I'm confused about the relationship between strong admissibility and weak admissibility for pointed diagrams in Heegaard Floer theory. For reference, here are Ozsváth-Szàbo's original definitions:
A pointed Heegaard diagram is called strongly admissible for the ${Spin}^c$ structure $\mathfrak{s}$ if for every nontrivial periodic domain $\mathcal{D}$ with $$\langle c_1(\mathfrak{s}),H(\mathcal{D})\rangle=2n \geq 0,$$ $\mathcal{D}$ has some coefficient $>n$. A pointed Heegaard diagram is called weakly admissible for $\mathfrak{s}$ if for each nontrivial periodic domain $\mathcal{D}$ with $$\langle c_1(\mathfrak{s}),H(\mathcal{D})\rangle =0,$$ $\mathcal{D}$ has both positive and negative coefficients.
Here's a concrete case that puzzles me: If $c_1(\mathfrak{s})$ is torsion, then it evaluates to zero on every homology class. In that case, it seems that a diagram is
- strongly admissible for $\mathfrak{s}$ if every nontrivial periodic domain $\mathcal{D}$ has some positive coefficient, and
- weakly admissible for $\mathfrak{s}$ if every nontrivial periodic domain $\mathcal{D}$ has both positive and negative coefficients.
If these conditions are to coincide for $c_1(\mathfrak{s})$ torsion (or at least for "strong" to be stronger), then it seems like any nontrivial periodic domain with positive coefficients must also have negative coefficients. Is this true?
Edit: For the non-torsion case... Suppose $\mathcal{D}$ is a nontrivial periodic domain. Then $$H(\mathcal{D}):=[\mathcal{D}+\ell \cdot \Sigma+n\cdot D^2],$$ where $n\cdot D^2$ are the attaching disks used to cap off $\mathcal{D}+\ell \cdot \Sigma$. Since $[\Sigma]=0$ in $H_2(Y)$, the class of $H(\mathcal{D})$ equals $[\mathcal{D}+n \cdot D^2]$ and therefore $$H(-\mathcal{D})=[-\mathcal{D}-n \cdot D^2]=-H(\mathcal{D})$$ since the orientation on the boundary components of $-\mathcal{D}$ being reversed requires the attaching disks to have their orientation reversed, too. Then if $\langle c_1(\mathfrak{s}),H(\mathcal{D})\rangle = 2n \geq 0$, we have $$\langle c_1(\mathfrak{s}),H(-\mathcal{D})\rangle = \langle c_1(\mathfrak{s}), -H(\mathcal{D})\rangle = - 2n \leq 0.$$ If $n \neq 0$, then strong admissibility appears to impose no constraints on the coefficients of $-\mathcal{D}$.