $z\in Z$, $\phi(z)\in J_1\Leftrightarrow \phi(z)\in J_2$, and this condition holds if and only if $\phi(Z)$ fails to separate $J_1$ and $J_2$. Thus $\phi$ is surjective if and only if $\Phi$ is injective.

  • $\begingroup$ Sorry where did you use the fact that $A$ and $B$ are unital $C^{\ast}-$ algebras? $\endgroup$
    – Math Lover
    Jul 31, 2020 at 18:06
  • $\begingroup$ I am not sure that I did, but Vesterstrom has it, so I felt safer with that hypothesis. In the non-unital case, the Stone-Weierstrass theorem requires "separates the points and vanishes nowhere", if I remember, so perhaps I did use it in line 2 of the second para. $\endgroup$ Jul 31, 2020 at 19:07

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.