NYMA (NASA Lewis Research Center)
Brook Park, Ohio, U. S. A.
The proposed presentation concerns an experiment in software engineering for scientific and engineering code. The experiment is the development and evaluation of an algorithmic procedure intended to analyze aspects of the meaning or semantics of scientific and engineering code. A proven ability to perform automated semantic analysis would reduce the time, risk, and effort of developing and using scientific and engineering codes.
The procedure being developed and evaluated involves taking a user's existing code, adding semantic declarations for some primitive variables, and parsing this annotated code using multiple, distributed expert parsers. These semantic parsers are designed to recognize programming conventions, formulae in different disciplines including physical and mathematical formulae and geometrical position in a numerical scheme.
The implementation of this analysis procedure will be described, as well as metrics of its ability to recognize real world subroutines. Further, the theoretical basis for this work will be explained.