The comparative analysis of the effectiveness of Goldberg and Holland’s classical models and their modifications using various options of the generational strategy is presented. The concept assuming that the number of individuals in a generation does not change is used in the classical genetic algorithms. An approach advancing the efficiency of standard Goldberg and Holland’s models through varying the number of individuals in a generation is considered. Various embodiments of the generational strategy are used to solve the homogeneous minimax scheduling problem related to the class of NP-complete problems. The computational experiment conducted for a various number of processors and works has shown that this approach can significantly improve the genetic algorithm efficiency by small changes in the standard models allowing obtain the solution that is closer to the accurate solution.
genetic algorithms, Goldberg model, Holland model, NP-complete problems, generational strategy, scheduling theory.
Введение. Теория расписаний — раздел дискретной математики, занимающийся проблемами упорядочения. Существуют различные варианты задач теории расписаний. Часть из них является NP-полными. NP-полные задачи образуют подмножество типовых задач в классе NP, к которым можно свести любую другую задачу из этого класса полиномиально быстрым алгоритмом решения [1, 8, 9]. В различных областях дискретной математики, комбинаторики и логики известно множество задач, принадлежащих к классу NP-полных задач. Для этих задач не найдены полиномиальные алгоритмы. Однако и не доказано, что таких алгоритмов не существует. Нахождение точного решения для задачи из класса NP-полных является практически невыполнимым. Поэтому для таких задач разрабатываются различные методы, позволяющие получить приближённое решение.
1. Kobak, V. G., Trotsyuk, N. I., Rozhkovskiy, B. A. Sravnitelnyye kharakteristiki modifikatsii modeli Khollanda pri pokolencheskoy strategii. Trudy Severo-Kavkazskogo filiala Moskovskogo tekhnich-eskogo universiteta svyazi i informatiki. [Comparative characteristics of Holland model modification under generational strategy. Proc. North Caucasian Branch of Moscow Technical University of Communications and Informatics.] Rostov-on-Don : Publ. Center «Universitet» SKF MTUSI, 2014, part 1, pp. 319–322 (in Russian).
2. Kobak, V. G., Trotsyuk, N. I. Sravnitelnyy analiz algoritmov: geneticheskogo s elitoy i Krona s geneticheskim nachalnym raspredeleniyem. Matematicheskiye metody v tekhnike i tekhnologiyakh : sb. trudov XXVI mezhdunar. nauch. konf. [Comparative analysis of algorithms: genetic one with elite and Crohn´s genetic initial distribution. Mathematical Methods in Engineering and Technology : Proc. XXVI Int. Sci. Conf.] Saratov, 2013, vol. 12, part 2, pp. 62–64 (in Russian).
3. Kobak, V. G., Trotsyuk, N. I. Ispolzovaniye pokolencheskoy strategii modeli Goldberga pri reshenii odnorodnoy minimaksnoy zadachi. [Application of Goldberg model generational strategy for homogeneous minimax problem solution.] Aspirant, 2014, no. 2, pp. 62–64 (in Russian).
4. Korneyev, V. V., et al. Bazy dannykh. Intellektualnaya obrabotka informatsii. [Database. Intelligent information processing.] Moscow : Nolidzh, 2000, 352 p. (in Russian).
5. Neydorf, R. A., Kobak, V. G., Titiov, D. V. Sravnitelnyy analiz effektivnosti variantov turnirnogo otbora geneticheskogo algoritma resheniya odnorodnykh raspredelitelnykh zadach. [Comparative analysis of alternative effectiveness of genetic algorithm tournament selection for solving homogeneous allocation problems.] Vestnik of DSTU, 2009, vol. 9, no. 3, pp. 410–418 (in Russian).
6. Kureychik, V. М. Geneticheskiye algoritmy i ikh primeneniye. [Genetic algorithms and their application.] Taganrog : TRTU Publ. House, 2nd red., 2002, 242 p. (in Russian).
7. Kureychik, V. М., Gladkov, L. A., Kureychik, V. V. Geneticheskiye algoritmy. [Genetic algo-rithms.] Moscow : Fizmatlit, 2006, 319 p. (in Russian).
8. Koffman, E. G. Teoriya raspisaniy i vychislitelnyye mashiny. [Scheduling theory and computers.] Moscow : Nauka, 1984, 336 p. (in Russian).
9. Pashkeyev, S. D., Menyazov, I. R., Mogilevskiy, V. D. Mashinnyye metody optimizatsii v tekhnike svyazi. [Machine optimization techniques in communication technology.] Moscow : Svyaz, 1976, 250 p. (in Russian).
10. Batishchev, D. I. Geneticheskiye algoritmy resheniya ekstremalnykh zadach. [Genetic algo-rithms for solving extremum problems.] Voronezh : VGTU, 1995, 69 p. (in Russian).
