A STOCHASTIC DOMINANCE ALGORITHM USING PIECEWISE LINEAR APPROXIMATIONS
针对现有随机占优算法用阶梯函数逼近连续分布导致误差累积的问题，提出一种用分段线性函数逼近累积分布的新算法，并在正态分布案例中与旧算法比较，约95%情况下结果一致。
ABSTRACT Current stochastic dominance algorithms use step functions to approximate the cumulative distributions of the alternatives, even when the underlying random variables are known to be continuous. Since stochastic dominance tests require repeated integration of the cumulative distribution functions, a compounding of errors may result from this type of approximation. This article introduces a new stochastic dominance algorithm that approximates the cumulative distribution function by piecewise linear approximations. Comparisons between the new and old algorithms are performed for normally distributed alternatives. In about 95 percent of all cases, the two algorithms produce the same result.