GRESS: gradient-enhanced software system. Version B. User's guide

Technical Report · Fri Apr 01 04:00:00 EST 1983

OSTI ID:6335889

Oblow, E M

GRESS is a FORTRAN compiler language than enhances conventional FORTRAN programs with analytic differentiation of arithmetic statements. GRESS handles any standard FORTRAN code which contains arithmetic text material as part of the source program. The GRESS compiler reads the FORTRAN source code text, redefines the variables and their storage locations, searches for arithmetic statements, translates the latter for gradient calculations, and then generates a new source program which now includes gradient capabilities in all arithmetic statements. GRESS, therefore, allows any standard FORTRAN code to be upgraded to calculate any derivatives required, whether they be for internal use in a calculation (e.g. for iteration) or for external use (e.g. sensitivity studies). The independent variables with respect to which the partial derivatives are calculated in the FORTRAN source code may be freely selected, and the only theoretical restriction is that the sequence of computation must be continuous and differentiable. The gradients are calculated according to the chain rule of differential standard chain rule. Thus, no numerical difference scheme is involved and the results are as accurate as any algebraic computations on the host ocmputer. In this process, the analytic formulas representing the derivatives are neither generated explicity nor saved. At any stage in the program execution, therefore, the only additional information available in the original code is the numerical value of the gradients.

• Research Organization: Oak Ridge National Lab., TN (USA)
• DOE Contract Number: W-7405-ENG-26
• OSTI ID: 6335889
• Report Number(s): ORNL/TM-8339; ON: DE83010463
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
990200* -- Mathematics & Computers
COMPUTER CODES
DOCUMENT TYPES
FORTRAN
G CODES
MANUALS
MATHEMATICS
PROGRAMMING LANGUAGES