Proof question on lemma included in paper on invariant sets of PDEs

82 Views Asked by At

I'm not exactly sure how to categorize this question so sorry if it's in the wrong place. I'm trying to understand one part of a lemma given in a paper on invariant sets for PDEs and I'm getting hung up on signs. I think it might just be me not understanding notation or forgetting some analysis. Anyway it looks like:

Let $G:\Bbb{R}^n \rightarrow \Bbb{R}$ be a smooth function, and let $dG$ and $d^2G$ denote the first and second derivatives of $G$. They note here that "the $v$ dependence of $dG$ and $d^2G$ is being suppressed." $v$ being a function of $x$ and $t$ that satisfies a generic form PDE: $$ \frac{\partial v}{\partial t}=\epsilon D \Delta v +\Sigma M^i \frac{\partial v}{\partial x^i}+f,\quad\epsilon>0. $$ And satisfies some boundary conditions. Continuing, with $\gamma\left(t\right)$ being a smooth curve in $\Bbb{R}^n$ and with dots denoting time derivatives, they note that: $$ \begin{split} \dot{\left(G\circ v \right)} &=dG\left(\dot{\gamma}\left(t\right) \right)\\ \ddot{\left(G\circ v \right)}&=dG^2\left(\dot{\gamma}\left(t\right) \right) +dG\left(\ddot{\gamma}\left(t\right) \right) \end{split}$$

Finally, suppose $v:\Bbb{R}^m \rightarrow \Bbb{R}^n$ is a smooth function with range in the set $\Sigma=\{v:G\left(v\right)\leq 0\}$ and that $G\circ v\left(x_0\right)=0$. Then for $\xi\in\Bbb{R}^m$ and $h\in\mathbb{R}$, $G\circ v\left(x_0+h\xi \right)\leq G\circ v\left(x_0\right)$. After this they make use of a Taylor expansion to show $dG\left(dv\left(\xi\right) \right)=0$ as $h$ goes to $0$.

I'm not sure if the information above the last paragraph is helpful, but I figured I'd include just in case. They use some of it later with a convexity argument to reach more conclusions. What's mixing me up here is how the range of $v$ is defined. I understand it to mean that $v$ maps points in $\Bbb R^m$ to points in $\Bbb R^n$ that, when $G$ is applied to them, give positive real numbers. So, if the value of $G\circ v\left(x_0\right)=0$ and I move in some direction $\xi$ and distance $h$ away from $x_0$, I would expect that that value is also going to be $\geq 0$ given the restrictions on the range of $v$. However, the inequality they give suggests that the quantity is $\leq 0$. Can you clarify?

Sorry for the long post.