2

I am having much trouble reading the proof Local Immersion Theorem in Guillemin & Pallock's Differential Topology on Page 15.

Now we try to augment $g$ so that the Inverse Function Theorem may be applied.

How does it mean here to "augement $g$"?

As $dg_0: \mathbb{R}^{k} \rightarrow \mathbb{R}^{l}$ is injective.

Why $dg_0$ is injective?

Could some one explain these for me? Thanks.

WishingFish
  • 2,412

1 Answers1

4

The Inverse Function Theorem requires that you are mapping between spaces of the same dimension. In the present case, you are mapping from a space of dimension $k$ to a space of dimension $l$, where $l \geq k$. Hence we need to modify $g$ somehow to get a map between spaces of, say, dimension $l$. Guillemin and Pollack achieve this by using $U \times \Bbb R^{l-k}$ in place of $U$ and making a new function $G: U \times \Bbb R^{l-k}$ that acts as $g$ on the $U$ part and as the identity on the $\Bbb R^{l-k}$ part.

The way the proof is set up, we have that $$g = \psi^{-1} \circ f \circ \phi,$$ where $f$ is an immersion at $x \in X$ and $\phi$, $\psi$ are local parametrizations. This means that $df_x$ is injective and $d\phi_0$, $d(\psi^{-1})_0$ are isomorphisms. By the chain rule, $$dg_0 = d(\psi^{-1})_0 \circ df_x \circ d\phi_0,$$ from which it follows that $dg_0$ is injective.

Henry T. Horton
  • 18,318
  • 5
  • 60
  • 77
  • So clear and helpful! Thank you Henry! – WishingFish Jun 30 '13 at 22:12
  • 1
    Did you mean to say $d(\psi^{-1})_y$ instead of $d(\psi^{-1})_0$? The map $\psi^{-1}$ goes from $Y$ to $V$, and it is not guaranteed that $Y$ contains $0$. Or did you mean the inverse of the differential of $\psi$ at $0$? – user557 Jan 31 '18 at 01:02