Skinny is a derivative-free algorithm based on pseudo-random direct
searches for solving constrained problems. The main feature of this
algorithm is the ability to handle equality constraints. The objective
function must be defined outside the feasible region, as the algorithm
works by enlarging the feasible set and iteratively reducing it.
This version was used to produce the numerical results presented in
- J. M. Martínez and F. N. C. Sobral, "Constrained Derivative-Free Optimization on Thin Domains", Journal of Global Optimization, 56(3), pp. 1217-1232, 2013. Accepted version Submitted version.
Instructions for building and extending Skinny can be found in the
author's
web page. Complete results
can be downloaded here.
The present algorithm was designed to work with Algencan version
2.4. Modifications
to work with version 3.1 are under development in dev branch.