An acceleration scheme for solving convex feasibility problems using incomplete projection algorithms (Q596679)

From MaRDI portal
scientific article
Language Label Description Also known as
English
An acceleration scheme for solving convex feasibility problems using incomplete projection algorithms
scientific article

    Statements

    An acceleration scheme for solving convex feasibility problems using incomplete projection algorithms (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    10 August 2004
    0 references
    An accelerated iterative projection method is presented for solving systems of linear inequalities. This method is an extension of the projection methods for solving systems of linear equations given by \textit{H. Scolnik, N. Echebest, M. T. Guardarucci}, and \textit{M. C. Vacchino} [Stud. Comput. Math. 8, 457--471 (2001; Zbl 1003.65027); Ann. Oper. Res. 117, 95--115 (2002; Zbl 1023.65026)]. The general scheme is similar to the incomplete projection algorithm, and therefore it is very convenient for parallel processing. The idea is that at the current iterate the set of violated constraints is splitted into subsets or blocks, in such a way that the required incomplete projection is obtained by combining exact projection onto simple convex sets. The new iterate is defined by the projection of the current iterate onto a separating hyperplane. The advantages of this approach are illustrated both theoretically and numerically.
    0 references
    aggregated projection methods
    0 references
    convergence acceleration
    0 references
    parallel computation
    0 references
    numerical examples
    0 references
    systems of linear inequalities
    0 references
    incomplete projection algorithm
    0 references
    0 references

    Identifiers

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