34.6 The Gelfand-Naimark Theorem

Theorem 34.6.1 (Gelfand-Naimark).label Let $A$ be a commutative unital $C^{*}$-algebra, then the Gelfand transform

\[\Gamma_{A}: A \to C(\Omega(A); \complex) \quad \Gamma_{A}(x)(\phi) = \phi(x)\]

is a unital $C^{*}$-isomorphism.

Proof, [Theorem II.9.4, Zhu93]. By construction $\Gamma_{A}$ is a unital algebra homomorphism.

To see that $\Gamma_{A}$ preserves involutions, let $y \in A$ be self-adjoint. By Proposition 33.8.2 and Proposition 34.3.6, $\Gamma_{A}(y)(\Omega(A)) = \sigma_{A}(y) \subset \real$, so $\Gamma_{A}(y) \in C(\Omega(A); \real)$. For any $x \in A$, write $x = \text{Re}(x) + i\text{Im}(x)$, where $\text{Re}(x)$ and $\text{Im}(x)$ are both self-adjoint, then

\begin{align*}\Gamma_{A}(x^{*})&= \Gamma_{A}(\text{Re}(x) - i\text{Im}(x)) \\&= \Gamma_{A}(\text{Re}(x)) - i\Gamma_{A}(\text{Im}(x)) \\&= \overline{\Gamma_A(\text{Re}(x)) + i\Gamma_A(\text{Im}(x))}= \overline{\Gamma_A(x)}\end{align*}

so $\Gamma_{A}(x^{*}) = \Gamma_{A}(x)^{*}$.

Now, for each $x \in A$, Corollary 34.3.5 and Proposition 33.8.2 imply that

\begin{align*}\norm{x}_{A}^{2}&= \sup\bracs{|\lambda|\ | \lambda \in \sigma_A(x^*x)}\\&= \sup\bracs{|\Gamma_A(x^*x)(\phi)|\ | \phi \in \Omega(A)}\\&= \sup\bracs{|\Gamma_A(x)(\phi)|^2\ | \phi \in \Omega(A)}\\ \norm{x}_{A}&= \norm{\Gamma_A(x)}_{u}\end{align*}

Thus $\Gamma_{A}$ is an isometry, and $\Gamma_{A}(A)$ is a closed subalgebra of $C(\Omega(A))$.

Since $\Gamma_{A}(1_{A}) = 1$, $\Gamma_{A}(A)$ contains constants. As $\Gamma_{A}(A)$ separates points and is closed under complex conjugation, $\Gamma_{A}(A) = C(\Omega(A))$ by the Stone-Weierstrass Theorem.$\square$

Corollary 34.6.2.label Let $A$ be a unital $C^{*}$-algebra, then the following are equivalent:

  1. (1)

    $A$ is commutative.

  2. (2)

    $\Gamma_{A}$ is a *-isomorphism.

  3. (3)

    $\Gamma_{A}$ is injective.

Corollary 34.6.3.label Let $A$ be a commutative unital $C^{*}$-algebra and $x \in A$ be normal, then:

  1. (1)

    $x$ is self-adjoint if and only if $\sigma_{A}(x) \subset \real$.

  2. (2)

    $x$ is unitary if and only if $\sigma_{A}(x) \subset \partial B_{\complex}(0, 1)$.

  3. (3)

    $x$ is positive if and only if $\sigma_{A}(x) \subset [0, \infty)$.

  4. (4)

    $x$ is a projection if and only if $\sigma_{A}(x) \subset \bracs{0,1}$.

Corollary 34.6.4.label Let $A$ be a unital $C^{*}$-algebra, then $A_{sa}$ is order complete if and only if $\Omega(A)$ is extremely disconnected.

Proof. By Theorem 34.6.1, $A$ and $C(\Omega(A); \complex)$ are isomorphic as $C^{*}$-algebras. In particular, $A_{sa}$ and $C(\Omega(A); \real)$ are isomorphic as ordered vector spaces, so $A_{sa}$ is order complete if and only if $C(\Omega(A); \real)$ is order complete. Thus the Stone-Nakano Theorem implies that $A_{sa}$ is order complete if and only if $\Omega(A)$ is extremely disconnected.$\square$

Corollary 34.6.5.label Let $(X, \cm, \mu)$ be a localisable measure space, then $\Omega(L^{\infty}(X))$ is extremely disconnected.

Proof. By Corollary 22.6.8, $L^{\infty}(X; \real)$ is order complete. By Corollary 34.6.4, $\Omega(L^{\infty}(X))$ is extremely disconnected.$\square$

Post a Comment

Name:Email:
Please enter the tag of the current page (152) to post the comment.
Tag: