Solving resource distribution nonlinear optimization problems in large block-structured systems with binding parameters
DOI:
https://doi.org/10.20535/SRIT.2308-8893.2016.3.07Keywords:
resource distribution problems, optimization models, approximation methods, nonlinear programming, decomposition algorithmsAbstract
Solving non-linear optimization problems with a block structure and binding parameters (variables) is realized by a combination of the approximation and decomposition approaches. The approximation method is chosen so that the decomposition of the mathematical programming problem can be performed without making any assumptions about the convexity or additive separability of objective functions and constraints. The coordinating and block sub-problems that are auxiliary in the approximation method, are solved using a finite number of steps. In the course of calculation, binding variables vary from step to step of the iterative process, providing a monotonic decrease of the value of the coordinating problem objective function; in other words, the amount of shared resources is changed in such a way that block subsystems operate more and more efficiently in terms of the efficiency of the whole system.References
Lesdon L.C. Optimizatsija bol'shih sistem / L.C. Lesdon. — M.: Nauka, 1975. — 432 s.
Tsurkov V.I. Dekompozitsija v zadachah bol'shoj razmernosti / V.I. Tsurkov. — M.: Nauka, 1981. — 352 s.
Laptin Ju.P. Nekotorye voprosy reshenija blochnyh nelinejnyh zadach optimizatsii so svjazyvajuschimi peremennymi / Ju.P. Laptin, N.G. Zhurbenko // Kibernetika i sistemnyj analiz. — 2006. — № 2. — S. 47–55.
Shefi R. Rate of Convergence Analysis of Decomposition Methods Based on the Proximal Method of Multipliers for Convex Minimization / R. Shefi, M. Teboulle // SIAM J. Optim. — 014. — 24, Issue 1. — P. 269–297. DOI: 10.1137/1309.10774.
Andersson H. A new decomposition algorithm for a liquefied natural gas inventory routing problem / H. Andersson, M. Christiansen, G. Desaulniers // International Journal of Production Research. — 2016. — 54, Issue 2. — P. 564–578. DOI: 10.1080/00207543.2015.1037024
Laptin Ju.P. Tochnye shtrafnye funktsii i vypuklye prodolzhenija funktsij v shemah dekompozitsii po peremennym / Ju.P. Laptin // Kibernetika i sistemnyj analiz. — 2016. — 52, № 1. — S. 93–104.
Kirik O.Ye. Pidkhid do rozv’jazannja nelinijnykh optymizatsijnykh zadach blochnoyi struktury zi zv’jazujuchymy obmezhennjamy / O.Ye. Kirik // Naukovi visti NTUU "KPI". — K., 2015. — № 5. — S. 32–38.
Kirik O.Ye. Alhorytmy linearyzatsiyi ta sprjazhenykh hradiyentiv dlja nelinijnykh zadach rozpodilu potokiv / O.Ye. Kirik // Naukovi visti NTUU "KPI". — K., 2007. — № 3. — S. 67–73.
Pshenichnyj B.N. Chislennye metody v ekstremal'nyh zadachah / B.N. Pshenichnyj, Ju.M. Danilin. — M.: Nauka, 1975. — 319 s.
Aleksandrova V.M. Optymizatsijni modeli ta alhorytmy dlja merezhevykh zadach rozpodilu resursiv / V.M. Aleksandrova, O.Ye. Kirik // Naukovi visti NTUU "KPI". — 2014. — № 4. — S. 39–45.
Pshenichnyj V.N. Vypuklyj analiz i ekstremal'nye zadachi / V.N. Pshenichnyj. — M.: Nauka, 1980. — 320 s.
Faddeev D.K. Vychislitel'nye metody linejnoj algebry / D.K. Faddeev, V.N. Faddeeva. — M.: Fizmatgiz, 1960. — 656 s.
