A working-set framework for sequential convex approximation methods

    Research output: Book/ReportReportResearch

    Abstract

    We present an active-set algorithmic framework intended as an extension to existing implementations of sequential convex approximation methods for solving nonlinear inequality constrained programs. The framework is independent of the choice of approximations and the stabilization technique used to guarantee global convergence of the method. The algorithm works directly on the nonlinear constraints in the convex sub-problems and solves a sequence of relaxations of the current sub-problem. The algorithm terminates with the optimal solution to the sub-problem after solving a finite number of relaxations.
    Original languageEnglish
    Publication statusPublished - 2008
    SeriesMat-Report
    Number2008-03

    Bibliographical note

    The report is available upon request to Mathias Stolpe (M.Stolpe@mat.dtu.dk).

    Keywords

    • Sequential convex approximation methods
    • Active-set method
    • Inequality constrained optimization

    Cite this