Singular Sturm-Liouville theory on manifolds (Q5955181)

The authors study the spectral properties of Schrödinger-type operators \(L=-\Delta_g +a(x)\) on a compact Riemannian manifold \((M,g)\), where \(a(x)\) is a real-valued potential defined and continuous, but not necessarily bounded, on \(\widehat M=M-\sigma\), where \(\Sigma\subseteq M\) is a set of measure zero. To be more precise, the paper addresses the problem of determining conditions, under which such operators enjoy well known principal properties of formally self-adjoint elliptic differential operators (with positive definite principal symbol) on closed manifolds, being essentially self-adjoint and Sturm-Liouville (i.e. possessing discrete spectrum tending to infinity), and extends to this more general setting the theory of elliptic boundary value problems. In order to give a more precise presentation of the results of the paper let us, after the authors, introduce the following terminology. Let \(q(u)\equiv \langle Lu,u\rangle\), and \(Q(u)\equiv \int_M(|\nabla u|^2+a(x)|u|^2) dV_g\). A self-adjoint extension of \(L\) is a \(B\)-extension if \(q(u)=Q(u)\); if \(\Sigma=\partial M\) this corresponds to imposing certain boundary conditions (e.g. Dirichlet or Neumann, depending on the core domain of \(L)\); essentially Dirichlet if the Dirichlet extension of \(L\) (by definition the Friedrichs extension of \(L\) with core domain \(C^\infty_0(\widehat M)\), smooth functions of \(\widehat M\) compact support; it is clearly a \(B\)-extension) is the only \(B\)-extension; and singular Sturm-Liouville if it is essentially Dirichlet and the Dirichlet extension is Sturm-Liouville. Principal results of the paper can be, very roughly, summarized as follows: If \(\Sigma\) is sufficiently regular (a smooth closed submanifold of codimension \(\geq 1\) or, more generally, a stratified subset of iterated cone-edge type- roughly a generalization of manifolds with cusps and edges) and \(a(x)\) does not escape to \(-\infty\) to quickly as \(x\to\Sigma\), then the Dirichlet extension of \(L\) is Sturm-Liouville. Moreover, under some more restrictive conditions on \(a(x)\), \(L\) is singular Sturm-Liouville, and under still more restrictive conditions on \(a(x)\), \(L\) is essentially self-adjoint, and then the operator's spectrum is insensitive to the ``boundary conditions''. Proofs of these results are based on certain generalization of standard arguments used to prove corresponding results in the absence of \(\Sigma\), and the principal tools are some ``Gårding-type'' inequalities, elliptic regularity, Rellich lemma, and continuity properties of \(L\) as a map between some weighted Sobolev spaces. For the proofs of the main theorems of the paper a careful analysis of the behavior of functions near \(\Sigma\) is required, and this is done, roughly speaking, by expressing \(L\) in ``normal'' coordinates along \(\Sigma\) (a generalization of normal coordinates along a submanifold) and using beautiful ``Hardy-type'' inequalities. The paper is very nicely written. It starts with more elementary results (for a smooth submanifold \(\Sigma)\) and then proceeds with generalizations to more complicated singular sets \(\Sigma\).