Population-based multi-level algorithms for solving single- and multi-objective global optimization problems

Principial Investigator: Ewa Gajda-Zgórska,

Catalogue number: DEC-2012/05/N/ST6/03433,

Scientific goal of the project:

Solving multi-objective global optimization problems and inverse problems is often a difficult and costly task. On the other hand, these types of problems are very important, because they appear in key areas of technology, business and medicine (i.e. design of chemical compounds, optimum design, flaw detection, search and exploitation of natural resources).
The aim of the project is to create stochastic algorithms which are capable of solving these types of problems efficiently. We will formulate a strategy which couples population-based algorithms with post-processing of the received samples by cluster analysis methods to obtain maximum information about solutions of the problem: additional knowledge about the location of the solution, the size of basins of attraction, connectivity of the Pareto set and its neighbourhood, etc.