3
$\begingroup$

Let $K \subseteq S^3$ be a knot in the $3$-sphere and assume there exists a smooth map $p \colon S^3\setminus K \to S^1$ which is a fiber bundle.

For every point $\require{enclose} \enclose{horizontalstrike}{t \in S^1}$ the fiber $\enclose{horizontalstrike}{p^{-1}(t)}$ together with $\enclose{horizontalstrike}{K}$ forms a compact 2-dimensional submanifold $\enclose{horizontalstrike}{F_t}$ of $\enclose{horizontalstrike}{S^3}$ such that $\enclose{horizontalstrike}{\partial F_t = K}$, though $\enclose{horizontalstrike}{F_t}$ may not be orientable/connected(?).

Question: Does there exists a smooth fiber bundle $q \colon S^3\setminus K \to S^1$ such that for every $t \in S^1$ the fiber $p^{-1}(t)$ together with $K$ forms an orientable connected compact 2-dimensional submanifold $F_t$ of $S^3$ with $\partial F_t = K$?

Edit: What I get from the comments:

  • For every $t \in S^1$ the fiber $p^{-1}(\{t\}) = F$ is orientable as locally we have that $(0,1) \times F$ is diffeomorphic to an open subset $U$ of $S^3\setminus K$. Since $S^3\setminus K$ is orientable, so is $U$ and then also $F$.

  • We can construct a smooth map $q \colon S^3 \setminus K \to S^1$ such that the fiber is connected: Assume $F$ has $n$ components. From the exact sequence $$ 0 \to \pi_1(F) \to \pi_1(S^3 \setminus K) \to \pi_1(S^1) = \mathbb Z \to \pi_0(F) = \mathbb Z_n \to 0 $$ we obtain a lift: $$ \require{AMScd} \begin{CD} S^3\setminus K @>\exists q>> S^1 \\ @V=VV @VVx \mapsto x^nV \\ S^3 \setminus K @>p>> S^1 \end{CD} $$ and $q$ is a smooth fiber bundle with connected fiber.

  • It remains to show that $\partial (F \cup K) = K$.

$\endgroup$
6
  • 1
    $\begingroup$ I confess I thought that $S^3\smallsetminus K$ fibring over $S^1$ was the definition of a fibred knot. But now I see that Wikipedia, say, gives a different definition: en.wikipedia.org/wiki/Fibered_knot. So I suppose your question is why the two definitions coincide? $\endgroup$
    – HJRW
    Commented Jun 22 at 7:47
  • 1
    $\begingroup$ As pointed out by HJRW, the interesting question here would be: if $S^3\setminus K$ fibers, why it does so in a way that $\partial F_t = K$? $\endgroup$ Commented Jun 22 at 15:26
  • 1
    $\begingroup$ If the fiber is disconnected, then it must consist of n disjoint isotopic Seifert surfaces. Hence the fibration factors through a map $S^1\to S^1$ which is a degree n cover (up to isotopy). $\endgroup$
    – Ian Agol
    Commented Jun 22 at 21:19
  • 1
    $\begingroup$ @HJRW - this question asker is perhaps new to higher mathematics, so will not understand that by "boundary of $S^3 - K$" you mean "boundary of the compact manifold $E_K = S^3 - n(K)$ where $n(K)$ is a small open tubular neighborhood of $K$"... $\endgroup$
    – Sam Nead
    Commented Jun 23 at 18:22
  • 1
    $\begingroup$ @SamNead: Yes, that's a useful clarification. $\endgroup$
    – HJRW
    Commented Jun 24 at 13:46

1 Answer 1

2
$\begingroup$

Suppose that $K$ is a smooth knot in the three-sphere $S^3$. We set $X = S^3 - K$. Suppose that $p \colon X \to S^1$ is a smooth fiber bundle. Let $F_z = p^{-1}(z)$ be the fiber lying over $z$. Exercise: the inclusion of $F_z$ into $X$ is $\pi_1$-injective.

We now take $U$ to be a small, closed tubular neighborhood of $K$. Note that $U - K$ is homeomorphic to a torus cross a ray. Let $T = \partial U$. We perturb $p$ to make the intersections of the surfaces $F_z$ and $T$ "as transverse as possible". Let $\mathcal{F}$ be the resulting foliation of $T$ given by the sets $T \cap F_z$. This foliation is possibly singular, but it has only finitely many singularities. Exercise: If $\mathcal{F}$ has center singularities, then we can alter $p \colon X \to S^1$ by a diffeotopy to remove them. (The solution to the exercise uses the fact that the $F_z$ are $\pi_1$-injective.)

Since the Euler characteristic of $T$ is zero, once the foliation $\mathcal{F}$ has no centers, it has no singularities at all. We now "foliate" $U$ by annuli $A_z$ where $\partial A_z = K \cup (T \cap F_z)$ and where $A_z \cap A_w = K$ iff $z \neq w$.

The surfaces $(F_z - U) \cup A_z$, after smoothing, have the desired properties.

$\endgroup$
1
  • $\begingroup$ I've been trying to understand your answer, but i struggle a lot. Can you explain what you mean by "center singularities" and how to remove them? Also how does the Euler characteristic being zero imply that a foliation has no singularities? $\endgroup$ Commented Jun 27 at 20:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .