Edit: According to the interesting comments of Michael Albanese and Nick L we revise the question as follows:

By manifold compactification of a manifold $M$ we mean a compact manifold $\tilde{M}$ which contains $M$ as an open dense subset.

Assume that $M$ is an open connected manifold which admits a manifold compactification. Does $M$ necessarily admit a manifold compactification with zero Euler characteristic?

  • 7
    $\begingroup$ @Neal: Once you remove the pinch point, what's left is homeomorphic to a cylinder which embeds in a torus. $\endgroup$ – Michael Albanese Nov 3 '17 at 20:32
  • 1
    $\begingroup$ Possible duplicate of Compactification of a manifold $\endgroup$ – Michael Albanese Nov 3 '17 at 20:49
  • 3
    $\begingroup$ No. The complement of a point in an oriented surface of genus 2 cannot be embedded in a torus or a Klein bottle. $\endgroup$ – Tom Goodwillie Nov 4 '17 at 12:38
  • 1
    $\begingroup$ Perhaps use the fact that the genus two surface contains a separating curve which is non-trivial in $\pi_{1}$, You just have to prove that the class of this curve doesn't die in the compactification. $\endgroup$ – Nick L Nov 5 '17 at 16:52
  • 1
    $\begingroup$ No it is trivial in homology but non-trivial in the fundamenetal group. This is discussed here math.stackexchange.com/questions/1031069/…. $\endgroup$ – Nick L Nov 12 '17 at 11:22

First note that odd dimensions the question of Euler characteristic $0$ is automatic, $M$ will embed in the orientable double cover of $\tilde{M}$, which will have $\chi = 0$ by Poincare Duality.

In even dimension = $2n$ (we assume $n > 1$), we recall the following fact. If $M_{1},M_{2}$ are compact connected manifolds then $\chi(M_{1} \# M_{2}) = \chi(M_{1}) + \chi(M_{2}) - \chi(S^{2n}) = \chi(M_{1}) + \chi(M_{2}) - 2$.

To prove that some embedding into any manifold implies embedding in to Euler characteristic $0$ manifold, it is sufficient to show that any integer is equal to the Euler characteristic of some manifold of dimension $2n$, since we can do connect sums on the complement of the embedding $M \hookrightarrow \tilde{M}$ (It is easy to see that the embedding can be changed so that this complement contains an open set) to shift the Euler characteristic of $\tilde{M}$ to the correct value.

In dimension $4$ we have $\chi(\mathbb{R}\mathbb{P}^{2} \times \mathbb{R}\mathbb{P}^{2}) =1$ and $\chi(\mathbb{C} \mathbb{P}^{2} ) = 3$. So for any $4$-manifold $N$ connect summing with $\mathbb{R}\mathbb{P}^{2} \times \mathbb{R}\mathbb{P}^{2}$ subtracts 1 from $\chi(N)$, connect summing with $\mathbb{C} \mathbb{P}^{2} $ adds one to $\chi(N)$, hence there is a $4$-manifold with Euler characteristic equal to any integer. In higher even dimension taking appropriate products with $\mathbb{R}\mathbb{P}^{2}$ will give the same result.

Edit 1 As Misha points out we can now ensure the image is dense by using the fact that every connected n-manifold is a compactification of an open n-cell.

Edit 2. Note that the above solution holds in even dimensions atleast $4$. I will give details a counterexample in the dimension $2$ case (which was pointed out Tom Goodwillie).

Let $S$ be an orientable surface of genus $g \geq 2$. I will show there is no embedding $I: S \setminus \{p\} \hookrightarrow S'$ (for some $p \in S$). Where $S'$ is a compact surface of Euler characteristic $0$ (i.e. a torus or a Klein bottle).

We argue by contractiction, suppose such an embedding $I$, exists. Let $C$ be the boundary of a small neighbourhood of $p$.

Suppose there is an embbeding $I: S \setminus \{p\} \hookrightarrow S' $, then image of $I(C)$ decomposes $S'$ as the connect sum of $S$ and compact surface $\Sigma$. In symbols: $S' = S \# \Sigma$. Hence by the formula for Euler characteristic of a connect sum given above $\chi(S') \leq -2$.

To see this note by the classification of surfaces $\chi(\Sigma) \leq 2$ and by direct computation $\chi(S) = 2-2g \leq -2$, the fact that $\chi(S') \leq -2$ now follows directly from the formula for Euler characteristic of a connect sum. This is the desired contradiction.

  • 3
    $\begingroup$ You can easily complete your solution by using the fact that every connected n-manifold is a compactification of an open n-cell. $\endgroup$ – Misha Nov 5 '17 at 2:48
  • $\begingroup$ Thank you for your answer.And I thank you @Misha for your comment. $\endgroup$ – Ali Taghavi Nov 5 '17 at 11:52
  • 4
    $\begingroup$ @Misha: will you please elaborate your comment: how does the fact you quoted help to find a dense embedding out of an open embedding? I would say one needs a more general "every connected n-manifold is a compactification of (any manifold diffeomorphic to) any nonempty open subset of its" (which I think is also true). $\endgroup$ – Pietro Majer Nov 12 '17 at 17:23

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.