新的等价类划分算法-计数法

目前，基于基数排序的等价类划分算法有较低的时间复杂度但存在以下不足：属性值跳跃性大时会产生大量空队列；排序后仍需O（|P‖U|）的时间才实现划分，求出等价类，排序没能发挥应有作用。为此，设计了一种新算法，通过属性值映射避免大量空队列产生，通过增加一个记录等价类长度信息的计数数组，排序后仅需O（|U|）就可实现划分，求出等价类。整个算法时间复杂度为O（|C‖U|），空间复杂度为O（|U|），为求等价类划分提供了一个新的解决办法。

Counting method-novel algorithm for computing partition

At present，the algorithm for computing partition based on radix sorting has lower time complexity than the others，but it also has the following shortcomings：it will create a lot of idle storage when there is a big increment between two attribution value；for computing the partition，its time complexity is as high as O（|P‖U|） after sorting.For improving the efficiency of computing partition，a novel algorithm for computing partition is designed.It does not create a lot of idle storage by mapping the attribution values to sequential natural number.And it records the length of equivalence classes to compute the partition by defining a count array.Its time complexity is as only low as O（|U|）.The time complexity of the new algorithm is O（|C‖U|），the space complexity is O（|U|）.It offers a new method for computing the partition of the universe U.