ACM Transactions on Mathematical Software · 1986 · 10 citations · 6 references
ComputingComputer ScienceMathematicsAlgorithmsProgramming LanguagesScientific ComputingComputational AlgebraNumerical AlgorithmsNumerical AnalysisNumerical Linear AlgebraComputational PrecisionError Analysis In ArithmeticFloating-point ArithmeticInterval Arithmetic
The program package ACRITH (High-Accuracy Arithmetic Subroutine Library) provides FORTRAN subroutines for the solution of several standard mathematical problems. The routines use floating point operations with extended precision and interval arithmetic and are designated especially for the solution of ill-conditioned problems. Test results for most of the routines are presented with emphasis on the practical usability of the package. It turns out that not all routines are of equal high quality and reliability; in the documentation, hints to the implemented numerical algorithms are completely missing, and the error messages are not always concise. Some possible alternatives like symbolic algebra systems or multiple precision packages are mentioned.
6
Rounding errors in algebraic processes
Journal of the Franklin Institute · 1964 · 1.3K citations
Rounding Errors in Algebraic Processes
Nature · 1964 · 459 citations