Partial decay on simple manifolds (Q750737)

Prescribing curvature in a class of complete Riemannian metrics is still a widely unexplored theme on noncompact manifolds. It was first considered by \textit{W.-M. Ni} [Indiana Univ. Math. J. 31, 493-529 (1982; Zbl 0496.35036)] for the scalar curvature, in the quasi-isometry class of \({\mathbb{R}}^ n\), \(n>2\), where it is required that the curvature function had partial decay of order \(p>2\) with respect to at least three orthonormal covectors (if L is a unit covector of \({\mathbb{R}}^ n\), a real function f has partial decay of order \(p\in (0,\infty)\) with respect to L provided \(| L(x) |^ p| f(x) |\) remains uniformly bounded on \(\{| x| >1\})\). The present article originated from an attempt [see the author, Commun. Partial Differ. Equations 14, 1127-1146 (1989; Zbl 0706.35045)] to understand the nature of Ni's partial decay condition. Simple manifolds are introduced as a natural framework: these are punctured compact N-manifolds, \(N>2\), equipped with charts at infinity ranging onto \(\{x\in {\mathbb{R}}^ N\), \(| x| >1\}\) and with Riemannian metrics which read uniformly equivalent to the standard euclidean metric in those charts. The first half of the paper is a self- contained account of linear elliptic theory on simple manifolds in the presence of partial decay. A new maximum principle is worth quoting from it: let u be subharmonic on \({\mathbb{R}}^ N\), \(N>2\), and have partial decay of order \(p>0\) with respect to some nonzero covector, then either \(u<0\) or \(u\equiv 0\). The second half of the paper is nonlinear. One part is on prescribing the scalar curvature in the quasi-isometry class of a simple manifold: nonexistence results and the Yamabe problem are discussed before Ni's theorem is extended. An extra section deals with the (analytically related) Lichnérowicz-York equation which occurs in the constraints problem of general relativity; simple solutions, more generally than asymptotically euclidean ones, may be of some physical interest. The last part is about the real elliptic Monge-Ampère equation \(\det (D^ 2u)=\exp (f)\), solved on \({\mathbb{R}}^ N\) using tools from the linear part, setting \(u=P+\phi\), P a quadratic polynomial satisfying \(\det (D^ 2P)\equiv 1\), f and \(\phi\) having partial decay with respect to a subspace taken, together with its orthogonal, globally invariant under the linear map \((D^ 2P)\). An equiaffine interpretation is given, and an application is drawn to the prescribed Ricci curvature problem posed in the Kähler class of \({\mathbb{C}}^ N\), \(N>2\); the result is amazingly similar to Ni's aforementioned one.