Distributed evolutionary optimization using Nash games and GPUs - Applications to CFD design problems

被引：16
作者：
Leskinen, Jyri [1 ]
Periaux, Jacques [1 ,2 ]
机构：
[1] Univ Jyvaskyla, Dept Math Informat Technol, FI-40014 Jyvaskyla, Finland
[2] Univ Politecn Cataluna Edifcio C1, Int Ctr Numer Methods Engn CIMNE, Barcelona 08034, Spain
来源：
COMPUTERS & FLUIDS | 2013年 / 80卷
关键词：
Computational fluid dynamics; Distributed computing; Geometry decomposition; GPUs; Nash games; Shape optimization; DIFFERENTIAL EVOLUTION; GENETIC ALGORITHMS;
D O I：
10.1016/j.compfluid.2012.03.017
中图分类号：
TP39 [计算机的应用];
学科分类号：
081203 ; 0835 ;
摘要：
In this paper we present new results obtained by a competitive game based distributed geometry decomposition method (GDM) for graphics processing unit (GPU) assisted shape design. The method replaces the original single-objective optimization problem by the search of Nash equilibria in a multiobjective optimization problem. Significant wall-clock time reductions obtained using the GDM approach and GPUs are presented and discussed. Tests on stacked NACA0012 airfoil geometries operating at subsonic and transonic regimes illustrate the potential of the new methodology. (C) 2012 Elsevier Ltd. All rights reserved.
引用
收藏
页码：190 / 201
页数：12
相关论文
共 33 条
  • [1] [Anonymous]
  • [2] [Anonymous], 8427 ICASE
  • [3] [Anonymous], 2004, GPU gems
  • [4] Arian E, PROG SYST CONTROL TH, V19
  • [5] Sparse matrix solvers on the GPU:: Conjugate gradients and multigrid
    Bolz, J
    Farmer, I
    Grinspun, E
    Schröder, P
    [J]. ACM TRANSACTIONS ON GRAPHICS, 2003, 22 (03): : 917 - 924
  • [6] Self-adapting control parameters in differential evolution: A comparative study on numerical benchmark problems
    Brest, Janez
    Greiner, Saso
    Boskovic, Borko
    Mernik, Marjan
    Zumer, Vijern
    [J]. IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2006, 10 (06) : 646 - 657
  • [7] Assembly of finite element methods on graphics processors
    Cecka, Cris
    Lew, Adrian J.
    Darve, E.
    [J]. INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING, 2011, 85 (05) : 640 - 669
  • [8] Topical perspective on massive threading and parallelism
    Farber, Robert M.
    [J]. JOURNAL OF MOLECULAR GRAPHICS & MODELLING, 2011, 30 : 82 - 89
  • [9] Gainaru A, 2011, LECT NOTES COMPUT SC, V6804, P102, DOI 10.1007/978-3-642-21916-0_12
  • [10] Goddeke Dominik, 2009, 2009 International Conference on High Performance Computing & Simulation (HPCS), P12, DOI 10.1109/HPCSIM.2009.5191718
1234