COMP 3203

Introduction to Data Structures

and Algorithms

Quick Sort
Quick Sort Algorithm
 Given an array of n elements (e.g., integers):
 If array only contains one element, return
 If array contains two element
 swap the elements if they are not in-order
 Else
 pick one element to use as pivot.
 Partition elements into two sub-arrays:
 Elements less than or equal to pivot
 Elements greater than pivot
 Quicksort the two sub-arrays
 Return results

2 By Dr. Farha Al Kharusi, DCS, SQU

Partitioning Array
 Given a pivot (can be any array element, for example the
first or the last), partition the elements of the array such
that the resulting array consists of:
 One sub-array that contains elements <= pivot
 Another sub-array that contains elements > pivot

 The sub-arrays are stored in the original data array.

 Partitioning loops swapping elements below/above pivot.

3 By Dr. Farha Al Kharusi, DCS, SQU

Quick Sort - Code
static void QuickSort(int[] data, int left, int right)
{
if ( right <= left +1) //0,1,or 2 items
{
if ( right == left + 1)// 2 items
compareSwap(data, left, right);
return;
}
int pivot = partition(data, left, right);
QuickSort(data, left, pivot-1);
QuickSort(data, pivot+1, right);
}

4 By Dr. Farha Al Kharusi, DCS, SQU

 Swap Function
static void compareSwap(int[] data,int left,int right)
{
if ( data[right] < data[left])
{
int temp = data[left];
data[left] = data[right];
data[right] = temp;
}
}

5 By Dr. Farha Al Kharusi, DCS, SQU

 Partition: Quick Sort - Code
static int partition(int[] data, int left, int right)
{
int pivot(left), i(left+1), j(right), temp;
while (j > i)
{
while (data[i] <= data[pivot]) i++;
while (data[j] > data[pivot]) j--;
if (i < j) //swap
{
temp = data[i];
data[i] = data[j];
data[j] = temp;
}//if
}//while

//Swap data[j] and data[pivot]

temp = data[pivot];
data[pivot] = data[j];
data[j] = temp;
return j;
}
6 By Dr. Farha Al Kharusi, DCS, SQU
Example
 We are given array of n integers to sort:

40 20 13 80 69 50 6 30 100

 Pick Pivot Element

There are a number of ways to pick the pivot element. In
this example, we will use the first element in the array:

40 20 13 80 69 50 6 30 100

7 By Dr. Farha Al Kharusi, DCS, SQU

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

8 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

9 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

10 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

11 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

12 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

13 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]

pivot_index = 0 40 20 13 80 69 50 6 30 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

14 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

15 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

16 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

17 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

18 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

19 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

20 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 69 50 6 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

21 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

22 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

23 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

24 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

25 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

26 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

27 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

28 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

29 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

30 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 0 40 20 13 30 6 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

31 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 4 6 20 13 30 40 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

32 By Dr. Farha Al Kharusi, DCS, SQU

Partition Result

6 20 13 30 40 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

<= data[pivot] > data[pivot]

33 By Dr. Farha Al Kharusi, DCS, SQU

Recursion: Quicksort Sub-arrays

6 20 13 30 40 50 69 80 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

<= data[pivot] > data[pivot]

34 By Dr. Farha Al Kharusi, DCS, SQU

Quick Sort of N elements: How many
comparisons?
N For first call, when each of N elements
is compared to the pivot value

2 * N/2 For the next pair of calls, when N/2

elements in each “half” of the original
array are compared to their own pivot values.

4 * N/4 For the four calls when N/4 elements in each

“quarter” of original array are compared to
their own pivot values.
.
.

. HOW MANY SPLITS CAN OCCUR?

35 By Dr. Farha Al Kharusi, DCS, SQU

Quicksort Analysis
 Assume that keys are random, uniformly distributed.
 What is best case running time?
 Recursion:
 Partition splits array in two sub-arrays of size n/2
 Quicksort each sub-array
 Depth of recursion tree? O(log2n)
 Number of accesses in partition? O(n)

36 By Dr. Farha Al Kharusi, DCS, SQU

Quicksort Analysis
 Assume that keys are random, uniformly distributed.
 Best case running time: O(n log2 n)
 Worst case running time?
 Assume first element is chosen as pivot.
 Assume we get array that is already in order:

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

37 By Dr. Farha Al Kharusi, DCS, SQU

Quicksort Worst Case
 Assume first element is chosen as pivot.
 Assume we have an array that is already in order:

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

38 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

39 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

40 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

41 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

i j

42 By Dr. Farha Al Kharusi, DCS, SQU

 1. while (data[i] <= data[pivot]) i++;
2. while (data[j] > data[pivot]) j--;
3. if(i < j)
swap data[i] and data[j]
4. while j > i, go to 1.
5. swap data[j] and data[pivot_index]

pivot_index = 0 2 4 10 12 13 50 57 63 100

[0] [1] [2] [3] [4] [5] [6] [7] [8]

<= data[pivot] > data[pivot]

43 By Dr. Farha Al Kharusi, DCS, SQU
Quicksort Analysis
 Assume that keys are random, uniformly distributed.
 Best case running time: O(n log2n)
 Worst case running time?
 Recursion:
 Partition splits array in two sub-arrays:
 one sub-array of size 0
 the other sub-array of size n-1
 Quicksort each sub-array
 Depth of recursion tree? O(n)
 Number of accesses per partition? O(n)

44 By Dr. Farha Al Kharusi, DCS, SQU

Quicksort Analysis
 Assume that keys are random, uniformly distributed.
 Best case running time: O(n log2n)
 Worst case running time: O(n2)!!!

45 By Dr. Farha Al Kharusi, DCS, SQU

Quick Sort of N elements: How many splits
can occur?
 It depends on the order of the original array elements!
 If each split divides the sub-array approximately in half,
there will be only log2N splits, and QuickSort is O(N
log2N).
 But, if the original array was sorted to begin with, the
recursive calls will split up the array into parts of unequal
length, with one part empty, and the other part containing
all the rest of the array except for split value itself. In this
case, there can be as many as N-1 splits, and QuickSort is
O(N2).

46 By Dr. Farha Al Kharusi, DCS, SQU

QuickSort Analysis
 Assume that keys are random, uniformly distributed.
 Best case running time: O(n log2n)
 Worst case running time: O(n2)!!!
 What can we do to avoid worst case?

47 By Dr. Farha Al Kharusi, DCS, SQU

Improved Pivot Selection
Pick median value of three elements from data array:
data[0], data[n/2], and data[n-1].

Use this median value as pivot.

48 By Dr. Farha Al Kharusi, DCS, SQU

Improving Performance of Quicksort
 Improved selection of pivot.
 For sub-arrays of size 3 or less, apply basic force search:
 Sub-array of size 1: trivial
 Sub-array of size 2:
 if(data[first] > data[second]) swap them
 Sub-array of size 3: left as an exercise

49 By Dr. Farha Al Kharusi, DCS, SQU

Performance Characteristics of Quicksort
Property 1: Quicksort uses about N2/2 comparisons in the
worst case.
The number of comparisons used for input that is already in
order (sorted) is
N + (N-1) + (N-2)+… + 2+1 = N(N+1)/2

The best case for quicksort is when each partition stage divides
the input exactly in half. This would make the number of
comparisons used by quicksort satisfy the divide-and-conquer
recurrence
CN = 2CN/2 + N
 The 2CN/2 covers the cost of sorting the two sub-lists;
 the N is the cost of examining each element, using the partitioning
pointer or the other.
50 By Dr. Farha Al Kharusi, DCS, SQU
Performance Characteristics of Quicksort
Property 2: Quicksort uses about 2N ln N comparisons
on the average.

1 N
CN = N +1+ å(Ck-1 + CN-k ) for N ³ 2
N k=1
C1 = C0 = 0

 The N+1 is the cost of partitioning.

 The rest comes from the observation that each element k is
likely to be the partitioning element with probability 1/N, after
which we are left with two random partitions of size k - 1 and
N-k

51 By Dr. Farha Al Kharusi, DCS, SQU

Reference
 Part of this slide set is prepared or/and extracted from
the following references:
 Data Structures & Algorithm Analysis in Java; Clifford A. Shaffer;
Dover Publications Inc.; 3rd edition, 2011 – Chapter 7
 Algorithms in C++; Robert Sedgewick; Addison-Wesely
Publishing Company Inc.; 3rd edition, 1998 – Chapter 7

52 By Dr. Farha Al Kharusi, DCS, SQU