Models for combining vector estimators of random parameters (Q809508)

From MaRDI portal





scientific article
Language Label Description Also known as
English
Models for combining vector estimators of random parameters
scientific article

    Statements

    Models for combining vector estimators of random parameters (English)
    0 references
    0 references
    0 references
    1990
    0 references
    The authors, \textit{D. Young} and \textit{J. Igwe} [ibid. 12, No.12, 1627-1642 (1989; Zbl 0708.62044)] reviewed and extended the theory of modelling the combining of vector estimators of a non-random parameter to attain a better estimator. It is our purpose in this paper to consider the problem of combining vector valued information and develop models for estimating a value of a random parameter by optimally combining the available information (in the form of data vectors, vectors of expert opinion, etc.) to obtain a better estimator of the random vector, X. In the early part of this paper we develop a calculus of variation for solving minimization problems when the function being minimized is a positive semi-definite matrix valued function of a matrix variable. The problem is to search through an admissible set of matrices that one which minimizes the positive semi-definite function with respect to the usual positive semi-definite (psd) ordering criteria. This criterion is sometimes called the Loewner partial ordering, and is simply that \(M_ 1<M_ 2\) if \(M_ 1\), \(M_ 2\) are both psd and \(M_ 2-M_ 1\) is psd and non-zero.
    0 references
    combining of vector estimators
    0 references
    combining vector valued information
    0 references
    data vectors
    0 references
    vectors of expert opinion
    0 references
    calculus of variation
    0 references
    minimization problems
    0 references
    positive semi-definite matrix valued function
    0 references
    matrix variable
    0 references
    admissible set of matrices
    0 references
    Loewner partial ordering
    0 references

    Identifiers