Abstract
A new method of estimatinga posteriori the statistical characteristics of the rounding errors of an arbitrary algorithm is presented. This method is based on a discrete model of the distribution of rounding errors which makes more accurate estimates possible. The analysis is given for both rounding and truncating arithmetic. Finally, some experimental results are reported.
Article PDF
Similar content being viewed by others
Avoid common mistakes on your manuscript.
References
P. Henrici,Elements of numerical analysis, Ch. 16, Wiley, New York, 1964.
T. E. Hull and J. R. Swenson,Tests of probabilistic models for propagation of roundoff errors, Comm. ACM 9 (1966), 108–113.
T. Kaneko and B. Liu,On local round-off errors in floating-point aritmetic, J. ACM 20 (1973), 391–398.
D. E. Knuth,Seminumerical algorithms, The Art of Computer Programming, Vol. 2, Sec. 4.2, Addison-Wesley, Reading, Massachusetts, 1969.
S. Linnainmaa,The representation of the cumulative rounding error of an algorithm as a Taylor expansion of local rounding errors), Master's Thesis, Department of Computer Science, University of Helsinki, Helsinki, Finland, 1972.
S. Linnainmaa,Analysis of some known methods of improving the accuracy of floating-point sums, BIT 14 (1974), 167–202.
S. Linnainmaa,Statistical estimation methods of rounding errors in floating-point computations, Report A-1975-1, Department of Computer Science, University of Helsinki, Helsinki, Finland.
J. B. Scarborough,Numerical mathematical analysis, 5th ed., Ch. I, Johns Hopkins Press, Baltimore, 1962.
D. W. Sweeney,An analysis of floating-point addition, IBM Systems Journal 4 (1965), 31–42.
M. Tienari,A statistical model of roundoff errors for varying length floating-point arithmetic, BIT 10 (1970), 355–365.
M. Tienari,On some topological properties of numerical algorithms, BIT 12 (1972), 409–433.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Linnainmaa, S. Towards accurate statistical estimation of rounding errors in floating-point computations. BIT 15, 165–173 (1975). https://doi.org/10.1007/BF01932690
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01932690