10
$\begingroup$

I am studying the paper of Baum-Connes-Higson to understand the Connes-Kasparov conjecture. In example 4.23, they discuss the case $G=\mathbb{R}^2$. I have constructed the Dirac operator, but I’m stuck with computing its index. According to the example in the article, we should obtain the Bott element in $K_0(C_r^*G))$, but I cannot see how to get there.

For the computation, we have the following setup (in the notation of [BCH]) $G=\mathbb{R}^2$, $K= \{ e \}$, $\mathfrak{g} = \mathfrak{k} \oplus \mathfrak{p}$ with $\mathfrak{k} =0$ and $\mathfrak{p} = \mathbb{R}^2$.

We take the following representation of the Clifford algebra on the spinor space $\Delta_2 = \mathbb{C}^2$: $$ e_1 \mapsto \begin{pmatrix} 0 & 1 \\ -1 & 0 \end{pmatrix}, \; \; e_2 \mapsto \begin{pmatrix} 0 & i \\ i & 0 \end{pmatrix}. $$

The spinor bundle is the trivial bundle $S = \mathbb{R}^2 \times \mathbb{C}^2$, and the sections are given by: $$ C_c^\infty(\mathbb{R}^2,S) =\left\{ \begin{pmatrix} f \\ g \end{pmatrix}: \mathbb{R}^2 \to \mathbb{C}^2 \right\}. $$ The bundle splits as $S = S^+ \oplus S^-$, where $f: \mathbb{R}^2 \to S^+$ and $g: \mathbb{R}^2 \to S^-$. The Dirac operator $D: \Gamma(S) \to \Gamma(S)$ is given by: $$ D= 2 \begin{pmatrix} 0 & \frac{\partial}{\partial \overline{z}} \\ -\frac{\partial}{\partial z} & 0 \end{pmatrix}, $$ where $z= x + i y$. Now as in [BCH], we Fourier transform, so $C_r^*(\mathbb{R}^2)$ becomes $C_0(\mathbb{R}^2)$, and the Dirac operator becomes: $$ D= i \begin{pmatrix} 0 & z \\ - \overline{z} & 0 \end{pmatrix}. $$ To calculate the index we only need $D^+: C_c^\infty(\mathbb{R}^2,S^+) \to C_c^\infty(\mathbb{R}^2,S^-)$. These modules complete to the Hibert modules $C_0(\mathbb{R}^2)$ over $A= C_0(\mathbb{R}^2)$, so we end up calculating the index of the operator: $$ D^+: C_0(\mathbb{R}^2) \to C_0(\mathbb{R}^2), \; f \mapsto i z \cdot f. $$

According to [BCH], the index of $D^+$ should give the Bott element, so: $$ \text{Index}(D^+) = \text{ker}[ iz ] - \text{ker}[ -i\overline{z}] = \left[ \frac{1}{1+z^2} \begin{pmatrix} 1 & \overline{z} \\ z & |z|^2 \end{pmatrix} \right] - \left[ \begin{pmatrix} 0 & 0 \\ 0 & 1 \end{pmatrix}\right]. $$

I first tried to compute it by explicitly finding a $T$ such that $1-TD^+$ and $1-D^+T$ are compact, but I couldn't find such $T$. Then I tried to look at the graph projector $G(tD)$, see (def 4.1, Aastrup) and use theorem 4.15. This seems to give the right result, but we need here that 0 is an isolated point in the spectrum of $D^+$, which is not true in our case.

Is there a way to fix this, or another way to calculate the index?

$\endgroup$

1 Answer 1

4
$\begingroup$

Solved it! (Edit: typo in formula for $\tilde{S}$ and $1-\tilde{T}\tilde{S}$, added i's and signs to obtain exactly Bott element)

Denote by $E=C_0(\mathbb{R}^2)$ our Hilbert module over $A=C_0(\mathbb{R}^2)$. First, we should make $D$ into a bounded operator, we take:

$$ T=\frac{iz}{\sqrt{1+|z|^2}}. $$

Also define: $$ S = \frac{-i\overline{z}}{1+|z|^2}, $$ then $1-TS=1-ST= \frac{1}{1+|z|^2}$, which is a compact operator on $E$. Hence $T$ is $A$-Fredholm, according to (def 3.1, Exel).

Now as $T$ is non-regular, we apply the procedure of (lemma 3.8, Exel) and define: $$ \tilde{T} = \frac{1}{\sqrt{1+|z|^2}}\begin{pmatrix} i z & 0 \\ - i & 0 \end{pmatrix}, \tilde{S} = \frac{1}{\sqrt{1+|z|^2}}\begin{pmatrix} -i \overline{z} & i \\ 0 & 0 \end{pmatrix}. $$

Then the index of $D$ is then given by (see definition 3.10): $$ [\ker \tilde{T} ] - [ \ker \tilde{T^*}] = [1-\tilde{S}\tilde{T}] - [ 1 - \tilde{T} \tilde{S} ] \in K_0(A). $$

We have: $$ 1-\tilde{T} \tilde {S} = 1 - \frac{1}{1+|z|^2} \begin{pmatrix} i z & 0 \\ -i & 0 \end{pmatrix} \begin{pmatrix} -i\overline{z} & i \\ 0 & 0 \end{pmatrix} = \frac{1}{1+|z|^2} \begin{pmatrix} 1 & z \\ \overline{z} & |z|^2 \end{pmatrix} $$ and $$ 1-\tilde{S}\tilde{T} = \begin{pmatrix} 0 & 0 \\ 0 & 1 \end{pmatrix}, $$

so we get precisely the Bott element.

$\endgroup$

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.