MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

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

A subset of a symplectic manifold is called strongly non-displaceable if it cannot be displaced by symplectomorphisms. A meridian in a $2$-torus is displaceable by a symplectomorphism, but not by a Hamiltonian diffeomorphism. I am reading "Rigid subsets of symplectic manifolds" by Entov and Polterovich and trying to understand Theorem 1.11 saying that the special fiber of the moment map is strongly non-displaceable. For an explicit example $S^2$ with moment map as the height function $H\colon S^2 \rightarrow [-1,1]$, the special fiber is the equator $X:= H^{-1}(0)$.

They actually prove that $X$ is superheavy, which means that $\zeta(K) \leq \sup_X K$ for all smooth function $K$ on $S^2$. Here symplectic quasi-state $\zeta$ is defined as $\zeta(K):= \lim_{r \rightarrow \infty}\frac{c(rK)}{r}$, where $c$ is the spectral invariant defined using filtered Floer homology. More explicitly, $$c(K):= \inf \{\alpha \mid PSS([S^2]) \in HF_*^{\alpha}(K)\}$$ where $PSS\colon QH_*(S^2) \rightarrow HF_*(S^2)$ is the PSS-isomorphism from quantum homology to Floer homology by Piunikhin-Salamon-Schwarz. The upper index $\alpha$ indicates that elements are coming from elements whose value of the action functional is less than $\alpha$.

According to their Proposition 4.3, if a function $K$ on $S^2$ is zero on a superheavy set, then $\zeta(K) = 0$. Because I am a beginner, I tried to compute this directly for $K=H$, but I got $\zeta(H)=1$. I must be wrong somewhere but I don't know where. The following is my computation.

I follow the sign convention of Entov and Polterovich. Take a non-integer $\epsilon>0$ so that the only orbits of period $1$ of Hamiltonian $K= \epsilon H$ are fixed points $N$ and $S$. Critical points of the action functional $$\mathcal{A}_K(\gamma, u)= \int_0^1 K(\gamma(t))dt - \int u^*\omega$$ are either $(N,u)$ or $(S,u)$, where $u\colon S^2 \rightarrow S^2$ is any map. If we compute the Conley-Zehnder index, for an integer $k$ with $k-1 < \epsilon < k$, we have $$\mu_{CZ}(N, u)= 1+k+4m, \quad \mu_{CZ}(S, u) = 1-k+4m$$ where $m$ is an integer representing the degree of $u$. So the differential is trivial and elements of $HF_*(K)$ are generated by $(N,u)$ and $(S,u)$. The image of the fundamental class by $PSS$ should be the north pole $(N, pt)$ and $\mathcal{A}_K(N,pt)= \epsilon$. We get $$c(\epsilon H) = \epsilon.$$ Taking $\epsilon \rightarrow \infty$, we have $\zeta(H) = 1$.

Where am I doing wrong?

[Added] the Conley-Zehnder index should be $$\mu_{CZ}(N, u)= 2k+4m, \quad \mu_{CZ}(S, u) = 2-2k+4m,$$ so odd degree elements don't appear.

share|cite|improve this question
up vote 1 down vote accepted

I don't think your computations of CZ indices are correct -- in particular, I am surprised you have a factor of $4$ showing up.

The key mistake, and the one that explains the key reason this theorem works, is that the image of the PSS map will be of a fixed grading (depending on your convention). The grading of $(N, pt)$ changes as $\epsilon \to \infty$ (i.e. as $k \to \infty$). To compensate for this, the image of the PSS map must be this North pole with a different capping. If the CZ indices are correct, this will give you an action that looks like $\epsilon - k$, which stays bounded, and thus when you divide this by $\epsilon$, the quantity goes to $0$ as desired.

share|cite|improve this answer
Thanks a lot. I didn't see that the image of 1 by PSS should change as $\epsilon$ varies. For CZ indices, the computation seems to be OK for me. 4 appears because it is twice the first Chern number of $S^2$, which is the difference for different cappings. – Hwang Mar 17 '14 at 0:38
Oh I am sorry, you are right. My computations of CZ indices are wrong. I should have put $2k-1$ instead of $k$. – Hwang Mar 17 '14 at 7:22
In actual fact, we were both wrong with our CZ indices originally :-) I'm glad this got you on the right track despite that. – Sam Lisi Mar 17 '14 at 13:12

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.