This package is a Python re-implementation of the GLIS algorithm for derivative-free global optimization using surrogate functions developed by A. Bemporad. You can find the original paper and implementation here.
This re-implementation was intended to be used for parameter tuning for the autonomous driving controllers (primary Stanley and MPC based) created at the EPFL Racing Team.
The main features are the same as the GLIS implementation in the version 2.4:
- global minimization of a general function
- support for box, linear and/or nonlinear constraints
We have not (yet) implemented the GLISp and C-GLIS(p) variants since they were not of primary interest for our application.