On a spectral flow formula for the homological index (Q904040)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On a spectral flow formula for the homological index
scientific article

    Statements

    On a spectral flow formula for the homological index (English)
    0 references
    0 references
    0 references
    0 references
    15 January 2016
    0 references
    This very interesting paper under review is concerned with the question of the equality of the (Fredholm) index with the spectral flow. This question goes back to Atiyah, Patodi and Singer [\textit{M. F. Atiyah} et al., Math. Proc. Camb. Philos. Soc. 79, 71--99 (1976; Zbl 0325.58015)]. The equality was established in [\textit{J. Robbin} and \textit{D. Salamon}, Bull. Lond. Math. Soc. 27, No. 1, 1--33 (1995; Zbl 0859.58025)] for certain families of self-adjoint unbounded operators on compact manifolds. An extension to operators which possess essential spectra was given in [\textit{F. Gesztesy} et al., Adv. Math. 227, No. 1, 319--420 (2011; Zbl 1220.47017)], but the results concern only self-adjoint operators which differ by a trace-class perturbation. This excludes differential operators on non-compact manifolds (albeit pseudodifferential operators are OK). The current paper under review establishes the equality for operators on non-compact manifolds, which do not necessarily have the Fredholm property. In order to achieve this, the authors work in an abstract setting, and they develop novel resolvent and adiabatic approximation methods. However, in the introduction they explain what their results amount to, in the context of [Zbl 0859.58025]: If \(M\) is a closed, odd-dimensional manifold and \(\{B_s\}_{s \in S^1}\) is a smooth family of self-adjoint operators, one considers the Dirac operator \(D = \partial_s + B_s\) on \(M \times S^1\). Taking the adiabatic limit, and using a certain trace formula for the spectral flow, established by Wojciechowski and independently by Getzler, the equality \(index(D) = sf\{B_s\}\) amounts to the trace formula: \[ Tr_{M \times S^1}((1+D^{\ast}D)^{-r}+(1+DD^{\ast})^{-r}) = C_{r + 1/2}\int Tr_M(\frac{dB(s)}{ds}(1 + B(s)^2)^{-(r+1/2)})ds \] What the authors prove in this context is that, although the none of the operators \((1+D^{\ast}D)^{-r}\), \((1+DD^{\ast})^{-r}\), \((1 + B(s)^2)^{-(r+1/2)}\) is trace-class, the particular combination of operators that appear in the trace formula remain trace class and the formula holds in much greater generality.
    0 references
    0 references
    0 references
    0 references
    0 references
    index theory
    0 references
    spectral flow
    0 references
    perturbations
    0 references
    0 references
    0 references