Let $A=\{f:[0,1]\to M_2(\mathbb{C}): $f continuous and $ f(0)=\begin{pmatrix} f_{11}(0) & 0 \\ 0 & f_{22}(0) \end{pmatrix}\}$ be a $C^*$-algebra with pointwise multiplication, involutions and norm $\|f\|=\sup\limits_{t\in [0,1]}\|f(t)\|_{op}$. I want to determine $\hat{A}=\{[\pi]:\pi$ is a irreducible representation of $A$}, (Here is $\pi\sim \rho :\iff$ there is an unitary operator $V:H_{\pi}\to H_{\rho}$ such that $V\pi(a)=\rho(a)V$ for all $a\in A$).

I want to use the following fact: If $I$ is a closed ideal in a $C^*$-algebra $A$, it is $$\hat{A}=\hat{A}_I\coprod \hat{A}_{A/I}=\widehat{I}\coprod \widehat{A/I},$$ where $\hat{A}_I=\{[\pi]\in \hat{A}: \pi(I)\neq 0\}$ and $\hat{A}_{A/I}=\{[\pi]\in \hat{A}: \pi(I)=0\}$.

Consider $I=\{f\in A: f(0)=\begin{pmatrix} 0 & 0 \\ 0 & 0 \end{pmatrix}\}$. $I$ is a closed Ideal in $A$, because $I$ is the kernel of the evaluation-homomorphism $\epsilon_0:A\to M_2(\mathbb{C}), f\mapsto f(0)$. Therefore $I$ is a $C^*$-subalgebra of $A$. Then it follows that $I$ is isomorphic to $C_0((0,1],M_2(\mathbb{C}))$ as $C^*$-algebra. Maybe we get $\widehat{A/I}$ now, because we know what $A/I$ is, it is $A/I=A/ \ker(\epsilon_0)\cong Im(\epsilon_0)\cong \mathbb{C}^2$. But here I'm stuck, I'm not sure what $\widehat{A/I}$ should be and I really don't know what $\hat{I}$ is, I assume $\hat{I}$ is homeomorphic to $(0,1]$, but I can't prove it.

I need that because I want to describe the hull-kernel-topology on $\hat{A}$.

I'm not sure if the question is ok for mathoverflow, I'm sorry if not. But I appreciate your help.

edit: is $\widehat{A/I}$ a discrete set with 2 elements?

  • 3
    $\begingroup$ Indeed, $\widehat{C_0(X,\mathbb M_n)}$ is homeomorphic to $X$. Do you see a natural map? (This gives you both $\widehat{A/I}$ and $\hat{I}$. It is not hard to describe the topology on $\hat A=\hat I\sqcup\widehat{A/I}$.) $\endgroup$ – Rasmus Sep 28 '15 at 7:42
  • $\begingroup$ yes, thank you, I remember! therefore $\hat{I}$ must be homeomorphic to $(0,1]$. I know that $\widehat{C_0(X,M_2)}$ equals the Gelfand space of $C_0(X,M_2)$ and the Gelfand space of $C_0(X,M_2)$ is homeomorphic to $X$, the homeomorphism is the point-evaluation-map. But I still don't know what $\widehat{A/I}$ is. It must be a discrete set with 2 elements, but I don't see why... $\endgroup$ – Sabrina Gemsa Sep 28 '15 at 9:14
  • 2
    $\begingroup$ $\mathbb C^2=C_0(\{a,b\},\mathbb M_1)$. $\endgroup$ – Rasmus Sep 28 '15 at 9:29
  • 2
    $\begingroup$ You're welcome! If everthing's clear now, I encourage you to write a detailed answer! :) $\endgroup$ – Rasmus Sep 28 '15 at 9:35
  • 2
    $\begingroup$ ok, I will do it:) (in a few hours) $\endgroup$ – Sabrina Gemsa Sep 28 '15 at 11:17

Let $X$ be a compact Hausdorff space, $A$ a unital $C^*$-algebra, first you can prove that $$\gamma:X\times \hat{A}\to \widehat{C_(X,A)},\; (x,[\pi])\mapsto [\pi\circ ev_x]$$ is a bijection, where $ev_x: C(X,A)\to A,\; f\mapsto ev_x(f):=f(x)$ is the evaluation map. Then you can deduce the same for $\widehat{C_0(X,A)}$ and $X\times \hat{A}$, if $X$ is local compact, Hausdorff and if $A$ is an arbitrary $C^*$-algebra.

[Note: $\widehat{C_0(X,A)}$ and $\hat{A}$ are homeomorphic if $\widehat{C_0(X,A)}$ and $\hat{A}$ are endowed with the kern-hull-topology and $X\times \hat{A}$ takes the product topology. ]

Therefore it is $\hat{I}=\widehat{C_0( (0,1], M_2(\mathbb{C}) )}=(0,1]\times \widehat{M_2(\mathbb{C}) }$.
Next, note that all irreducible $*$-representations $\pi:M_2(\mathbb{C})\to L(H_{\pi})$ are equivalent to the irreducible $\ast$-representation $id:M_2(\mathbb{C})\to M_2(\mathbb{C})\cong L(\mathbb{C}^2)$, i.e. $ \widehat{M_2(\mathbb{C}) }=\{[id]\}$. So $\hat{I}$ is basically $(0,1]$.

To obtain $\hat{A}$ we also need $\widehat{A/I}$, because as sets we have the equality $$\hat{A}=\widehat{I}\coprod \widehat{A/I}.$$ It is $\widehat{A/I} =\widehat{ \mathbb{C}\oplus \mathbb{C} }=\hat{\mathbb{C}}\coprod \hat{\mathbb{C}}$, a discrete set with 2 elements.

In summary: $\hat{A}$ has the form $\{\pi_t|t\in (0,1]\}\cup \{\chi_1,\chi_2\}$ with $\pi_t(f)=f(t)$ and $\chi_i(f)=f_{ii}(0)$.

Thanks to Rasmus Bentmann for the help.

| cite | improve this answer | |
  • $\begingroup$ You're welcome! Some details concerning the topology are still missing. Can you work them out? $\endgroup$ – Rasmus Sep 30 '15 at 19:08
  • 1
    $\begingroup$ ok, I promise I will add some details the next days =). $\endgroup$ – Sabrina Gemsa Sep 30 '15 at 19:14

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.