Kreĭn-like extensions and the lower boundedness problem for elliptic operators (Q652458): Difference between revisions

The paper is devoted to the study of lower boundedness and Kreĭn-like extensions of elliptic operators. For realizations \({\widetilde A}\) of strongly elliptic \(2m\)-order differential operators \(A\) (on bounded smooth domains \(\Omega \subset {\mathbb R}^n\)), it is known that the realizations are lower bounded if and only if the associated operators \(T\) (determining the boundary conditions) are lower bounded (and this may be reached by using the fact that the inverse of the Dirichlet realization \(A_\gamma\) is compact). When dealing with the application of the abstract theory to the case where \(\Omega\) is an exterior domain (the complement of a compact smooth set in \({\mathbb R}^n\)), the Dirichlet solution operator \(A_\gamma^{-1}\) is not compact, and it has been an open question whether one always could conclude from lower boundedness of \(T\) the lower boundedness of \({\widetilde A}\). In the present work, the author answers this question positively (by using that the boundary is compact and using principles and results for pseudodifferential boundary operators). Consequent results on Gårding-type inequalities are also exposed. Moreover, associated with the lower boundedness questions, generalizations of the Kreĭn--von Neumann extension arise in a natural way. In the present work, the author considers certain Kreĭn-like extensions \(A_a\) for \(a\in{\mathbb R}\). In the case of a bounded domain, their essential spectrum is resumed to the single point \(a\), and a pertinent question is to look for the asymptotic behavior of the eigenvalue sequence converging to \(+\infty \). This is analysed in this paper by showing that the number \(N_+(t;A_a)\) of eigenvalues in \([r,t]\) (for some \(r>a\)) has the asymptotic behavior \[ N_+(t;A_a)-c_A\, t^{n/2m}=O(t^{(n-1+\varepsilon )/2m})\;\;\;\; \text{for}\;\;\;\; t\to\infty, \] for any \(\varepsilon >0\) (and with the same constant \(c_A\) as for the Dirichlet problem).