MathOverflow is a question and answer site for professional mathematicians. 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

Let $\alpha$ be an orientation-preserving automorphism of the torus $T^2 = S^1 \times S^1.$ Since the mapping torus $M_{\alpha} = T^2 \times [0, 1] / (x, 0) \sim (\alpha(x), 1)$ is an orientable compact 3-dimensional manifold, it is $spin^c.$

But is there a $spin^c$ structure on $M_{\alpha}$ that restricts to the trivial $spin^c$ structure on $T^2 \times [1/4, 3/4]$ ?

Thank you !

share|cite|improve this question
Actually, every $spin^c$-structure on $T^2\times (0,1)$ is the restriction of one on $M_\alpha$. The $spin^c$-structures form a torsor for $H^2$, and the restriction map $H^2(M_\alpha)\to H^2(T^2\times (0,1))$ is surjective. – Tim Perutz Aug 16 '11 at 22:48
up vote 3 down vote accepted

Yes. Moreover, there's a trivialization of the tangent bundle of $M_\alpha$ that restricts to the standard trivialization of the tangent bundle of $T^2 \times [1/4,3/4]$. By "the standard" trivialization I mean one that's invariant under the action of $T^2$. You construct it by hand -- take the invariant one and apply your gluing map $(x,0) \sim (\alpha(x),1)$. The gluing map does not preserve the trivialization on the boundary, so on one end of the boundary you need to rotate it a little, to ensure that it is preserved.

So what I'm saying is you can do it for $spin$ structures, and every $spin$ structure induces a $spin^c$ structure, giving you what you want.

edit in response to Alain's comment: I will describe a trivialization of the tangent bundle of $T^2 \times [0,1]$ which descends to a trivialization of the tangent bundle of $M_\alpha$, and this works for any $\alpha \in SL_2 \mathbb Z$.

Think of $T^2$ as $\mathbb R^2 / \mathbb Z^2$. So I will define the trivialization up on the space $\mathbb R^2 \times [0,1]$. Coordinates in $\mathbb R^2 \times [0,1]$ will be given by $(x,y,z)$. A trivialization of the tangent space to $(x,y,z)$ is a matrix in $GL_3(\mathbb R)$. So what we need to do is describe a matrix $A_{(x,y,z)} \in GL_3(\mathbb R)$ which respects all the identifications on the boundary (and tangent spaces) and which depends continuously on $(x,y,z)$.

$SL_2 \mathbb Z \subset GL_2(\mathbb R)$ is in the path-component of the identity matrix, so given $\alpha \in SL_2 \mathbb Z$, let $\beta : [0,1] \to GL_2(\mathbb R)$ be the path such that $\beta(0) = I$ and $\beta(1) = \alpha$. Then $A_{(x,y,z)} = \overline{\beta(z)}$ is our trivialization of the tangent bundle of $\mathbb R^2 \times [0,1]$, and this descends (by design) to the tangent bundle of $M_\alpha$. In the above, if $\beta(z) = \pmatrix{ b_{11} & b_{12} \cr b_{21} & b_{22} }$, let $\overline{\beta(z)} = \pmatrix{ b_{11} & b_{12} & 0 \cr b_{21} & b_{22} & 0 \cr 0 & 0 & 1 }$

share|cite|improve this answer
@ Ryan: Can you be more specific on "rotating a little"? Since $\alpha$ is not isotopic to the identity (assuming $\alpha\neq Id$), I don't see how you do it. – Alain Valette Aug 16 '11 at 21:52
You've got much more flexibility when it comes to trivializations of the tangent bundle -- the $SL_2 \mathbb Z$ restriction is only for the gluing map. Anyhow, I hope the edit helps. – Ryan Budney Aug 16 '11 at 22:53
@ Ryan, after editing: OK, thank you for the explanation. Meanwhile I also learned of a theorem by Steenrod, that every orientable 3-manifold is parallelizable! – Alain Valette Aug 17 '11 at 12:28

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.