Experiences with the application of the ADIC automatic differentiation tool to the CSCMDO 3-D volume grid generation code

Abstract

Abstract. Automatic differentiation {AD) is a methodology for developing reliable sensitivityenhanced versions of arbitrary computer programs with little human effort. It can vastly accelerate the use of advanced simulation codes in multidisciplinary design optimization, since the time for generating and verifying derivative codes is greatly reduced. In this paper, we report on the application of the recently developed ADIC automatic differentiation tool €or ABSI C programs to the CSCMDO multiblock three-dimensiond volume grid generator. The ADIC-generated code can easily be interfaced with Fortran derivative codes generated with the ADIFOR AD too1 for FOR-TRAN 77 progritIIis, thus providing efficient sensitivity-enhancement techniques for multilasguage, multidiscipline problems.

Echtzeit-Implementierung von NMPC

“…Several AD software tools are being developed by the applied mathematics and computing research community, all having both common and distinct features, regarding languages supported (FORTRAN 77, FORTRAN 90, C, C++). These tools include ADIFOR [BGH92], AUTO-DERIV [SPF00], ADIC [Bis96], ADOL-C [GDJ96], Odysee [FP98], and TAMAC [Gie97]. In the MATLAB environment, there is ADMIT-1 [CV97], ADMAT [CV98], MAD [AMR00], and ADiMAT [Veh04].…”

Effect of Random Geometric Uncertainty on the Computational Design of a 3D Flexible Wing

“…The CFD volume mesh needed by the flow analysis code was generated using a version of the CSCMDO [29] grid generation code. The associated grid SDs needed by the flow sensitivity analysis were generated with an automatically differentiated version of CSCMDO [30]. The 45,000 grid point baseline volume mesh required by CSCMDO and used in the present flexible wing examples was obtained with the Gridgen™ code.…”

Simultaneous aerodynamic analysis and design optimization (SAADO) for a 3-D rigid wing

“…Associated grid sensitivity derivatives needed by the flow sensitivity analysis were generated with an automatically differentiated version of CSCMDO. 33 In addition to the parameterized surface mesh and accompanying gradients, CSCMDO requires a baseline volume mesh of similar shape and identical topology. The 45,000 grid point baseline volume mesh of C-O topology used in the present flexible wing examples was obtained with the Gridgen™ code.…”