2
$\begingroup$

Let $A$ be a $C^*$ algebra. Let $a \in A$, then there exists $c \in A$ such that $a=cc^*c$.

This fact is used from example (1) of Prop 4.25. How does one show this?

$\endgroup$
  • $\begingroup$ The construction of $c$ is sketched in the proof of Prop. 4.25. Where do you have problems? $\endgroup$ – MaoWao Mar 16 at 18:17
  • $\begingroup$ Ah, ok, I did not read the proof, and jumped to the example for some intuition. $\endgroup$ – Bryan Shih Mar 16 at 18:36
1
$\begingroup$

$a^*a \ge 0$ so the function $t \mapsto \frac{t^{1/6}}{\sqrt{t+\frac1n}}$ is continuous on $\sigma(a^*a) \subseteq [0,\infty]$ and satisfies $0\mapsto 0$.

Thus it makes sense to define a sequence $(c_n)_n$ as $$c_n = a\left(a^*a + \frac1n\right)^{-1/2}(a^*a)^{1/6} \in A$$

Denote $d_{mn} = \left(a^*a + \frac1m\right)^{-1/2} - \left(a^*a + \frac1n\right)^{-1/2}$ for $m,n\in\mathbb{N}$. We have \begin{align} \|c_m - c_n\|^2 &= \|a \,d_{mn}(a^*a)^{1/6}\|^2 \\ &= \|(a^*a)^{1/6}d_{mn}(a^*a)d_{mn}(a^*a)^{1/6}\|\\ &= \|{d_{mn}}^2(a^*a)^{4/3}\|\\ &= \|{d_{mn}}(a^*a)^{2/3}\|^2\\ &= \left\|\left(a^*a + \frac1m\right)^{-1/2}(a^*a)^{2/3} - \left(a^*a + \frac1n\right)^{-1/2}(a^*a)^{2/3}\right\|^2\\ &\xrightarrow{m,n\to\infty} \|(a^*a)^{1/6}-(a^*a)^{1/6}\|^2 \\ &= 0 \end{align}

because $$f_n(t) = \frac{t^{2/3}}{\sqrt{t+\frac1n}} \xrightarrow{n\to\infty} t^{1/6}$$ uniformly on $\sigma(a^*a)$.

Hence $(c_n)_n$ converges to an element $c \in A$. We claim that $c$ is the desired element.

\begin{align} cc^*c &= \lim_{n\to\infty} c_nc_n^*c_n \\ &= \lim_{n\to\infty} a\left(a^*a + \frac1n\right)^{-1/2}(a^*a)^{1/6}(a^*a)^{1/6}\left(a^*a + \frac1n\right)^{-1/2}(a^*a)\left(a^*a + \frac1n\right)^{-1/2}(a^*a)^{1/6}\\ &= \lim_{n\to\infty} a\cdot (a^*a)^{3/2}\left(a^*a + \frac1n\right)^{-3/2}\\ &= a \cdot 1\\ &= a \end{align} because $$g_n(t) = \frac{t^{3/2}}{\left(t+\frac1n\right)^{3/2}} \xrightarrow{n\to\infty} 1$$ uniformly on $\sigma(a^*a)$. The limit function $1$ does not satisfy $0 \mapsto 0$ but we can adjoin a unit to the algebra $A$ so that it makes sense.

Therefore $a = cc^*c$.

This proof was mostly from Pedersen's $C^*$-algebras and their Automorphism Groups, Lemma $1.4.4.$ and $1.4.5.$

$\endgroup$

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.