I would like to understand the proof of Lemma 1, page 339 in this book. Very briefly, the context is as follows: we have even dimensional oriented conformal manifold with the Hodge star operator chosen and we would like to construct a Fredholm module based on this data. In order to do so, we need an auxiliary operator, denoted by $S$ which has the property that the graph of $S$ coincides with the image of exterior derivative. There are several points which are not clear for me in the argument:

Q1. Why any differential form in $\mathcal{H}_0^+$ which is orthogonal to harmonic forms may be written as $\frac{1+\gamma}{2} d\alpha$ for some $\alpha$?

Hodge theory ensures us that such $\omega$ lies in the image of $d+d^*$ and the claim should somehow follow from the fact that $d^*=-*d*$ but I don't see how since there is no obvious commutation relation between the grading $\gamma$ and $d$.

Q2. I don't see why $\|\frac{1+\gamma}{2} d \alpha \|_2 =\| \frac{1-\gamma}{2}d\alpha \|_2$.

In fact after simple computation this is equivalent to the fact that $\langle d \alpha, \gamma d\alpha \rangle = -\langle \gamma d \alpha, d \alpha \rangle$. But how to get this minus sign? I thought that $\gamma$ should be self adjoint.

Q3. Even if we have this equality, does it really imply that $d\alpha$ is determined uniquely?

Leaving this specific context for a moment, we can imagine two vectors in $\mathbb{R}^2$ say $\xi_1=(1,1)$ and $\xi_2=(-1,1)$ and a projection $P(x,y):=(0,y)$ Then $P\xi_1=P\xi_2$ and $\| P \xi_1 \|= \| (I-P)\xi_1 \|$ but $(I-P)\xi_1 \neq (I-P)\xi_2$. In our context the role of $P$ is played by $\frac{1-\gamma}{2}$ (which indeed is a projection).

I suspect that this uniqueness is needed in order to define $S$ in such a way that $S (\frac{1-\gamma}{2} d\alpha):=\frac{1+\gamma}{2} d \alpha$. But even so I still don't see why the graph of $S$ should be equal to the image of $d$ (image or closure of image?) since I can't get rid of the harmonic part in the expression $\omega+S \omega$ for $\omega \in \mathcal{H}_0^{-}$ (which is mapped to 0 by $S$).

I realize that these are technical questions and can be asked separetly, nevertheless I think that the whole context may be here important therefore I decded to ask all my questions in one post-I hope that this will be fine.


Here is the answer to the questions. (Recall $V$ is a closed oriented conformal manifold of dimension $2n$). According to the Hodge-deRham decomposition, any $\omega\in L^2(V,\Lambda^nT^*V) $can be written $$\omega=h+d\alpha+d^*\beta$$ where $h\in L^2(V,\Lambda^nT^*V)$ is harmonic, $\alpha\in W^{1,2}(V,\Lambda^{n-1}T^*V)$ and $\beta\in W^{1,2}(V,\Lambda^{n+1}T^*V)$, moreover $\alpha$ (resp. $\beta$) is unique provided it is weakly coexact (resp. weakly exact). In fact, there is a unique $\eta\in W^{2,2}(V,\Lambda^{n}T^*V)$, orthogonal to the space of harmonic form such that: $$\omega=h+(dd^*+d^*d)\eta.$$ Now if $\gamma\omega=\omega$ and $\omega$ is orthogonal to the space of harmonic form then $$\omega=d\alpha+d^*\beta \ \mathrm{and}\ \omega=\gamma d\alpha+\gamma d^*\beta=d^*\tilde\alpha+d\tilde \beta$$ where (for some $\varpi,\varpi'\in \{1,i\}$) $\tilde\alpha=\pm\varpi \gamma\alpha$ and $\tilde\beta=\pm\varpi' \gamma\beta$, by uniqueness we get $$d\alpha=d\tilde \beta=\gamma d^*\beta \ \mathrm{and}\ d^*\beta=\gamma d\alpha=d^*\tilde\alpha$$ so that $$\omega=(1+\gamma)d\alpha=\frac{1+\gamma}{2} d\alpha.$$

For the next question, we have $$\langle d\alpha,\gamma d\alpha\rangle_{L^2} =\pm\varpi \langle d\alpha,\star d\alpha\rangle_{L^2}=\pm\varpi\int_V d\alpha\wedge d\alpha=\pm\varpi\int_V d(\alpha\wedge d\alpha)=0.$$


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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