I am trying to understand the differences between several definitions used in many texts in symplectic topology. Let $(M,\omega)$ be a symplectic manifold, and $c_1 \in H_2(M,\mathbb{Z})$ be its first Chern class.

The minimal Chern number $N_M$ of $M$ can be defined as the positive generator of the group $\langle c_1, H_2(M,\mathbb{Z}) \rangle$. On the other hand, I have seen another definition in Salamon's lectures on Floer homology: $$ N_M := \underset{k > 0}{\inf} \lbrace \exists v : S^2 \to M \ | \ \int_{S^2} v^*c_1 = k \rbrace. $$ If $N \neq \infty$, then Salamon writes that $\langle c_1, \pi_2(M) \rangle = N_M \mathbb{Z}$.

I would like to understand the meaning of this definition, and its relation to the first one. In particular, what does $c_1(A)$ means, for an element $A \in \pi_2(M)$ ? Is $\pi_2$ in this setting a notation for its image under the Hurewicz homomorphism ?

Another related definition confuses me as well: a symplectic manifold $(M,\omega)$ is called monotone if there exists $\tau \in \mathbb{R}$ such that $[\omega] = \tau c_1$, where $[\omega] \in H^2(M,\mathbb{R})$ is the cohomology class of $\omega$. In Salamon's lecture, the definition is again different: $$ \exists \tau \in \mathbb{R}, \quad \forall v : S^2 \to M, \ \int_{S^2} v^*\omega = \tau \int_{S^2} v^* c_1. $$ What is the relation between these two definitions ? Is the second equivalent to saying that $\langle \omega, \pi_2(M) \rangle = \tau \langle c_1, \pi_2(M) \rangle$, for a suitable meaning of this last equation ?

Thanks for your help !


1 Answer 1


1) Yes. Let $f: S^2 \to M$ be a representative of $A \in \pi_2(M)$. Then $$\int_{S^2} f^* c_1 = c_1(A).$$ You may think of Salamon's minimal Chern number as the minimum integer that you get when you evaluate the first Chern class on all spheres.

2) Those are two different definitions of monotone. One is that $\omega$ is proportional to $c_1$ on all of $H^2$, whereas Salamon's definition is only restricted to spheres, which is a weaker requirement.

However, in the case that $M$ is simply connected, the Hurewicz map is an isomorphism, so the two definitions are equivalent.

For Hamiltonian Floer homology, you wish control two things: energy, and bubbling, and it turns out that it suffices to have monotonicity only on $\pi_2$.

Given two holomorphic cylinders $u_1, u_2 \in \mathcal{M}(x,y)$, the Fredholm index difference is given by $$ \mu(u_1) - \mu(u_2) = 2 c_1(u_1 \# u_2)$$ i.e. by the first Chern class evaluated on the resulting torus. The Fredholm index $\mu(u)$ is computed by the difference Conley-Zehdner indices of the two ends of $u$ of a symplectic trivialization. But since the ends of the cylinders are contractible, you end up computing the Chern class on a sphere $A$. Monotonicity implies that the difference in energy is $E(u_1) - E(u_2) = \tau c_1(A)$, so the energy is constant on all moduli spaces between two orbits of a given dimension. This implies compactness (modulo bubbling) for index 1 components of $\mathcal{M}(x,y)$. For $\tau >0$, you kill a second bird, which is the bubbling.

  • $\begingroup$ Thanks a lot. If I may ask a few more questions: 1- what is the index difference, are you talking about the Conley-Zehnder index ? By the way, why using this instead of the Maslov index since we are dealing with loops? 2- I don’t understand your statement about the energy and the index difference. Could you be more precise ? If the index difference is $c_1(A)$, then how could the energy be constant ? 3- In case $\tau = 0$, then it is said that there might be infinitely many connecting orbits. Why, and how the introduction of Novikov rings bypasses this issue ? Thanks again for your help ! $\endgroup$
    – BrianT
    Apr 29, 2019 at 5:29
  • 1
    $\begingroup$ By index difference I mean $\mu(u_1) - \mu(u_2)$ for $u_1, u_2 \in \mathcal{M}(x,y)$, where $\mu(u) = \mu_{CZ}(x, D) - \mu_{CZ}(x, u \# D)$, and $D$ is a disk whose boundary is the orbit. $\endgroup$ Apr 29, 2019 at 16:55
  • $\begingroup$ So the index $\mu(u)$ depends on this choice of $D$ right ? $\endgroup$
    – BrianT
    Apr 29, 2019 at 17:09
  • $\begingroup$ It does not. I edited my post, for clearer details $\endgroup$ Apr 29, 2019 at 17:30
  • $\begingroup$ I am not sure if I understand the $\tau > 0$ part. If $\tau > 0$, then there are no holomorphic spheres with negative Chern number. But why are these a problem in the first place? $\endgroup$
    – warzasch
    Jan 21 at 17:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .