Review and Unification of Methods for Computing Derivatives of Multidisciplinary Computational Models

作者: Joaquim R. R. A. Martins , John T. Hwang

DOI: 10.2514/1.J052184

关键词:

摘要: This paper presents a review of all existing discrete methods for computing the derivatives computational models within unified mathematical framework. framework hinges on new equation, unifying chain rule, from which can be derived. The computation is described as two-step process: evaluation partial and total derivatives, are dependent derivatives. Finite differences, complex-step method, symbolic differentiation discussed options It shown that these building blocks with assembled using algorithmic differentiation, direct adjoint methods, coupled analytic multidisciplinary systems. Each presented applied to common numerical example demonstrate compare various approaches. ends discussion current challenges possib...

参考文章(110)
J. Utke, U. Naumann, OpenAD/F: User Manual ,(2006)
W Gropp, P Brune, A Dener, S Abhyankar, K Rupp, B Smith, S Balay, D May, T Munson, D Kaushik, H Zhang, K Buschelman, M Knepley, R Mills, L Curfman McInnes, L Dalcin, J Brown, P Sanan, D Karpeyev, M Adams, S Zampini, Eijkhout, PETSc Users Manual Argonne National Laboratory. ,(2019)
Andrea Saltelli, Marco Ratto, Terry Andres, Francesca Campolongo, Jessica Cariboni, Debora Gatelli, Michaela Saisana, Stefano Tarantola, Global Sensitivity Analysis: The Primer ,(2008)
Valérie Pascual, Laurent Hascoët, TAPENADE 2.1 user's guide INRIA. pp. 78- ,(2004)
James Reuther, Juan J. Alonso, Joaquim R. Martins, Aero-Structural Wing Design Optimization Using High-Fidelity Sensitivity Analysis ,(2001)
J Utke, OpenAD : algorithm implementation user guide. Other Information: PBD: 13 May 2004. ,(2004) , 10.2172/834715
B. F. Blackwell, K. J. Dowding, Sensitivity Analysis and Uncertainty Propagation of Computational Models John Wiley & Sons, Inc.. pp. 443- 469 ,(2009) , 10.1002/9780470172599.CH14