Torus actions on small blowups of \(\mathbb C\mathbb P^2\) (Q1047586)

An action of a torus \(T^k\) on a symplectic manifold \((M,\omega)\) is Hamiltonian if there is a moment map \[ \Phi: M \to (\mathfrak t^k)^* \cong \mathbb R^k \] with \(d\Phi_j = -\iota(\xi_j)\omega\), where \(\Phi_j\) is the \(j\)-th component of \(\Phi\), \(\iota(-)\) is interior multiplication and \(\xi_j\) is a (Hamiltonian) vector field. It is a famous result that the image of \(\Phi\) is a convex polytope. If \(\text{dim}(M) = 2k\), then Delzant showed that \((M,\omega)\) may be reconstructed just from knowledge of the polytope; hence, the polytope determines the manifold and the action. These manifolds are the symplectic versions of \textit{toric manifolds} and, in some sense, their study is completely combinatorial. Even the symplectic procedure known as blowing up a point corresponds to a simple cutting-a-corner procedure on the polytope. In dimension \(4\), it is known that a toric manifold arises either from equivariantly blowing up \(\mathbb C \mathbb P^2\) a number of times or blowing up a Hirzebruch surface a number of times. As the author points out, however, confronted by a specific manifold, it is unclear how to know whether it \textit{is} such a blowup. The main result of the present paper concerns manifolds \((M_k,\omega_\epsilon)\) obtained by doing \(k\) simultaneous blowups on \(\mathbb C \mathbb P^2\) of equal sizes \(\epsilon >0\). Then, if \((M_k,\omega_\epsilon)\) is a toric manifold \((M_\Delta,\omega_\Delta)\) for a Delzant polytope \(\Delta\) and \(\epsilon \leq (3k\,2^{2k})^{-1}\), then \((M_\Delta,\omega_\Delta)\) is obtained as \(k\) equivariant blowups of \(\mathbb C\mathbb P^2\) of size \(\epsilon\). This is saying that there is a unique way for \((M_k,\omega_\epsilon)\) to arise as a toric manifold. The proof is by induction based on showing that some exceptional divisor class is represented by an embedded invariant symplectic sphere and that the equivariant blowdown along this sphere produces \((M_{k-1},\omega_\epsilon)\). Then, it is possible to work down to \(\mathbb C \mathbb P^2\) with its standard toric action. Now, a result of the author and Karshon says that \(\mathbb C\mathbb P^2\) does not admit four or more toric blowups of equal sizes. Hence, \((M_k,\omega_\epsilon)\) with \(\epsilon\) as above, admits a toric action if and only if \(k \leq 3\).