The GALAHAD project was born from the success of LANCELOT, a previously released software package for large-scale nonlinear optimization. As for LANCELOT, the aim is to transform state-of-the-art in numerical optimization algorithms (both with and without constraints) into high quality software tools that can be widely used by the research community and industry.
GALAHAD's objective is to build a library of nonlinear optimization packages, using an approach where the tools are vieewed as complementary. GALAHAD contains, for instance, tools for pre-processing, a powerful nonlinear equation and nonlinear least-squares solver as well as two state-of-the-art quadratic programming solvers. It also contains an upgraded version of the LANCELOT package.
The project emphasis in on a complete vision of this solution process, including the design of algorithms, their theoretical justification via a fully fledged convergence analysis, a high quality computer implementation and a very torough numerical testing procedure, notably within acknowledged testing environments such as CUTEr.