Shylo P. Algorithms for solving separate classes of large scale discrete optimization problems

The thesis establishes new computational tools for solving the following classes of large scale discrete optimization problems: set covering problem, quadratic assignment problem, and the problem of finding the size of the largest error-correcting code for a z-channel, which reduces to the problem of finding the maximal independent set in a graph. The solution methods based on the global equilibrium search, random local search, TABU and branch and bound methods are presented. The comparative analysis of the empirical computational performance is presented, comparing the best known algorithm from the literature and the algorithms proposed in this thesis. The results confirm the advantages of the proposed methods in terms of the solution quality and time. The parallel implementations of the portfolios and teams based on the global equilibrium search for solving maximum cut and unconstrained binary quadratic programming problems demonstrate high potential for improving computational times and solution quality on the large scale problems.