Reference: McDuff-Salamon's book J-holomorphic curves and Symplectic Topology, second edition.

In section 6.7, the book introduces the moduli space of $\{J_z\}$-holomorphic curves (see page 184 for example), where $\{ J_z \}_{z\in \Sigma}$ is a smooth family of almost complex structure parametrized by the point $z$ of the domain $\Sigma$. What could be advantages of this compared to the ordinary moduli space for an almost complex structure $J$ independent of $z$?

In particular, when this book discusses the gluing theorem in chapter 10, we also consider parametrized almost complex structures. How does such a slightly general gluing theorem apply more widely than the gluing theorem only for a fixed almost complex structure $J$? I am just wondering if it is really necessary to consider the most general cases at the cost of simplicity.

  • 1
    $\begingroup$ To show that the floer homology is independent of the chosen acs, one needs to consider families of such. $\endgroup$
    – Thomas Rot
    Mar 20, 2017 at 6:14
  • $\begingroup$ And when dealing with group actions (an equivariant scenario), you'll want domain-dependent $J$ to help with transversality issues. $\endgroup$ Mar 20, 2017 at 7:16
  • $\begingroup$ In projecteuclid.org/euclid.jsg/1210083200, these domain dependent almost complex structures have been applied to obtain enough transversality to define a version of genus zero Gromov-Witten invariants. Perhaps this could be more easily done by considering inhomogeneous perturbations of the equation as is done in work of Ruan-Tian, FOOO, Pardon etc. In any case, to obtain transversality for GW theory, one seems to need to consider some generalization of the usual pseudoholomorphic curve equation for surface-independent J. $\endgroup$ Mar 20, 2017 at 19:41

1 Answer 1


Sorry if I repeat answers already given. This is an answer I got from Auroux's lectures notes on Fukaya category. In defining the Lagrangian intersection Floer homology, an immediate issue is the transversality of the intersection. But that is not all: even if you have transverse intersection, you need transversality of the moduli space $\mathcal{M}(p,q;\beta,J)$ of $J$-holomorphic strips between $p$ and $q$ to ensure that it is a manifold.

Here, transversality means that at each $u \in \mathcal{M}(p,q;\beta,J)$, the linearised operator of $\bar{\partial}_J$ is surjective. This property does not hold in general for a fixed almost-complex structure $J$. However, it does hold for a generic family of almost-complex structure $(J_z)_{z \in \Sigma}$. Depending on what you look at, you may need only a dependence on one real parameter etc.


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.