A recent paper proves the existence of homeomorphic but not diffeomorphic Mazur manifolds (see also examples of exotic pairs of contractible Stein manifolds).

Let's call them $M_1$ and $M_2$. If we glue $W= M_1 \cup_{\partial M_1=\partial M_2} M_2$, then we get a manifold homeomorphic to $S^4$ by Freedman's theorem. Doubling $M_1$ or $M_2$ yields the standard smooth $S^4$ by a theorem of Mazur. But I'm wondering if $W$ is diffeomorphic to $S^4$?

Actually, I don't know if twisting the double of an Akbulut cork can yield an exotic $S^4$? Any exotic pair of manifolds are related by twisting along a cork, so I guess I'm asking whether anything more is known when the complementary contractible manifolds are homeomorphic?

  • 2
    $\begingroup$ It can be worth saying that the double is guaranteed to be S^4 only if the Mazur manifold M you start with is made of 0-, 1-, and 2-handles along an Andrews-Curtis trivial presentation. This is not guaranteed for any contractible M, not even if made only of 0-, 1-, and 2-handles. $\endgroup$ Apr 14, 2020 at 14:25
  • $\begingroup$ @BrunoMartelli I see, yes I just had in mind the classic sort of example with one 0,1 & 2-handle like the examples in the paper. $\endgroup$
    – Ian Agol
    Apr 14, 2020 at 15:17

1 Answer 1


enter image description here

It is diffeomorphic to $S^4$. I have drawn the Kirby picture, please let me know if it is not clear.

(Also I'm sorry that I don't know how to draw Kirby picture in computer so I do old fashioned drawing on notebook)

The Key ideas of the proof are,

1)When we upside down a compact 4 manifold $M$ with 0,1 and 2 handle, then 0 handle becomes 4 handle and 1 handle become 3 handle. Now when we upside down the 2 handle, the co-core become the new attaching circle, which is an unknotted meridian to the orginial attaching cirle of the 2 handle. And the new attaching framing is trivial,i.e, 0. [This is what we use to draw the kirby picture of $M_1\cup M_2$, using the fact that they have identical boundary]

2) 0 framed meridian of a knot always helps to resolve all the crossings of a knot (handle slides).[ This is what we use to simplify the Kirby picture]

3) If a 1 handle and a 2 handle (in the picture a dotted 1 handle and the attaching circle of the 2 handle) geometrically intersect at one point, then they cancel each other. Similarly a 0 framed unknotted 2 handle get canceled by a 3 handle. [this is what we used in last two steps to cancel all but only one 0 handle and one 4 handle and thus we get $S^4$].

  • 1
    $\begingroup$ Okay, thanks, I think I get the idea. $\endgroup$
    – Ian Agol
    Apr 12, 2020 at 19:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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