在python中将快速排序随机化，递归问题

Question

I am implementing randomized quick sort. For now I have created a function ChoosePivot(A,N) . This returns a random pivot and its location in input array. Then I switch that random pivot with first element in array so that pivot in Partition (A,l,r) is always first element. For Now ChoosePivot(A,N) is also returning first element of the array but I plan to modify it later.

Following is my code:

def QuickSort(A,N):
    if (N == 1):
        return
    pivot, pivot_pos = ChoosePivot(A,N)
    l = 0
    r = len(A)
    # Preprocessing, swapping pivot position with first element so that first element remains pivot always
    temp = A[0]
    A[0] = A[pivot_pos]
    A[pivot_pos] = temp

    A, i= Partition(A,l,r)
    print A,i
    # If i-1 == 0 this means that there is no left subarray
    if (i-1 != 0):
        print "Unsorted array"
        print A[0:i-1]
        QuickSort(A[0:i-1],i-1)
        print "Left call"
        print A

    if (N-i !=0): 
        print "Unsorted array"
        print A[i:N]
        QuickSort (A[i:N], N-i)
        print "Right call"
        print A

Following is my Partition(A,l,r)

def Partition(A,l,r):
    # Now first element is the pivot
    i= l + 1
    pivot = A[l]
    for j in range(l+1, r):
        if (A[j] < pivot):
            #swap (A[j], A[i])
            temp_1 = A[i]
            A[i] = A[j]
            A[j] = temp_1
            i = i+1
    #swap (A[i-1], A[l])
    temp_2 = A[i-1]
    A[i-1] = A[l]
    A[l] = temp_2
    return A, i 

ChoosePivot(A,N) Simply returns the first element in array for now

def ChoosePivot(A, N):
    #print A
    return A[0], 0

Input array I used is as follows:

Test_in = [3,8,2,5,1,4,7,6]
print Test_in

QuickSort(Test_in, len(Test_in))
print Test_in 

Please see that I can see code working at lower end of recursion. I did dry run code with pen and paper and I can see with print statements that it does sort the sub arrays but when array is returned finally, it has not changed. I thought its something related to value vs reference calling but I found that python calls by reference. So there should not be any problem there. Posting output as well and pointing out exactly where it went wrong.

Answer 1

I found the bug. Problem is when I call QuickSort again, I slice the original list and pass it as argument. When list is sliced python creates a new list and copied the sliced list into new list with same name. All changes happen in the new list deep within the recursion so they are not reflected back to the top. As solution I passed entire array and passed the list indices as arguments to the array.

There was still another bug. Condition before calling QuickSort which is i-1!=0 and Ni!=0 are not correct as pivot could be any index in array. I created separate left and right indices for left and right subarray and tested left_index < right_index .

Thanks