LANCELOT, a package for the solution of large-scale nonlinear optimization problems

Project: Research

Project Details

Description

The LANCELOT project covers the development of a mathematical programming package, with special emphasis on the solution of large-scale nonlinear and/or nonconvex optimization problems. It features a theoretical part (convergence analysis) and a computer science aspect (package implementation and validation).

The package is based on the application of a trust-region method for unconstrained problems or with bound constraints only. The general inequality constraints are transformed into equality constraints by introducing slack variables, and all equality constraints are handled via an augmented Lagrangian technique. The package reads problems description in a standard input format (SIF).

LANCELOT was awarded the Beale-Orchatd-Hayes prize in 1994.
AcronymLANCELOT
StatusFinished
Effective start/end date1/09/871/09/00

Keywords

  • nonlinear optimization
  • mathematical programming
  • large scale problems
  • convergence analysis