Sorry if the title is confusing, I don't know if the terminology is standard. For my homework this week I have to prove the following:

Let $M$ be a smooth manifold and let $p \in M$. Let $I_p$ denote the set of smooth functions $f:M \to \mathbb{R}$ such that $f(p)=0$. Let $I_p^2$ denote the set of sums $\sum_{i=1}^k f_ig_i$ where $k$ is a nonnegative integer and $f_i,g_i \in I_p$. Show that the quotient vector space $I_p/I_p^2$ is canonically isomorphic to the cotangent space $T_p^*M$.

I found the map $f \mapsto \phi_f$, where $\phi_f(v)=vf$ for $v \in T_pM$, and I've proved every part except injectivity. I've tried finding a basis for $I_p/I_p^2$ to show that it has the same dimension as $T_pM$, and I've tried showing that if $f=g \bmod I_p^2$ then there exist derivations taking $f$ and $g$ to different numbers, and I've tried showing that for any smooth function vanishing at $p$ and not in $I_p^2$ there is a derivation taking it to a nonzero number. I tried working in coordinates but that didn't seem to help. Does anyone have any hints? Thanks!

  • 1
    $\begingroup$ I think you can choose as a basis of $I/I^2$ the set of coordinate functions $x_1,...,x_n$, since they determine the first order behaviour (Taylor series) of the function at the point. Does this make sense to you? $\endgroup$ – Franco Feb 17 '15 at 5:01
  • $\begingroup$ I thought about that, since I showed surjectivity by showing that the dual basis of $T_p^*M$ is the image of $\{x^1+I^2,\ldots,x^m+I^2\}$, but I didn't think to use Taylor series. I'll try that. $\endgroup$ – desi Feb 17 '15 at 5:43
  • $\begingroup$ @Franco, desi: you are not allowed to use coordinate functions because they are not defined on all of $M$. $\endgroup$ – Georges Elencwajg Feb 17 '15 at 7:49

This is obviously related to Taylor expansion, as sugggested by @Franco in the comment. Indeed, it is quite trivial if you have the following form of Taylor expansion (Theorem A.58 in John Lee's smooth manifold):

Let $U \subset \mathbb R^n$ be open and $p \in U$, and $f\in C^k(U)$ with $1 \le k\le \infty$. Then there is $g_1, \cdots g_n \in C^{k-1}(U)$, $g_i(p) = 0$ and

$$f(x) = f(p) + \sum_{i=1}^n \frac{\partial f}{\partial x_i}(p) (x_i - p_i) + \sum_{i=1}^n g_i(x) (x_i - p_i)$$

From this, you have that if $f(p) = 0$ and $\nabla f(p) = 0$, then there are smooth $f_1, \cdots, f_n$ and $g_1, \cdots , g_n$ so that $f = f_1 g_1 + \cdots f_ng_n$. That is, $f\in I_p^2$.

Of course now everything is local. But one can use a cutoff function to reduce the general case to this one.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.