Trouble with structure sheaf, derivations, tangent space

65 Views Asked by At

I'm teaching myself differential geometry, and the author I've been reading opts to work with germs of locally defined functions on a (not necessarily smooth) manifold to describe the tangent space. So, in what follows, $\mathcal{O}^{(k)}_{M,p}$ will denote the set of germs of $C^k$ functions defined in some open neighborhood of $p\in M$, and $\mathcal{S}^{(k)}_{M,p}$ will denote the subset of germs which are stationary at $p$, i.e. where all partial derivatives vanish in any chart. The tangent space $T_p M$ is then defined as the set of linear forms $L:\mathcal{O}^{(k)}_{M,p} \to \mathbb{R}$ which vanish on $\mathcal{S}^{(k)}_{M,p}$, which is isomorphic to the quotient dual $\left( \mathcal{O}^{(k)}_{M,p} / \mathcal{S}^{(k)}_{M,p} \right)^*$.

I'm trying to prove the correspondence between tangent vectors and linear derivations. The partial differentiation operation in $\mathcal{O}^{(k)}_{M,p}$ exhibits a property similar to the Leibniz rule.

$$ \left( \dfrac{\partial}{\partial x_i} \right)_p \left( \mathbf{f}\mathbf{g} \right) = \mathbf{f}(p) \left\lbrace \left( \dfrac{\partial}{\partial x_i} \right)_p \mathbf{g} \right\rbrace + \left\lbrace \left( \dfrac{\partial}{\partial x_i} \right)_p \mathbf{f} \right\rbrace \mathbf{g}(p)$$

Strictly speaking, a derivation should be of the form $D:R\to M$ where $R$ is a ring and $M$ is an $R$-bimodule (possibly $R$ itself). In the above, the pesky evaluation at the point $p$ serves to get in the way. We can consider the partial derivative map as a binary operator

$$ \left( \dfrac{\partial}{\partial x_i} \right) : \mathcal{O}^{(k)}_{M,p} \times U \to \mathbb{R} : (\mathbf{f},q) \mapsto \left( \dfrac{\partial}{\partial x_i} \right)_q \mathbf{f} $$

where $U$ is any open neighborhood of $p$. Then the Leibniz rule can take its proper form.

$$ \left( \dfrac{\partial}{\partial x_i} \right) \left( \mathbf{f}\mathbf{g} \right) = \mathbf{f} \left\lbrace \left( \dfrac{\partial}{\partial x_i} \right) \mathbf{g} \right\rbrace + \left\lbrace \left( \dfrac{\partial}{\partial x_i} \right) \mathbf{f} \right\rbrace \mathbf{g} $$

My question is this: how can we show that every tangent vector $u\in T_p M$ corresponds bijectively to a derivation $D: \mathcal{O}^{(k)}_{M,p} \to \mathcal{O}^{(k)}_{M,p}$? How might we extend this to $\mathcal{O}^{(k)}_{M,U}$, the germs of functions defined on some open subset, or eventually to $C^{(k)}(M,\mathbb{R})$, the global functions on $M$?