Separating Language Dependent and Independent Tasks for the Semantic Transformation of Numerical Programs

J. Utke and U. Naumann (USA)


source-to-source transformation, language independence, auto matic differentiation


Numerical models increasingly move to advanced programming paradigms for instance using object oriented features of C++ or the more recent Fortran standards. Sophisticated models for com plex real world problems for instance in oceanography pose in themselves a significant computational challenge. The semantic transformations of those models done for example for sensitiv ity or uncertainty analysis become rather complex. To allow for reuse of the transformation implementations we have to identify a separation of programming-language-dependent from indepen dent implementation portions. Furthermore we have to find an ap proach to handle the language dependent transformation portions that is flexible and involves marginal coding effort. We show the range of options, introduce our solution and discuss the practica bility and the consequences of the choices.

Important Links:

Go Back