2
$\begingroup$

Suppose $X$ and $Y$ are compact metric spaces. Let $\varphi\colon C(X)\to M_{n}(C(Y))$ be any $*$-homomorphism. If $\pi$ is an irreducible representation of $M_{n}(C(Y))$, then $\pi$ is unitarily equivalent to a point evaluation $\textrm{ev}_{y}$. The $*$-homomorphism $\textrm{ev}_{y}\circ\varphi\colon C(X)\to M_{n}(\mathbb{C})$ is a representation of $C(X)$. Since it's a finite-dimensional representation, we can find a unitary $u_{y}\in M_{n}(\mathbb{C})$ and a set of points $X_{y}=\{x^{y}_{1},\ldots,x^{y}_{n}\}\subset X$ such that for all $f\in C(X)$, $$ (\varphi\circ f)(y)=(\textrm{ev}_{y}\circ\varphi)(f)=u_{y} \begin{pmatrix} f(x^{y}_{1}) & 0 & \cdots & 0\\ 0 & f(x^{y}_{2}) & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ 0 & 0 & \cdots & f(x^{y}_{n}) \end{pmatrix} u_{y}^{*}. $$

My question is:

Does the Hausdorff distance of the set $\{x_{1}^{y},\ldots,x_{n}^{y}\}$ vary continuously in $y$? More precisely, if $y_{m}\to y$ in $Y$, does $D_{m}:=\max_{1\leq i\leq n}\min_{1\leq j\leq n}d_{X}(x^{y}_{i},x^{y_{m}}_{j})\to 0$ as $m\to \infty$?

This question is motivated by the simpler case where $\varphi$ gives rise to continuous functions $\lambda_{1},\ldots,\lambda_{n}:Y\to X$ satisfying $$ \varphi(f)=\begin{pmatrix} f\circ\lambda_{1} & 0 & \cdots & 0\\ 0 & f\circ\lambda_{2} & \cdots & 0\\ \vdots & \vdots & \ddots & \vdots\\ 0 & 0 & \cdots & f\circ\lambda_{n} \end{pmatrix}. $$

In this case for each $y\in Y$, $X_{y}=\{\lambda_{1}(y),\ldots,\lambda_{n}(y)\}$, and the result holds.

I asked this question some weeks ago on stack exchange, but did not get an answer: https://math.stackexchange.com/questions/2787579/closeness-of-points-in-the-irreducible-decomposition-of-a-c-algebra-repres

$\endgroup$
2
$\begingroup$

Yes, it varies continuously. To see this, suppose it did not, i.e., ${\rm max}_i {\rm min}_j d(x^y_i, x^{y_m}_j) \not\to 0$. Passing to a subsequence, we can assume that ${\rm min}_j d(x^y_i, x^{y_m}_j)$ is bounded away from $0$ for some fixed value of $i$. Now find $f \in C(X)$ such that $f(x_i) = 1$ and $f(x^{y_m}_j) = 0$ for all $m$ and $j$. Then ${\rm ev}_{y_m}\circ \phi(f) = 0$ for all $m$ but ${\rm ev}_y\circ \phi(f) \neq 0$. But $\phi(f) \in M_n(C(Y)) \cong C(Y; M_n)$ so that $y \mapsto {\rm ev}_y(\phi(f))$ must be a continuous map from $Y$ into $M_n$, which is a contradiction.

$\endgroup$

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.