I am currently reading the excelent book by Lawson and Michelsohn on Spin Geometry. In chapter 3 paragraph 1, they define a differntial operator of order $m$ to be a $\mathbb{C}$-linear map of $\mathbb{C}$-vector bundles of ranks $p$ respectively $q$, as $P\colon E\longrightarrow F$, in the local coordinate $(U,x)$ and local trivializations $\Phi_E\colon \left.E\right|_U\longrightarrow U\times\mathbb{C}^p,\Phi_F\colon \left.F\right|_U\longrightarrow U\times\mathbb{C}^q$ by $P=\sum_{\left|\alpha\right|\leq m}A^\alpha(x)\frac{\partial^{\left|\alpha\right|}}{\partial x^\alpha}$.

When applying a change of trivialization, say $g_E\colon U\cap V\longrightarrow \operatorname{GL}_p(\mathbb{C})$ and $g_F\colon U\cap V\longrightarrow \operatorname{GL}_p(\mathbb{C})$ they then claim that in the new trivializations $\Psi_E\colon \left.E\right|_V\longrightarrow V\times\mathbb{C}^p,\Psi_F\colon \left.F\right|_V\longrightarrow V\times\mathbb{C}^q$ the operator $P$ is of the form $P=g_F(\sum_{\left|\alpha\right|\leq m}A^\alpha(x)\frac{\partial^{\left|\alpha\right|}}{\partial x^\alpha})g_E^{-1}=\sum_{\left|\alpha\right|\leq m}\tilde A^\alpha(x)\frac{\partial^{\left|\alpha\right|}}{\partial x^\alpha}$ where $\tilde A^\alpha=g_F A^\alpha g_E^{-1}$ whenever $\left|\alpha\right|=m$. I have tried to do this computation for some time now but I cannot seem to do it. If $\sigma\in\Gamma(E)$ is given under the identification $\Gamma(\left.E\right|_U)\leftrightarrow (C^\infty(M,\mathbb{C}))^p$ by $\sigma_\Phi=(\sigma_\Phi^1,\dots,\sigma_\Phi^p)$ in the trivialization $\Phi_E$ and by $\sigma_\Psi=(\sigma_\Psi^1,\dots,\sigma_\Psi^p)$ in $\Psi_E$, then $\sigma_\psi^j=\sum_{i=1}^p (g_E^{-1})_{ij}\sigma_\Phi^i$. So by the generalised Leibniz' rule we find $$ \frac{\partial^{\left|\alpha\right|}}{\partial x^\alpha}\sum_{i=1}^p (g_E^{-1})_{ij}\sigma_\phi^ i=\sum_{\nu\leq\alpha}\begin{pmatrix}\alpha \\ \nu\end{pmatrix}\sum_{i=1}^p\frac{\partial^{\left|\nu\right|}}{\partial x^\nu}(g_E^{-1})_{ij}\frac{\partial^{\left|\alpha-\nu\right|}}{\partial x^{\alpha-\nu}}\sigma_\Phi^i, $$ but I don't think that helps much. If anyone knows how to do this computation, I would be eager to know!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.