Let $M$ and $N$ two manifolds which have the same dimension, $f:M\to N$ a map $\mathcal{C^\infty}$. We suppose that $M$ is compact and we have $b$ a regular value of $f$.

First, I have to prove that $f^{-1}(\{b\})$ is a compact and finite set.

For compacity, I thought about the topology of $\mathbb{R}^m$ using the fact that if a set is closed and bounded it's compact. I tried to define a map $\mathbb{R}^m\to \mathbb{R}^m$ using homeomorphism $\phi$ and $\psi$ but I don't know if it is a good method. To prove that it is finite I have no idea.

Second, I have to prove that there exists an open set $V$ of $N$ which contains $b$ such as for all $c\in V$ we have $\mid f^{-1}(\{c\}) \mid\ge\mid f^{-1}(\{b\}) \mid$.

Thanks in advance !

  • $\begingroup$ If $b$ is a regular value, $f^{-1}(b)$ is a manifold of dimension $\text{dim} M - \text{dim} N = 0$ by the local submersion theorem. So that's a discrete set of points. That's a closed set because it's a level set. Now $M$ is compact, so it's compact as it's a closed subspace. Compact discrete sets are always finite. $\endgroup$ – Balarka Sen Jun 3 '16 at 18:45

Since $f$ is continuous, $f^{-1}(b)$ is closed. It is compact since it is a closed subset of a compact set.

Let $x\in f^{-1}(b)$, since $f$ is regular, $df_x:T_xM\rightarrow T_bN$ is an isomorphism. You can work locally and restrict $f$ to a neighborhood $U$ of $x$ such that $f(U)\subset V$ where $V$ is a neighborhood of $b$ and $U$ and $V$ are diffeomorphic to open subsets of $R^n$. This implies that the differential $f_{\mid U}:U\rightarrow V$ is invertible so it is a local diffeomorphism. We deduce that there exists an open subset $x\in U_x$ such that the restriction of $f$ to $U_x$ is a diffeomorphic onto its image. This implies that $f^{-1}(b)\cap U_x=\{x\}$. Thus $f^{-1}(b)$ is discrete. A discrete compact set is finite.

  • $\begingroup$ I don't understand the last implication $\endgroup$ – Maman Jun 3 '16 at 15:59
  • $\begingroup$ If $X$ is discrete and compact, suppose that $X$ is infinite, you can construct a sequence $(x_n)$ of points pairwise distinct. It has an accumulation point $x\in X$, but you have $U_x$ such that $U_x$ is open and $U_x=\{x\}$, thus since there exists $N$ such that $n>N$ implies $x_n\in U_x$, $x_n=x, n>N$ this is a contradiction with the fact that the points of the sequence are pairwise distinct. $\endgroup$ – Tsemo Aristide Jun 3 '16 at 16:07
  • $\begingroup$ For the second question, we can say that if $c$ is not a regular value of $f$ then $f^{-1}(\{c\})$ is not finite ? $\endgroup$ – Maman Jun 3 '16 at 17:14
  • $\begingroup$ It is possible, take for example the constant map whose image is $y$, $y$ is not a regular value and it's inverse image maybe infinite $\endgroup$ – Tsemo Aristide Jun 3 '16 at 17:17
  • $\begingroup$ And for the open set I would choose $f_{\mid U} (U)=V$, the equality is given if we take $c=b$ and for the other cases $f^{-1}(\{c\})$ which is infinite. But the inverse image for $c$ is not necessary in $U$. $\endgroup$ – Maman Jun 3 '16 at 17:21

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.