1
$\begingroup$

It is well-known that the Toeplitz algebra $\mathcal{T}$ (that I view as concrete subalgebra of $\mathbb{B}(\ell^2(\mathbb{N})$) is the universal algebra generated by an isometry, that is, for any $C^*$-algebra $A$ and an isometry $v \in A$, there exists a unique algebra $\mathcal{T} \rightarrow A$ such that $S \mapsto v$, where $S$ denotes the shift operator in $\mathcal{T}$.

The proof of this fact that I know is done using the Wold decomposition (explained e.g. in Wegge-Olsen, 3.F). He then writes "there are probably easier ways (of which I'm ignorant) to do this". Is anyone here not ignorant of easier ways? In particular, is there a "more $C^*$-algebraic" proof?

Turning things around, we can just define $\mathcal{T}$ as the universal $C^*$-envelope of the $*$-algebra generated freely by $S$ and $S^*$. This then has the universal property by abstract nonsense, but the proof I know that this algebra is isomorphic to the "concrete one" again uses Wold decomposition.

$\endgroup$

1 Answer 1

2
$\begingroup$

The proof for Cuntz-Toeplitz algebras, i.e. algebras generated by some number of isometries with orthogonal ranges, is pretty $C^{\ast}$-algebraic. Let $s$ be the universal isometry and let $\mathcal{T}_{u}$ be the universal $C^{\ast}$-algebra generated by an isometry. We want to show that the $\ast$-homomorphism $\Phi: \mathcal{T}_{u} \to \mathcal{T}$ defined by $s \mapsto S$ is injective. There are circle actions on both of these actions, given just by multiplication by a scalar. They commute with $\Phi$, so there is an induced map between the fixed-point subalgebras. Moreover, there are faithful conditional expectations onto these fixed-point subalgebras, so if the restriction of $\Phi$ is injective, then $\Phi$ itself is injective as well. The fixed point subalgebras are spanned by the elements $t_n := s^{n} (s^{\ast})^{n}$ ($T_n:= S^{n} (S^{\ast})^{n}$). The span of first $m$ $t_n$'s is actually a finite dimensional subalgebra and just by checking the dimension you may prove that the restriction of $\Phi$ to such a subalgebra is injective, hence isometric. It follows easily that $\Phi$ restricted to the fixed-point subalgebra is isometric.

I don't think this is really an easier way, but it definitely works in more general situations.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.