论文信息 - Quantization errors in floating-point arithmetic

Quantization errors in floating-point arithmetic

In this paper, the quantization of the mantissa in a normalized floating-point number is investigated. A necessary and sufficient condition is given for the mantissa to have a reciprocal probability density. A model to represent a floating-point quantizer with the mantissa having a reciprocal density is developed. The first- and second-order statistical properties of this model are studied.

A. Sripad | D. Snyder | A. Sripad | D. Snyder

[1] R. W. Hamming,et al. On the distribution of numbers , 1970, Bell Syst. Tech. J..

[2] J. Richard Swenson,et al. Tests of probabilistic models for propagation of roundoff errors , 1966, CACM.

[3] B. Liu,et al. Error analysis of digital filters realized with floating-point arithmetic , 1969 .

[4] A. Sripad,et al. A necessary and sufficient condition for quantization errors to be uniform and white , 1977 .

[5] Donald Ervin Knuth,et al. The Art of Computer Programming, Volume II: Seminumerical Algorithms , 1970 .

[6] Richard W. Hamming,et al. Numerical Methods for Scientists and Engineers , 1963 .

[7] Bede Liu,et al. On Local Roundoff Errors in Floating-Point Arithmetic , 1973, JACM.

[8] Roger S. Pinkham,et al. On the Distribution of First Significant Digits , 1961 .

[9] A. Fettweis. On properties of floating-point roundoff noise , 1974 .

[10] James Hardy Wilkinson,et al. Rounding errors in algebraic processes , 1964, IFIP Congress.

[11] Richard Goodman,et al. Convergence Estimates for the Distribution of Trailing Digits , 1976, JACM.

[12] C. K. Yuen,et al. Theory and Application of Digital Signal Processing , 1978, IEEE Transactions on Systems, Man, and Cybernetics.