How do normal coordinates take effect in the proof of evolution equations?

70 Views Asked by At

Though the title includes "(geometric) evolution equations", my question is really more of how to use normal coordinates to help us prove an identity involving components of a tensor. My sole experience with these coordinates is the proof of the second Bianchi identity $$\nabla\mathrm{Rm}(X,Y,Z,V,W)+\nabla\mathrm{Rm}(X,Y,V,W,Z)+\nabla\mathrm{Rm}(X,Y,W,Z,V)=0.$$ To prove this identity, we show the LHS vanishes at an arbitrary point $p$. The justification can be tedious, but if we choose normal coordinates centered at $p$, then large amounts of effort can be dispensed. The whole process is fine with me, but some other techniques seem to be needed if I want to prove instead that $$\frac{\partial}{\partial t}g_{ij}=2Hh_{ij},\tag{1}$$ which describes how the induced metric $g:=F^*k$ on the immersion $F:M\to(N,k)$ evolves under the mean curvature flow $$\partial_t F_t=H\nu.$$ Those $h_{ij}$ are the components of the second fundamental form. Now fix $p\in M$. If I succeed in justifying (1) at $p$ with normal coordinates $(x^i)$ centered at $p$, then can I conclude that $$\frac{\partial}{\partial t}\widetilde{g}_{\alpha\beta}=2H\widetilde{h}_{\alpha\beta}\tag{2}$$ for any (NOT necessarily normal) coordinates $(\widetilde{x}^\alpha)$ around $p$?

This question may sound stupid, but I need someone experienced to check if my idea works:

At $p$, the rule for a change of coordinates tells us $$\widetilde{g}_{\alpha\beta}=g_{ij}\frac{\partial x^i}{\partial\widetilde{x}^\alpha}\frac{\partial x^j}{\partial\widetilde{x}^\beta}.$$ Now, at $p$, $$\begin{align} \frac{\partial}{\partial t}\widetilde{g}_{\alpha\beta}&=\frac{\partial x^i}{\partial\widetilde{x}^\alpha}\frac{\partial x^j}{\partial\widetilde{x}^\beta}\frac{\partial}{\partial t}g_{ij}\\ &=\frac{\partial x^i}{\partial\widetilde{x}^\alpha}\frac{\partial x^j}{\partial\widetilde{x}^\beta}(2Hh_{ij})\\ &=2H\left(\frac{\partial x^i}{\partial\widetilde{x}^\alpha}\frac{\partial x^j}{\partial\widetilde{x}^\beta}h_{ij}\right)\\ &=2H\widetilde{h}_{\alpha\beta}. \end{align}$$ Did I get it? Thank you.