Derivative free numerical algorithms for optimization

Project: Research

Project Details

Description

The project's purpose is to design new algorithms for numerical optimization, that do not use the problem function's derivatives nor attempt to explicitly approximative them. The resulting algorithms should be relatively insensitive to noise on the function values.
AcronymDFO
StatusActive
Effective start/end date1/03/94 → …

Keywords

  • noisy functions
  • derivative free methods
  • numerical optimization
  • multivariate interpolation
  • trust regions
  • simulation
  • algorithms

Research Output

File
  • 7 Downloads (Pure)
    Open Access
    File
  • A derivative-free trust-funnel method for equality-constrained nonlinear optimization

    Rodrigues Sampaio, P. & Toint, P. L., 1 May 2015, In : Computational Optimization and Applications. 61, 1, p. 25-49 25 p.

    Research output: Contribution to journalArticle

  • Projects

    Activities

    Università di Firenze

    Philippe Toint (Visiting researcher)
    15 Sep 201815 Oct 2018

    Activity: Visiting an external institution typesVisiting an external academic institution

    Università di Firenze

    Philippe Toint (Visiting researcher)
    1 Oct 201731 Oct 2017

    Activity: Visiting an external institution typesVisiting an external academic institution

    BFO: a versatile and trainable Brute Force Optimizer

    Philippe Toint (Invited speaker)
    11 Mar 2016

    Activity: Talk or presentation typesOral presentation