Lecture 24 Recursive sorting II FIT 1008&2085 Introduction to Computer Science
COMMONWEALTH OF AUSTRALIA Copyright Regulations 1969 WARNING This material has been reproduced and communicated to you by or on behalf of Monash University pursuant to Part VB of the Cop yright Act 1968 (the Act).The material in this communication may be subject to copyright under the Act.Any further reproduction o communication of this material by you may be the subject of copyright protection under the Act. Do not remove this notice. Overview • To review what a “divide and conquer” algorithm is
• To review in more depth two different “divide and
Combine 3 Divide and Conquer: Sorting General Idea
• Merge Sort has a simple split and a elaborate combine
• Quick Sort has a elaborate split and a simple combine
4 Quicksort Top-10 algorithms 20th century (SIAM)
5 Quick Sort • Partition the list • Sort the first part (recursively) • Sort the second part (recursively) • (Combine: there is nothing to do!)
6 Partition • Choose an item in the list, called it the pivot. • The first part consists of all those items which are less than the pivot. • The second part consists of all those items larger than or equal to the pivot (except the pivot). so we need to do some shuffling around to organise the data after we have chosen a pivot
7 after doing the partition we have all the data less than the pivot, the pivot, then all the data greater than the pivot then we recursively sort each section based on this partition. why the combination step is trivial as every element in the first section of the partition is smaller than every element in the second section of the partition; this renders the merge operation as trivial.
Partition x<p p p <= x Sort should read x >= p Sort
• Partition: Elaborate, based on a pivot p.
• Combination: Simple append, pivot in the middle.
8 Ideally, what should the pivot be? D - the middle element of the output list as the output list is sorted therefore making the middle element of that list the median the reason why the input list is undesireable is because we will be shuffling items around the median is chosen so that we have a divide by 2 division giving us a logarithmic time complexity.
A) The smallest element of the list.
B) The largest element of the list.
C) The middle element of the input list.
D) The middle element of the output list.
E) Something else.
F) It actually makes no difference.
9 Given a list of size N, how efficiently can the median be computed?
A) O(1)
B) O(log N) the calculate the median element of a list it is a O(N) operation as we need to look at each part of the data in the input set
C) O(N)
D) O(N log N) In practise we might not pick exactly the median each time we want to divide as this will be come costly instead we can choose an element that will give approximately.a half-and-half split thus we can pick the median of a subset of the lists elements
E) O(N^2)
11 Example Partition
array: 5 89 35 14 24 15 37 13 20 7 70
start:0 end:10
13 Example Partition
array: 5 89 35 14 24 15 37 13 20 7 70
Randomly choose a pivot, which
happens to be in the middle
14 Example Partition
array: 5 89 35 14 24 15 37 13 20 7 70
partition: 5 89 35 14 24 15 37 13 20 7 70
result 7 14 5 13 15 35 37 89 20 24 70
pivot position: 4 note that the pivot defines the boundaries
sort first half (using QS), 15sort second half (using QS) Quicksort
16 Quicksort
17 p swap with first element
p index k
p x<p p >= x index increases if necessary index k always increases
p x<p p >= x
x<p p p >= x
18 Example Partition randomly pick element in position 5
array: 5 89 35 14 24 15 37 13 20 7 70
15 89 35 14 24 5 37 13 20 7 70
19 Example Partition
15 89 35 14 24 5 37 13 20 7 70
k:1
20 Example Partition index:0
15 89 35 14 24 5 37 13 20 7 70
k:1
21 Example Partition index:0
15 89 35 14 24 5 37 13 20 7 70
k:2
22 Example Partition index:0
15 89 35 14 24 5 37 13 20 7 70
k:3
23 Example Partition index:1
15 14 89 35 89 14 24 5 37 13 20 7 70
k:3
24 Example Partition index:1
15 89 14 35 89 24 5 37 13 20 7 70
k:4
25 Example Partition index:1
15 89 14 35 89 24 5 37 13 20 7 70
k:5
26 Example Partition index:2
15 89 14 5 89 24 35 37 13 20 7 70
k:5
27 Example Partition index:2
15 89 14 5 89 24 35 37 13 20 7 70
k:6
28 Example Partition index:2
15 89 14 5 89 24 35 37 13 20 7 70
k:7 etc…
29 Example Partition index:4
15 89 14 5 13 7 35 37 89 20 24 70
k:11
30 Example Partition index:4
15 7 89 14 5 13 15 7 35 37 89 20 24 70
x < 15 x >= 15
After last swap, pivot is in the
correct position
31 Example Partition
7 14 5 13 15 35 37 89 20 24 70
quick_sort quick_sort 7 5 13 14 20 24 35 37 70 89
32 p swap with first element
p index k
p x<p p >= x index
p x<p p >= x
x<p p p >= x
33 34 37 Found an element that belongs in 1… index swap and update index to maintain invariant
38 What is the time complexity of the partition method?
A) O(log N)
B) O(N)
C) O(N log N)
D) O(N2)
39 What is the best-case time complexity of quicksort?
A) O(log N)
B) O(N)
C) O(N log N)
D) O(N2)
41 Quicksort: Number of partitions depends on the pivot
p Partition x<p p p <= x
pivot
Best case: The size of the problem is reduced by half with
every partition Worst case: The size of the problem is reduced by 1 with every partition 43 Quick sort’s best case partition is O(n) n n
n height n/2 n/2
n is n/4 n/4 n/4 n/4 O(log n n) n/8 n/8 n/8 n/8 n/8 n/8 n/8 n/8
Running time in the best case: O(n log n)
44 Quick sort’s worst case
n partition is O(n) n n-1 n-1 n-2 n-2 n-3 n-3
... n-4
n (n+1) _______ 2
Running time in the worst case: O(n2)
45 Summary
Best case Worst case
Quicksort O(n log n) O(n2)
Mergesort O(n log n) O(n log n)
How common is quicksort’s worst case?
Not too common if choosing a random pivot. 51 Summary • Divide and Conquer and Recursive Algorithms (for sorting).
