[Math] How to show that $\|T\|^2=\|T^*T\|$ for a bounded linear operator $T$

adjoint-operatorsfunctional-analysishilbert-spacesnormed-spacesreal-analysis

I need to show that for a bounded linear operator, $T$, on a Hilbert space:
\begin{align*}
\|T\|^2=\|T^*T\|
\end{align*}
All I have so far:

\begin{align*}
\|T^*T\|&=\sup\{|\langle T^*Tf,g \rangle |:\|f\|\le1,\|g\|\le1\}
\\ &= \sup\{|\langle Tf,Tg \rangle|:\|f\|\le1,\|g\|\le1\}
\\ &\le \sup_{\|f\|\le1}\|Tf\|\sup_{\|g\|\le1}\|Tg\|
\end{align*}

Not too sure what to do from here (or even if this is right so far)

Any help is appreciated.

Best Answer

$ \newcommand{\norm}[1]{\left\|{#1}\right\|} \newcommand{\ip}[1]{\left\langle{#1}\right\rangle} $You might find it easier to use an alternative but equivalent definition of the operator norm: if $S \in B(H)$, then $$ \norm{S} = \sup_{\norm{x}=1} \norm{Sx}. $$

Now:

  1. On the one hand, $$ \norm{T} = \sup_{\norm{x}=1} \norm{Tx} = \sup_{\norm{x}=1} \sqrt{\ip{Tx,Tx}}. $$ Since $\ip{Tx,Tx} = \ip{x, T^\ast T x} \leq \norm{T^\ast T}\norm{x}$, what can you conclude?

  2. On the other hand, $$ \norm{T^\ast T} = \sup_{\norm{x}=1} \norm{T^\ast Tx}. $$ Since $\norm{T^\ast T x} \leq \norm{T^\ast} \norm{Tx}$, where $\norm{T^\ast}=\norm{T}$ (why?), what can you conclude?

In terms of wider context, the identity $\norm{T^\ast T} = \norm{T}^2$ is called the $C^\ast$-identity, and is the key fact that makes the theory of $C^\ast$-algebras (e.g., $B(H)$ for $H$ a Hilbert space) so much easier (for lack of a better word) than the theory of more general Banach ($\ast$-)algebras, just as the theory of Hilbert spaces is so much easier than the theory of more general Banach spaces.

Related Question