Applying Algorithm Unions to Specific Classes of Boolean Programming Problems
摘要
The paper considers unions (portfolios and teams) of algorithms for solving a number of complex Boolean programming problems. Significant attention is given to the experimental study of the developed unions. For example, for the complex quadratic assignment problem tai100a, which remains a considerable challenge for researchers worldwide, a new record was achieved using a portfolio of 16 algorithms, modifications of the tabu search algorithm. The use of teams consisting of four such algorithms made it possible to improve this record. The acceleration factors obtained for solving the shortest-cover problem using portfolios of random iterative local search algorithms, compared to a single algorithm, approach a linear acceleration factor.