After a long winter, OOL 0.2.0 is on the road!
This release includes the Spectral Projected Gradient method (SPG), which is suited to large-scale problems due to its low memory requirements.
The PGRAD and GENCAN were validated with the CUTEr test problem suite, ensuring the reliability of this implemantation.
Most of the OOL Reference Manual was rewriten and/or improved. Besides, two new chapters ("Quick Start" and "Brief Theoretical Introduction") were included into the OOL Reference Manual.... read more
The Nonmonotone Spectral Gradient Method (SPG) is the third method implemented in the OOL. This method is suited for minimization of differentiable functions subjected to bound constraints. It is a competitive algorithm with low memory requirements, with the spectral steplength, a clever choice for the steplength which really improves the convergence speed.
It is available in the CVS root. A new realease is coming soon.
Araucaria Foundation, the State of Parana Research Foundation, has approved a one-year term project to financially supports the OOL development.