Qubist is a powerful MATLAB® toolbox for massive-scale parameter search and global optimization, designed for problems in science, engineering and other technical disciplines. All of the Qubist optimizers are designed for optimization problems with a single objective or multiple objectives, and their capabilities span the range of problems encountered in these fields. The two most powerful “front-line” optimizers in the Qubist package (Ferret and Locust) are designed with a built-in parallel computing system that does not require the MATLAB® parallel computing toolbox. A few clicks on the graphical Qubist node manager turns a standard non-parallel calculation into a parallel run that can take advantage of today's multi-core computers and clusters. Built-in visualization and analysis tools help you to explore your optimal solution sets and understand your results. Qubist provides a complete, integrated software solution with sufficient power and capability to tackle even the toughest parameter search and optimization problems.
The Qubist platform contains five complementary optimization algorithms:
a parallel multi-objective genetic algorithm with sophisticated machine-learning features
a parallel multi-objective particle swarm algorithm
a “Simple Approach to a Multi-Objective Simplex Algorithm”
an innovative simulated anneal/genetic algorithm hybrid code
a unique experimental “Semi-Global Solution Spray” algorithm.
All Qubist optimizers are all interchangeable. Set up your problem for one of them, and effortlessly discover how other optimization techniques perform on your problem.