10
$\begingroup$

It is known that every orientable 3-manfiold can be obtained as a ramified cover of S3 with a ramification (of some order) at a link in S3. I am curious if there is a reasonable characterization of 3-manifolds that cover 3-torus?

Added. Notice that such a manifold is enlargeble, so it does not admit a metric of positive scalar curvature, so for example a connected sum of n copies of S2 x S1 does not admit a ramified cover of T3 (as far as I understand).

$\endgroup$
8
  • $\begingroup$ Any conjectures? Clearly there is an onto homomorphism $\pi_1M\to Z^3$; can you do better? $\endgroup$ Nov 14 '09 at 23:01
  • $\begingroup$ I don't really know anything about this question, and can not do better than the homeomorphism to Z^3. For every M^3 its connected sum with several T^3 admits a cover to T^3 :) $\endgroup$ Nov 14 '09 at 23:58
  • $\begingroup$ Conjecture: M is a branched cover of T^3 iff the first homology group of M has rank at least three. $\endgroup$
    – Sam Nead
    Nov 15 '09 at 0:59
  • $\begingroup$ (M assumed to be closed, orientable, connected.) $\endgroup$
    – Sam Nead
    Nov 15 '09 at 1:00
  • 1
    $\begingroup$ No, in fact I did not read the proof for S^3. Also the existence of onto homeo $\pi_1M\to Z^3$ as well having rank of first homology at lest 3 is no sufficient because covers of T^3 don't admit a metric of positive scalar curvature, so the conncted sum of 3 S^2xS^1 will be a counterexample. $\endgroup$ Nov 15 '09 at 10:34
12
$\begingroup$

Note that a branched covering induces an injection of rational cohomology rings, by transfer considerations. Therefore the cohomology of a manifold that is a branched covering of $T^3$ must contain three classes of degree 1 whose triple cup product is nontrivial.

This condition on a manifold $M^3$ implies the existence of a map $M^3\to T^3$ of nonzero degree. Passing to a covering space of $T^3$ if necessary we obtain such a map that is also surjective on $\pi_1$. Assuming the resulting map is of degree $\ge 3$, the main result of [Edmonds, Allan L.Deformation of maps to branched coverings in dimension three. Math. Ann. 245 (1979), no. 3, 273--279.] implies that this map is homotopic to a branched covering.

$\endgroup$
0

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.