Two applications of nets (Q2346171)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Two applications of nets
scientific article

    Statements

    Two applications of nets (English)
    0 references
    0 references
    0 references
    29 May 2015
    0 references
    The authors develop two applications of nets to classical problems of functional analysis. The common point is that both of them use nets for describing properties of spaces and operators in a (let us say) non-standard way. The first one gives a description of Bochner integrable functions. The experienced functional analyst is used to assume that nets and countably additive structures do not fit very well. Surprisingly enough, in this case the authors show that this point of view is useful for extending the notion of Bochner integrability to the case of locally convex-valued measurable functions. They do that following the original definition of integrable function given by Bochner as limit of sequences of simple functions -- in this case with nets --, instead of the Gelfand-Pettis approach based on the action of functionals of the dual on the corresponding integrals, sometimes called barycentric formulas. The notion of a Bochner approximable function is introduced here: A function as above having values in a locally convex space \(V\) is Bochner approximable if it can be approximated by simple functions for each continuous seminorm. Using then nets of simple functions, they prove for instance in Theorem 1.16 that, if \(V\) is complete, then every Bochner approximable \(V\)-valued function is Bochner integrable, and if \(V\) is quasi-complete, then every bounded Bochner approximable function in \(V\) is Bochner integrable. In the second part of the paper, the authors analyze a new version of trace class operators on Hilbert spaces, which are called to be of determinant class. A bounded operator \(A\) on a Hilbert space \(H\) is of determinant class if the limit \[ \det(A)= \lim_F \det (A_F) \] exists in \(\mathbb C\) and is different from \(0\), where the limit is computed over the net \(\det(A_F)\) indexed by the directed set of all finite-dimensional subspaces of \(H\). The authors prove in this case that if \(A\) is a normal operator of determinant class, then \(T=1 - A\) is trace class and the determinant \(\det(A)=\det(1-T)\) is a Fredholm determinant. Some open questions are also presented.
    0 references
    Bochner integral
    0 references
    net
    0 references
    determinant
    0 references
    trace class operator
    0 references

    Identifiers

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