Maple programs for generating efficient FORTRAN code for serial and vectorised machines Original Research Article

We present two packages written in the language of the Maple Computer Algebra system for creating efficient FORTRAN code. The package called Macrofort provides the user with tools for creating optimised FORTRAN code. Through a realistic application, namely the computation of molecular integrals arising in the area of quantum chemistry, we show how Macrofort can generate optimised FORTRAN code for vectorised machines. The second package called Transfor allows the user to efficiently convert Matrix operations, as written in Maple, into Basic Linear Algebra Subroutines (BLAS). These are the essential building-blocks of fast numerical computation in research and industry. BLAS routines are available for serial, vectorised and parallel machines. These tools provide results beyond the capacities of modern-day compilers.