Nonsmooth exclusion test for finding all solutions of nonlinear equations (Q616162)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Nonsmooth exclusion test for finding all solutions of nonlinear equations
scientific article

    Statements

    Nonsmooth exclusion test for finding all solutions of nonlinear equations (English)
    0 references
    0 references
    0 references
    7 January 2011
    0 references
    This is a great paper about Newton-methods for finding all real solutions of nonlinear equations in a connected compact subset of \({\mathbb R^n}\). After a comprehensive review of the existing Newton methods, quasi-Newton type methods, interval Newton-type methods, the authors reformulate the root finding problem into a global optimization problem and introduce the McCormick non-smooth convex relaxation and its properties for developing a branch-and-bound algorithm. Furthermore, they establish an inclusion relation between the solution set of the relaxed problem and the solution set of the original non-convex problem and propose a process to generate automatically starting points for a Newton-type method. Based on the componentwise Krawczyk operator and interval Newton operator with Gauss-Seidel based root inclusive and exclusive tests, they refine the variables bounds and restrictive monotonically tests in the proposed branch-and-bound algorithm to ensure uniqueness of a solution in the inclusion and convergence of the iteration. Efficiency and performance of the proposed algorithm and its various variations are discussed and demonstrated extensively with numerical examples.
    0 references
    0 references
    systems of nonlinear equations
    0 references
    global optimization
    0 references
    interval analysis
    0 references
    convex relaxation
    0 references
    Newton methods
    0 references
    quasi-Newton type methods
    0 references
    interval Newton-type methods
    0 references
    McCormick's non-smooth convex relaxation
    0 references
    branch-and-bound algorithm
    0 references
    non-convex problem
    0 references
    Krawczyk operator
    0 references
    Gauss-Seidel
    0 references
    inclusion
    0 references
    convergence
    0 references
    algorithm
    0 references
    numerical examples
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers

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