Here is what I have to show:
Let $(M,\mathcal{F})$ and $(N,\mathcal{F}')$ be smooth manifolds of class $C^{\infty}$ and let $\Psi:M\to N$ be a continuous map. Show that the following conditions are equivalent:
- In local coordinates, $\Psi$ is given by a smooth map.
- $\Psi$ is a morphism of ringed spaces.
Definitions (manifolds are assumed Hausdorff):
- Let $(M,\mathcal{F})$ be a topological manifold $M$ (i.e. a topological space locally homeomorphic to $\mathbb{R}^{m}$) equipped with a sheaf of functions $\mathcal{F}$. It is said to be a smooth manifold of class $\mathcal{C}^{\infty}$ if every point has a neighbourhood isomorphic to the ringed space $(\mathbb{R}^{n},\mathcal{G})$ where $\mathcal{G}$ is a ring of functions of this class.
- A coordinate system on an open subset $U$ of a smooth manifold $(M,\mathcal{F})$ is an isomorphism between $(U,\mathcal{F})$ and an open subset in $(\mathbb{R}^{m},\mathcal{G})$ where $\mathcal{G}$ are functions of the same class on $\mathbb{R}^{m}$.
- $\implies$ 2.
My first problem is what does "In local coordinates, $\Psi$ is given by a smooth map" mean? When I first learned Differential Geometry, it used to mean that for some appropriate charts $(U,\phi)$ and $(V,\varphi)$ of $M$ and $N$, we had $$\varphi\circ\Psi\circ\phi^{-1}\vert_{\phi\left(\,\Psi^{-1}(V)\cap U\,\right)}:\mathbb{R}^{m}\to\mathbb{R}^{n}\in C^{\infty}$$ I do not understand how to make a link between the two definitions if they are indeed equivalent.
This problem left aside, if I take the "new" definition and if I understand correctly, $\Psi$ has to be smooth for some coordinate system on $V$, an open subset in $N$? Let $\Lambda_{V}$ be a coordinate system on $V$.
I need to prove that $\Psi$ is a morphism of ringed spaces, i.e. $\forall\,V\,\text{open}\subset N,\,\forall\,f\in\mathcal{F}'(V)$ we have $$f\circ\Psi\in\mathcal{F}(\Psi^{-1}(V))$$
For all $f\in\mathcal{F}'(V)$, we have $f\circ\Lambda_{V}^{-1}\in\mathcal{G}(\Lambda_{V}(G))$ since $\Lambda_{V}^{-1}$ is well-defined and is a morphism of ringed spaces: $$\Lambda_{V}^{-1}:(\Lambda_{V}(V),\mathcal{G}(V))\to(V,\mathcal{F}'(V))$$
We also have
$$\Lambda_{V}\circ\Psi:\Psi^{-1}(V)\subset M\to\Lambda_{V}(V)\subset\mathbb{R}^{n}$$
But now, if it is the right way to begin, how can I go further?