There are many conjecturally-equivalent three-manifold Floer homologies, of which my understanding is the most-computable is Heegaard Floer homology.

What is the (Heegaard) Floer homology of a connect sum of $k$ copies of $S^2 \times S^1$? Is it trivial?

I am an outsider, and so please forgive me if my question is not quite well formed. E.g. I gather that HF depends on a $\mathrm{Spin}^c$ structure, so at best I would understand the answer for all choices; the value of Heegaard Floer homology is really an exact triangle, not a vector space, so I guess I'm asking for that triangle; perhaps there are other technicalities I am unaware of.


1 Answer 1


The explicit calculation of this Heegaard-Floer homology (at least, $HF^-$, and of the unique $\text{Spin}^c$ structure for which $c_1(\mathfrak s) = 0$) was carried out in the paper in which it was introduced, and indeed is an important part of the proof that Heegaard-Floer homology is well-defined. See Lemma 9.1. One also has the isomorphisms $HF^+ \cong \Bbb Z[U,U^{-1}]/U\Bbb Z[U] \otimes H_*(T^g;\Bbb Z)$ and $HF^\infty \cong \Bbb Z[U^{-1},U] \otimes H_*(T^g;\Bbb Z)$. The exact sequence relating the triple ($HF^+, HF^-, HF^\infty)$ is exactly what you think it is. When the $\text{Spin}^c$ structure is nonzero, all the Floer groups are zero. (These four groups are all computable by hand, as done in the original paper.)

In this case, the Seiberg-Witten Floer homology is computable, because $S^2 \times S^1$ has positive scalar curvature, as do all connect-sums with itself by Gromov-Lawson. When $c_1(\mathfrak s)$ is not torsion (in this case, not zero), all the Floer groups are trivial, and in the case that $c_1(\mathfrak s) = 0$ all the Floer groups are explicitly calculable, as is the exact sequence (one of the connecting maps is zero) - for the precise result, see Chapter 36 of Kronheimer-Mrowka, "Monopoles and 3-manifolds".

To compare these results, one should note the relationship (now proved, thanks to Colin-Ghiggini-Honda and Kuthulan-Lee-Taubes) $\overline{HM} \cong HF^\infty$, $\check{HM} \cong HF^-$, $\widehat{HM} \cong HF^+$. $\widehat{HF}$ corresponds to a reduced Seiberg-Witten Floer homology not described in that chapter of Kronheimer-Mrowka, $\widetilde{HM}$.

  • $\begingroup$ If anyone knows how to get the wide-check that really should be over HM-to, please feel free to edit it in... I don't know how to do that in MathJaX. $\endgroup$
    – mme
    Commented Apr 20, 2016 at 6:57
  • 2
    $\begingroup$ If someone's interested in the absolute grading and the action by $H_2$ of the 3-manifold: the grading on $HF^+$ needs to be shifted by $-g/2$ (so that the bottom-most homogeneous element sits in degree $-g/2$); and the action is so-called "standard" in this case. I think that the reference for both statements is the Absolutely graded paper by Ozsváth and Szabó. $\endgroup$ Commented Apr 20, 2016 at 8:08

Your Answer

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

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