ABSTRACT
A scheme is proposed for permitting a user of conventional procedural programming languages (initially, Standard FORTRAN) to test actual error propagation in numerical calculations. The process is to be fully mechanistic so that, with no human resequencing required or permitted, a "numerical procedure debugging" tool is made available. Other goals include a quantification of the order-of-precision decision for specified accuracy, provision of an observational tool for determining word length requirements, and an automatic facility for utilizing other kinds of arithmetic interpretively in executing existing programs and program segments. The experimental package consists, in effect, of a compiler from FORTRAN source language into an artificial machine language in which arithmetic operations produce, in addition to numerical results, a measure of the current accuracy of each result operand.
- Carr, J. W., III, Error Analysis in Floating-Point Arithmetic, Comm. ACM 2, 10--15 (1959). Google ScholarDigital Library
- Chartres, B. A., and Geuder, J. C., Computable Error Bounds for Direct Solution of Linear Equations J. ACM 14, 63--71 (1967). Google ScholarDigital Library
- Hammer, C., Statistical Validation of Mathematical Computer Routines, Proc. Spring Joint Comput. Conf. pp. 331--333 (1967). Google ScholarDigital Library
- Householder, A. S., Generation of Errors in Digital Computation, Bull. Amer. Math. Soc. 53, 234--247 (1947).Google ScholarCross Ref
- von Neumann, J., and Goldstine, H. H., Numerical Inverting of Matrices of High Order, Bull. Amer. Math. Soc. 53, 1021--1099 (1947).Google ScholarCross Ref
- Goldstine, H. H., and von Neumann, J., Numerical Inverting of Matrices of High Order, II, Proc. Amer. Math. Soc. 2, 188--204 (1951).Google ScholarCross Ref
- Moler, C. E., Accurate solution of Linear Algebraic Systems---a Survey, Proc. Spring Joint Comput. Conf. pp. 321--324 (1967). Google ScholarDigital Library
- Moler, C. B., Iterative Refinement in Floating-Point, J. ACM 14, 316--321 (1967). Google ScholarDigital Library
- Moore, R. E., Automatic Error Analysis in Digital Computation, Lockheed M&S Div. Tech. Rept. LMSD-48421, January 1959.Google Scholar
- Ralson, A., and Wilf, S., "Mathematical Methods for Digital Computers." Wiley, New York, 1962.Google Scholar
- Oettli, W., and Prager, W., Compatibility of Approximate Solution of Linear Equations with Given Error Bounds for Coefficients and Right-Hand Sides, Numer. Math. 6, 405--409 (1964).Google ScholarDigital Library
- Rigal, J. L., and Caches, J., On the Compatibility of a Given Solution with the Data of a Linear System, J. ACM 14, 543--548 (1967). Google ScholarDigital Library
- This paper gives and proves theorems regarding the numerical validity of solutions to linear systems. It suggests an improved stop test for iterative processes, taking into account numerical uncertainties of both system and solution. It generalizes the results of Oettli and Prager.Google Scholar
- Todd, J., The Problem of Error in Digital Computation, in "Error in Digital Computation" (L. B. Rail, ed.), Vol. 1 (Proc. of Seminar Univ. of Wisconsin, October 1964), pp. 3--44. Wiley, New York, 1965.Google Scholar
- Wilkinson, J. H., "Rounding Errors in Algebraic Processes." Prentice-Hall, Engel-wood Cliffs, New Jersey, 1963.Google Scholar
- Kelly, L. G., "Handbook of Numerical Methods and Applications." Addison-Wesley, Reading, Massachusetts, 1967.Google Scholar
- Turner, L. R., and Moshos, G. J., A Scheme for Obtaining Signed Error Indication, NASA--LERC Internal Technical Memorandum, 1963; also delivered in 1963 as an IEEE "Conference Paper Only"; remark: Turner and Moshos used this scheme as a part of an iterative method for improving numerical accuracy.Google Scholar
- Ashenhurst, R. L., and Metropolis, N., Unnormalized Floating-Point Arithmetic, Unpublished memorandum, Institute for Computer Research, Univ. of Chicago, Chicago, Illinois, 1958 (28 pages).Google Scholar
- Metropolis, N., and Ashenhurst, R. L., Significant Digit Computer Arithmetic, IRE Trans. Electron. Comput. 7, 265--267 (1958).Google ScholarCross Ref
- Ashenhurst, R. L., MANIAC III Arithmetic System, Proc. Spring Joint Comput. Conf. 195--202 (1962). Spartan Books, Washington, D.C., 1962. Google ScholarDigital Library
- Ashenhurst, R. L., Function Evaluation in Unnormalized Arithmetic, J. ACM 11, 168--187 (1964). Google ScholarDigital Library
- Ashenhurst, R. L., Techniques for Automatic Error Monitoring and Control, in "Error in Digital Computation " (L. B. Rail, ed.), Vol. 1 (Proc. of Seminar Univ. of Wisconsin, October 1964), pp. 43--59. Wiley, New York, 1965.Google Scholar
- Ashenhurst, R. L., and Metropolis, N., Error Estimation in Computer Calculation, Amer. Math. Monthly 72, Part II, 47--58 (1965).Google Scholar
- Metropolis, N., and Ashenhurst, R. L., Radix Conversion in an Unnormalized Arithmetic System, Math. Comput. 19, 435--441 (1965).Google ScholarCross Ref
- Metropolis, N., and Ashenhurst, R. L., "Basic operations in an unnormalized arithmetic system," IEEE Trans. Electron. Comput. 12, 896--904 (1965).Google Scholar
- Ashenhurst, R. L., Experimental Investigation of Unnormalized Arithmetic, in "Error in Digital Computation" (L. B. Rall, ed.), Vol. 2 (Proc. Seminar Univ. Wisconsin, November 1965), pp. 3--35. Wiley, New York, 1965.Google Scholar
- Cheydleur, B. F., Binary Notations in Automatic Computer Algorithms and Operation Codes, ACM 3rd Natl. Conf. 1949, Oak Ridge, Tennessee (conference paper pre-print only). Also, Significance Arithmetic Notation and Algorithms, unpublished internal memorandum, Naval Ordnance Lab., pp. 1--19, 1955.Google Scholar
- Goldstein, M., Significance Arithmetic on a Digital Computer, Commun. ACM 6, 111--117 (1963). Google ScholarDigital Library
- Gray, H. L., and Harrison, C., Jr., Normalized Floating-Point Arithmetic with an Index of Significance, Proc. Eastern Joint Comput. Conf. pp. 244--248 (1959). Google ScholarDigital Library
- Robinson, G., Some Theoretical Bases for Significance Arithmetic in FLIP, Argonne Natl. Lab. (internal) Tech. Memo. No. 42, pp. 1--17, 1963.Google Scholar
- Gregory, R. T., On the Design of the Arithmetic Unit of a Fixed Word Length Computer from the Standpoint of Computational Accuracy, IEEE Trans. Electron. Comput. 15, 255--257 (1966).Google ScholarCross Ref
- Kanner, H., Number Base Conversion in a Significant Digit Arithmetic, J. ACM 12, 242--246 (1965). Google ScholarDigital Library
- Miller, R. H., An Example in Significant Digit Arithmetic, Commun. ACM 7, 21 (1964). Google ScholarDigital Library
- Richtmyer, R. D., The Estimation of Significance, AEC Res. and Develop. Rept. NYO-9083, 1960.Google Scholar
- Cody, W. J., The Influence of Machine Design on Numerical Algorithms, Proc. Spring Joint Comput. Conf. pp. 305--309 (1967). Google ScholarDigital Library
- Graham, R. M., Compiling Technique and Algorithms, Univ. of Michigan Summer Session Notes, Advanced Automatic Programming, Univ. of Michigan, Ann Arbor, June 1963 (31 pages). Also, Bounded Context Translation, Proc. Spring Joint Comput. Conf. (1964).Google Scholar
- Matula, D. W., Base Conversion Mappings, Proc. Spring Joint Comput. Conf. pp. 311--318 (1967). Google ScholarDigital Library
- von Neumann, J., and Goldstine, H. H., Planning and Coding of Problems for an Electronic Computing Instrument, unpublished internal memorandum, Institute for Advanced Study, Princeton Univ., Princeton, New Jersey, 1947. (Note: While this report has long been out of print, informal copies are widely available.)Google Scholar
- Pope, D. A., and Stein, M. L., Multiple Precision Arithmetic, Commun. ACM 3, 652--654 (1960). Google ScholarDigital Library
- Hastings, C., "Approximations for Digital Computers," Princeton Univ. Press, Princeton, New Jersey, 1955. Google ScholarDigital Library
Recommendations
Understanding the accuracy of Nanbu's numerical Coulomb collision operator
We investigate the accuracy of and assumptions underlying the numerical binary Monte Carlo collision operator due to Nanbu [K. Nanbu, Phys. Rev. E 55 (1997) 4642]. The numerical experiments that resulted in the parameterization of the collision kernel ...
Transformation of a PID Controller for Numerical Accuracy
Numerical programs performing floating-point computations are very sensitive to the way formulas are written. Several techniques have been proposed concerning the transformation of expressions in order to improve their accuracy and now we aim at going a ...
Numerical approximation of the Vlasov-Poisson-Fokker-Planck system in one dimension
A numerical method is developed for approximating the Vlasov-Poisson-Fokker-Planck system in one dimension. This system of equations is a mathematical model for an electrostatic plasma in which collisions between the electron distribution and a ...
Comments