\(L^ 2\)-index theorem for elliptic differential boundary problems. (Q5937861): Difference between revisions

From MaRDI portal
Set OpenAlex properties.
Importer (talk | contribs)
Changed an Item
 
Property / arXiv ID
 
Property / arXiv ID: math/9810133 / rank
 
Normal rank

Latest revision as of 11:18, 18 April 2024

scientific article; zbMATH DE number 1621008
Language Label Description Also known as
English
\(L^ 2\)-index theorem for elliptic differential boundary problems.
scientific article; zbMATH DE number 1621008

    Statements

    \(L^ 2\)-index theorem for elliptic differential boundary problems. (English)
    0 references
    0 references
    17 July 2001
    0 references
    The main result of the paper is an extension of the Atiyah \(L^2\)-index theorem for regular coverings of a compact manifold to differential elliptic boundary value problems. In order to put things in context let us consider the following situation: Let \(M\) be a compact manifold with non-vacuous boundary \(\partial M\) and let \(\widetilde M\) be a normal covering of \(M\) with deck transformation group \(\Gamma\). Let \(P= (A,T): C^\infty_0(E)\to C^\infty_0(F)\oplus C^\infty_0(Y)\) be a differential elliptic boundary value problem (\(E\) and \(F\) are vector bundles over \(M\) and \(Y\) is a vector bundle over \(\partial M\)) and let \(\widetilde P\) be the lift of \(P\). Then \(\widetilde P\) commutes with the action of \(\Gamma\), and the von Neumann index of \(\widetilde P\) is defined using the natural structure of the Hilbert module over the von Neumann group algebra \(N(\Gamma)\) on both \(\ker(\widetilde P)\) and \(\text{co\,ker}(\widetilde P)\) and the trace function \(\text{Sp}_\Gamma\) for \(\Gamma\)-equivariant operators: \[ \text{index}_\Gamma(\widetilde P)= \dim_\Gamma\ker(\widetilde P)-\dim_\Gamma\text{coker}(\widetilde P)= \text{Sp}_\Gamma(H_0)- \text{Sp}_\Gamma(H_1), \] where \(H_0\) resp. \(H_1\) is the projection on \(\ker(\widetilde P)\) resp. \(\text{coker}(\widetilde P)\). The main theorem of the paper states that \(\text{index}_\Gamma(\widetilde P)= \text{index}(\widetilde P)\) similarly as in the case of manifolds with empty boundary. The proof of the theorem is analogous to the original proof of Atiyah: It is shown that \(\text{index}_\Gamma(\widetilde P)\) can be computed from any parametrix of \(P\) with support sufficiently close to the diagonal, i.e. given any such a parametrix \(Q\) with \(PQ=1- S_1\) and \(QP =1- S_0\) all the operators involved in these equations can be lifted to \(\widetilde M\) and then \(\text{index}_\Gamma(\widetilde P)= \text{Sp}_\Gamma(\widetilde S_0)- \text{Sp}_\Gamma(\widetilde S_1)\); moreover for lifts of operators with smooth kernels it is shown that \(\text{Sp}_\Gamma(\widetilde S)= \text{Sp}(S)\), which completes the proof. The paper contains also some other interesting results e.g. Rellich lemma in the setting of Sobolev spaces with group actions and some extensions of the well known properties of the trace of trace-class operators to the case of \(\Gamma\)-traces.
    0 references
    index of boundary value problem
    0 references
    von Neumann dimension
    0 references
    Boutet de Monvel calculus
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references