Complete prolongation for infinitesimal automorphisms on almost complex manifolds (Q2268159)

Let \((M,J)\) be an almost complex manifold. An automorphism of \(M\) is a smooth map \(f: M \rightarrow M\) such that \(J \circ df = df \circ J\). An infinitesimal automorphism of \(M\) is a vector field \(X\) such that \(\mathcal{L}_XJ=0.\) The paper under review is devoted to the study of the germs of infinitesimal automorphisms. The set of germs is closed under Lie bracket and is called \textit{symmetry algebra} in the paper. In the integrable case infinitesimal automorphisms coincide with (real parts of) holomorphic vector fields, hence the symmetry algebra is an infinite dimensional vector space over the complex numbers. The authors give examples of nonintegrable almost complex manifold both with finite dimensional and with infinite dimensional symmetry algebra. For \(x\in M\), denote by \(\mathcal{V}_x\) the image of the Nijenhuis tensor \(N_{J,x} : T_xM \times T_xM \rightarrow T_xM\). \(\mathcal{V}_x\) is a \(J\)-invariant subspace of \(T_xM\). If \(\dim_\mathbb{R} M=4\), then \(\mathcal{V}_x\) is either trivial or of real dimension 2 (Prop. 1.4). If \(N_J \neq 0\) on all \(M\), then \(\mathcal{V} \) is a complex line bundle, called the \textit{torsion bundle}. The torsion bundle is \textit{non-degenerate} at \(x\in M\) if there is a local section \(W\) of \(\mathcal{V}\) such that \([W, JW] (x) \not \in \mathcal{V}_x\) (Def. 1.5). The main result (Thm. 1.6) asserts that if \(M\) is an almost complex manifold with \(\dim_\mathbb{R} M=4\), \(N_J\neq 0\) on \(M\) and \(\mathcal{V}\) is nondegenerate at \(x\), then the symmetry algebra at \(x\) has real dimension at most 4. The proof is based on the tecnique of prolongations. It is shown explicitely that (under the above hypotheses) the equation for infinitesimal automorphisms admits a complete prolongation of order 1. It follows that any germ is determined by its 0-jet. The needed material on prolongations is recalled in \S 1.3 and the proof is contained in \S 2. In \S 3 there is an example of an almost complex manifold \(M\) with \(\dim_\mathbb{R} M=4\), \(N_J\neq 0\) on \(M\), nondegenerate torsion bundle and symmetry algebra of dimension 4. This shows that 4 is the optimal bound under the assumptions of the theorem. In \S 4 there is an example of an almost complex manifold \(M\) with \(\dim_\mathbb{R} M=4\), \(N_J\neq 0\) on \(M\) and infinite dimensional symmetry algebra. This shows that the nondegeneracy assumption is essential.