On the topology of positively curved 4-manifolds with symmetry (Q1121538)

The authors pose the following question. Which compact 4-manifolds admit positively curved Riemannian metrics with at least one infinitesimal isometry, in other words, a nontrivial Killing field? The main result of their paper answers this question. Theorem 1. Let \(M^ 4\) be closed orientable and positively curved. If \(M^ 4\) has a nontrivial Killing vector field, then \(M^ 4\) is homeomorphic to \(S^ 4\) or \(CP^ 2\). Corollary 1. Let \(M^ 4\) be closed, nonorientable and positively curved. If \(M^ 4\) has a nontrivial Killing vector field, then \(M^ 4\) is two-fold covered by \(S^ 4\). Corollary 2. \(S^ 2\times S^ 2\) does not admit a positively curved Riemannian metric with a nontrivial Killing field. The proof runs as follows. Technical speaking, the existence of a nontrivial Killing vector field on a compact Riemannian manifold M is equivalent to the existence of a nontrivial \(S^ 1\)-action on M. Let \(F(S^ 1,M)\) be the fixed point set of such an \(S^ 1\)-action on M. Proving Theorem 1, the authors show \[ F(S^1,M) = \begin{cases} \chi(M) &\text{ isolated points,}\\ S^2 &(\chi(M)-2 \text{ isolated points}). \end{cases} \] Furthermore, \(\chi (F(S^ 1,M))=\chi (M)\). Using this, the authors show \(\chi(M)\leq 3\). Then the result follows from Synge's theorem and Freedman's results.