Testing results

We constantly test the efficiency of our optimization algorithms using various classes of test functions and optimization problems of complex real-life systems. Based on the results of these tests we may conclude that IOSO Technology Algorithms:

• are comparable in performance with well known highly efficient optimization algorithms for smooth single extremum function (comparison);

• have distinct advantages in comparison to other algorithms while considering non-smooth function, which are more frequently encountered in real-life problems;

• could be efficiently used to solve problems with various types of objective functions: smooth, non-differentiable, stochastic, dependent on mixed variables, the objective functions that could potentially have the regions where they can not be evaluated;

• have good global properties and in most cases allow the users to solve the problems with multiple optima;

• are not sensitive to “noise” when solving stochastic optimization problems;

• have unique capabilities when applied to multiobjective, multilevel, and parallel optimization.
  

The results of some the test tasks performed by WaveFront (Japan)  (pdf, 1Mb) 

© Sigma Technology 2001,2009.