Algorithmic Differentiation for an efficient CFD solver

We illustrate the benefits of Algorithmic Differentiation (AD) for the development of aerodynamic flow simulation software. In refining the architecture of the elsA CFD solver, developed jointly by ONERA and Safran, we consider AD as a key technology to cut development costs of some derivatives of interest, namely the tangent, adjoint, and Jacobian. We first recall the mathematical background of CFD applications which involve these derivatives. Then, we briefly present the software architecture of elsA (Cambier et al. [12]) and the design choices which give it its HPC capability while highlighting how these choices strongly constrain the applicability of AD. To meet our efficiency requirements, we select the Source-Transformation approach to AD through the Tapenade tool which is justified by a series of experiments and benchmarks. Finally, we present results on large scale configurations.