论文信息 - Summation of imprecise numbers

Summation of imprecise numbers

Abstract An algorithm is presented for computer summation of imprecise numbers that may have quite disparate magnitudes and variances. Each number is represented by its expected value and variance. Exact numbers form a subset and are treated in a consistent manner. Cancellation effects are minimized.

N. Metropolis

[1] Jack M. Wolfe. Reducing truncation errors by programming , 1964, CACM.

[2] Michael A. Malcolm,et al. On accurate floating-point summation , 1971, CACM.

[3] Robert L. Ashenhurst,et al. Significant Digit Computer Arithmetic , 1958, IRE Trans. Electron. Comput..

[4] Peter Linz,et al. Accurate floating-point summation , 1970, CACM.