Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm stuck on the following problem: let $S$ be a compact orientable hypersurface in the symplectic manifold $(M,\omega)$. Prove that there exists a smooth function $H: M \to \mathbb R$ such that $0$ is a regular value of $H$ and $S \subset H^{-1}(0)$.

Since $S$ and $M$ are orientable, I can find a tubular neighborhood $N\simeq S \times (-\epsilon, \epsilon)$, which is open in $M$ and with $S$ corresponding to $S \times \{0\}$. Then $S$ is the inverse image of the regular value 0 under the projection onto the second factor $N \to (-\epsilon,\epsilon)$. Is there a clean way to see that I can extend this map to all of $M$ such that 0 remains a regular value?

share|cite|improve this question
use a partition of unity – user8268 Apr 21 '11 at 17:16
@user8268: Can you be more specific? If I simply extend the map by multiplying by a bump function then 0 will no longer be a regular value. – Eric O. Korman Apr 21 '11 at 17:17
@Eric: sorry, it was too sloppy. Since $M$ and $S$ are orientable, $S$ cuts $M$ to two pieces. You can take $H$ which is $+1$ on one piece, $-1$ on the other piece, and grows from $-1$ to $1$ on a tubular neighbourhood of $S$. (first construct that function on your tubular neighbourhood so that it's constant $\pm1$ away from $S\times(-\epsilon/2,\epsilon/2)$ and then extend it to $\pm1$ on the rest of $M$) – user8268 Apr 21 '11 at 17:38
@user8268: It looks like in your construction $S = H^{-1}(0)$? That can't work in general. – Eric O. Korman Apr 21 '11 at 17:49
ah, I'm on leave of absence from my mind :) But one more attempt: take a smooth function on $\mathbb{R}$ which is $1$ away from $(-\epsilon,\epsilon)$ and has two regular zeros, one at $0$ and the other at (say) $\epsilon/2$. This gives you a function on $S\times(-\epsilon,\epsilon)$ where $0$ is a regular value and $H^{-1}(0)$ are two copies of $S$. Extend this function by $1$ to the resto of $M$. – user8268 Apr 21 '11 at 18:22

Perhaps my idea is silly, but could not you take $H=f\circ pr_2$? where $pr_2$ is your projection on the second factor and $f$ is a real-valued smooth function on $]-\epsilon,+\epsilon[$ such that: its support is compact and $f(0)\neq 0$ is one of its regular values. For example you could take f(x)=exp(1/(x−2ϵ/3))exp(−1/(x+ϵ/3)) for x∈]−ϵ/3,2ϵ/3[ and f(x)=0 otherwise.

The proposed $H$, defined on $N$, has compact support $S\times\mathrm{supp}(f)$, so it prolongs to the smooth function on $M$ vanishing outside $N$. In such a way, you get a smooth function $H$ on $M$ with a constant value $f(0)$ on $S$ which is not one of its singular values.

share|cite|improve this answer
Do you really mean composition? $pr_2$ is not defined on all of $M$, only on an open set. If I multiply $pr_2$ by a bump function to extend it to $M$ then 0 will not be a regular value (since the function will be identically zero off of $N$). – Eric O. Korman Apr 21 '11 at 17:12
I mean the composition by $\circ$. Yes, $f\circ pr_2$ is defined only on the tubular neighborhood $N$, but its support is the compact $S\times\mathrm{supp}(f)$, so it extends to the smooth function on $M$ which vanishes outside $N$. By construction $f\circ pr_2$ is constant on $S$ with a value $f(0)\neq 0$ which is not singular. – Giuseppe Apr 21 '11 at 17:22
take for example $f(x)=\exp(1/(x-2\epsilon/3))\exp(-1/(x+\epsilon/3))$ for $x\in]-\epsilon/3,2\epsilon/3[$ and $f(x)=0$ otherwise. – Giuseppe Apr 21 '11 at 17:31
Say $H(S) = c$. Then in general $H^{-1}(c)$ will have to contain more than $S$. How can I guarantee other points in $H^{-1}(c)$ are not critical? – Eric O. Korman Apr 21 '11 at 17:52
Sure $H^{-1}(c)$ could contain more than $S$. It is properly $pr_2^{-1}(f^{-1}(f(0)))$ by constuction, and does not contain singular points because $pr_2$ is non singular and because I requested $f(0)$ is not a singular value for $f$( so in particular $c=f(0)\neq 0$. – Giuseppe Apr 21 '11 at 18:53

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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