Variation of the counting sort algorithm?

Question

I realize this isn't the standard counting sort as stated in CLRS, but what does my simplified version lack that the standard counting sort has? I realize mine only sorts positive integers, but that should be pretty easy to tweak (by using maps).

def count_sort(array):
    maximum = max(array)
    minimum = min(0, min(array))
    count_array = [0]*(maximum-minimum+1)

    for val in array:
        count_array[val] += 1

    sorted_array = []
    for i in range(minimum, maximum+1):
        if count_array[i] > 0:
            for j in range(0, count_array[i]):
                sorted_array.append(i)

    return sorted_array

array = [3,2,-1,1,5,0,10,18,25,25]
print array
print count_sort(array)

Edit: The reason why I thought this wasn't the standard counting sort was because the algorithm covered in the MIT OpenCourseware lecture seemed slightly more involved (http://youtu.be/0VqawRl3Xzs?t=34m54s).

Answer 1

You're doing something odd with your min and max. Try this:

def count_sort(array):
    maximum = max(array)
    minimum = min(array)
    count_array = [0]*(maximum-minimum+1)

    for val in array:
        count_array[val-minimum] += 1

    sorted_array = []
    for i in range(minimum, maximum+1):
        if count_array[i-minimum] > 0:
            for j in range(0, count_array[i-minimum]):
                sorted_array.append(i)

    print sorted_array

array = [3,2,-1,1,5,0,10,18,25,25]
print array
count_sort(array)

Answer 2

I don't see problems with the way you're count-sorting, anyway some suggestions to your code come to mind that maybe can interest you.

For example the line:

minimum = min(0, min(array))

could be replaced with:

minimum = min(0, *array)

If you ran your code in python-2.x use xrange() instead of range() , so you could change:

for i in range(minimum, maximum+1):

with:

for i in xrange(minimum, maximum+1):

For the last loop you don't actually need range() at all, check out the question pythonic way to do something N times , so you could change:

for j in range(0, count_array[i]):
    sorted_array.append(i)

with:

for _ in itertools.repeat(None, count_array[i]):
    sorted_array.append(i)

Answer 3

Do you mean http://en.wikipedia.org/wiki/Counting_sort ?

If yes, it's almost similar to your code. But it has one improvement: "The relative order of items with equal keys is preserved here, ie this is a stable sort.". So if you are sorting key-value dictionary order of the values will remain the same if they have equal keys.