Sorting operation partition, argpartition, sort, argsort in numpy

Reference resources ：

Explain in detail ：https://blog.csdn.net/m0_37192554/article/details/107459712

Usually used to full sorting np.sort, The complexity is between and . If the array to be sorted is particularly large , The time cost is not small . But generally, the sorting of extremely large arrays does not need to be full , Maybe you just need to take Top K. At this point, you can use partial sorting , such as numpy.partition and numpy.argpartition.

partition The function functions as follows ： Given a array, Rearrange the elements to make the k The element of bit happens to be the first k The element of bit . All Biti k Bit elements small elements are placed in front of it , All Biti k The elements with large bit elements are placed behind it . in other words , This article k Bit elements divide the entire array into two parts , The front is smaller than it , The back is bigger than it . But the elements inside each part are not sorted .

kth It can also be negative
Represents an array a pass the civil examinations 2 Large elements are elements 8 In the sorting completion array b The last of 2 In a position , Index b[-2] It's about
Of course , What's the use of this ？ One of the uses is if we have a very large array , If we want to find the biggest one 10 Number , What shall I do? , We can use sort function , Then take it out after sorting , But this method is time-consuming when the array is too large , So we can use partition function ,partition Sort the array only once , find k The number of positions is enough , And yes k The order of position left and right doesn't care , Less work means faster efficiency