实验变差函数分治算法

Divide Algorithm of Experiment Variation Function

目的为优化实验变差函数算法,提高计算效率.方法采用布尔沙-沃尔夫模型对样品坐标进行坐标转换,利用网格分割的方法进行分区计算空间任意方向变差函数.结果比较分治算法与常规算法的计算次数,提出实验变差函数分治算法,得出常规算法与分治算法计算量的倍数关系式.结论实验结果表明:在大样本情况下分治算法计算效率高于普通的实验变差函数算法,计算结果可靠.

In order to optimize the algorithm and improve the efficiency of calculating experiment variation, this paper introduces divide algorithm of experiment. Bursa-Wolf model is adopted to convert samples coordinates. We can calculate any orientation variation function according to mesh segmentation and coordination conversion. Comparing calculation times between common algorithm and divide algorithm, we propose divide algorithm of experimental variation function. Experimental results indicate that the calculation times of divide algorithm on large numbers of samples are fewer than those of common algorithm, and the calculation results are credible.