Filter methods for nonlinear problems

Project: Research

Project Details

Description

Study of the convergence properties (both theoretical and numerical) of algorithms based on filters for the solution of nonlinear problems. This notion, introduced by Fletcher and Leyffer, frees the algorithm designer from the monotonicity constraints that are common in classical approaches. It also suggests practical new methods that appear to be very efficient for solving optimization problems as well as systems of algebraic equations.
StatusFinished
Effective start/end date2/02/0231/08/11

Keywords

  • numerical experiment
  • algorithm
  • Optimization
  • convergence

Research Output

How much do approximate derivatives hurt filter methods?

Sainvitu, C. & Toint, P., 1 Jul 2009, In : RAIRO - Operations Research. 43, 3, p. 309-329 21 p.

Research output: Contribution to journalArticle

File
  • 12 Downloads (Pure)

    A brief history of filter methods

    Fletcher, R., Leyffer, S. & Toint, P., 2007, In : SIAG/ OPT Views-and-News. 18, 1, p. 2-12 11 p.

    Research output: Contribution to journalArticle

    A filter-trust-region method for simple-bound constrained optimization

    Sainvitu, C. & Toint, P. L., 1 Oct 2007, In : Optimization Methods and Software. 22, 5, p. 835-848 14 p.

    Research output: Contribution to journalArticle

    Open Access
    File
  • 55 Downloads (Pure)

    Projects