next up previous contents
Next: Logarithmic divergence of spectra Up: Spectral Growth and Dixmier Previous: Spectral Growth and Dixmier   Contents

Definition of spectral triples

We start with the definition of the main concept in noncommutative geometry.


\begin{defn}
A (unital) {\bf spectral triple} $(\sA, \sH, D)$ consists of:
\beg...
...(D - \la)^{-1}$ is compact, when
$\la \notin \spec(D)$.
\end{itemize}\end{defn}

For now, and until further notice, all spectral triples will be defined over unital algebras. The compact-resolvent condition must be modified if $\sA$ is nonunital: as well as enlarging $\sA$ to a unital algebra, we require only that the products $a(D - \la)^{-1}$, for $a \in \sA$ and $\la \notin \spec(D)$, be compact operators.


\begin{example}
Let $(M,\eps)$ be an oriented compact boundaryless manifold whi...
...er to these
spectral triples as \lq\lq standard commutative examples''.
\end{example}

Note that, if $\la,\mu \notin \spec(D)$, then $(D - \la)^{-1} - (D - \mu)^{-1} = (\la - \mu)(D - \la)^{-1}(D - \mu)^{-1}$ --this is the famous ``resolvent equation''-- since

\begin{displaymath}
(D - \la) \bigl((D - \la)^{-1} - (D - \mu)^{-1}\bigr) (D - \mu)
= (D - \mu) - (D - \la) = \la - \mu.
\end{displaymath}

Thus $(D - \la)^{-1}$ is compact if and only if $(D - \mu)^{-1}$ is compact, so we need only to check this condition for one value of $\la$. In the same way, we get the following useful result.


\begin{lem}
$D$ has compact resolvent if and only if $(D^2 + 1)^{-1}$ is compact.
\end{lem}


\begin{proof}
We may take $\la = -i$, since the selfadjointness of $D$ implies ...
... bounded operator $T$ is
compact if and only if $T^*T$ is compact.
\end{proof}

By the spectral theorem $(D^2 + 1)^{1/2}-\vert D\vert = f(D)$, where $f\: \bR \to \bR$ is the continuous function $f(\la) := \sqrt{\la^2+1} - \vert\la\vert = \frac{1}{\sqrt{\la^2+1} + \vert\la\vert}$; and $0 < f(\la) \leq 1$ for all $\la \in \bR$, so that $\Vert f(D)\Vert \leq 1$. Or more precisely: the operator $f(D) := (D^2 + 1)^{1/2} - \vert D\vert$, defined initially on $\Dom D$, extends to a bounded operator on $\sH$, of norm at most 1.

In many arguments to come, we shall employ $\vert D\vert$ and $\vert D\vert^{-1}$ as if we knew that $\ker D = \{0\}$. However, even if $\ker D \neq \{0\}$, we can always replace $\vert D\vert$ by $(D^2 + 1)^{1/2}$ and $\vert D\vert^{-1}$ by $(D^2 + 1)^{-1/2}$, at the cost of some extra calculation.


next up previous contents
Next: Logarithmic divergence of spectra Up: Spectral Growth and Dixmier Previous: Spectral Growth and Dixmier   Contents
Pawel Witkowski 2006-03-14