Scribd
Upload
39 views

Elementary Sorting: For I in (0 ,... N) : Findindexofminimum Swap (I, I - Min)

Selection sort, insertion sort, and shellsort are three elementary sorting algorithms. Selection sort has O(n2) runtime and uses (n-1) + (n-2) + ... + 1 + 0 ≈ n2 comparisons and exchanges. Insertion sort's runtime depends on how sorted the input array is - from O(n) if sorted to O(n2) if reverse sorted. Shellsort improves on insertion sort by allowing exchanges of elements further than one position away.

Uploaded by

Andrew Lee
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
39 views

Elementary Sorting: For I in (0 ,... N) : Findindexofminimum Swap (I, I - Min)

Selection sort, insertion sort, and shellsort are three elementary sorting algorithms. Selection sort has O(n2) runtime and uses (n-1) + (n-2) + ... + 1 + 0 ≈ n2 comparisons and exchanges. Insertion sort's runtime depends on how sorted the input array is - from O(n) if sorted to O(n2) if reverse sorted. Shellsort improves on insertion sort by allowing exchanges of elements further than one position away.

Uploaded by

Andrew Lee
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Elementary Sorting

1
1

Selection Sort

for i in [0 ,... n ) :
find index of minimum
swap (i , i_min )
2

Uses (n 1) + (n 2) + . . . + 1 + 0 N2 comparisons and N exchanges,


giving O(n2 ) running time no matter the size/sort order of input array.

2
1

Insertion Sort

for i in [1 ,... n ) :
j = i
while ( a [j -1] > a [ j ]) :
swap (j , --j )

To sort a random array with distinct keys, insertion sort uses 41 N 2


comparisons and 41 N 2 exchanges on average, because given a random
data set, we can expect each entry to move halfway back.
If the array is sorted, insertion sort makes N 1 comparisons and 0
exchanges. If the array is reverse-sorted, it makes 12 N 2 comparisons and
21 N 2 exchanges.
In a partially sorted array with cN inversions, insertion sort runs in
linear time and the number of exchanges is equal to the number of inversions,
and the number of comparisons is exchanges + (N 1).

Shellsort

Idea: insertion sort that moves entries more than one position at a time by
h-sorting the array. An h-sorted array is h interleaved, sorted subsequences
- every pair separated by a distance of h is sorted.
Shellsort h-sorts an array for a decreasing sequence of values h. Use
insertion sort because large increments at the beginning means small subarrays, and when the increment grows smaller, the array is very nearly sorted.
The mathematical fact that makes shellsort possible is that a g-sorted array
is still g-sorted after being h-sorted.
Knuth recommends using the sequence xn = 3xn1 +1 for shellsort sizes,
and Sedgewick recommends a merge of (94!)(92i )+1 and 4i (32i )+1
that goes 1, 5, 19, 41, 109, 209, 505, 929, 2161, 3905 . . ..
The worst-case number of comparisons using the 3x+1 model is O(n3/2 ),
and nobody has determined an accurate model for the average-case running
time of the algorithm yet.

Elementary Sorting
4

Applications of Sorting

4.1

Shuffling

Shuffle sort: generate a random real number for each array entry, then sort.
Provided no duplicate values, this randomly permutes the array. However,
this is lower-bounded by O(n log n) due to the limit on comparison-based
sorting.
The Knuth shuffle is an algorithm that shuffles an array to a uniformly
random permutation in linear time.
2

for i in [0 ,... n ) :
swap (i , rand (0 , i ) )

4.2

Convex Hull

The convex hull of a set of N points is the smallest perimeter convex


fence enclosing the points. Equivalent definitions include the smallest
convex set containing all points, smallest area convex polygon enclosing the
points, or the convex polygon enclosing the points whose vertices are points
in the set.
Important properties of the convex hull:
1. Can traverse by making only CCW turns
2. The vertices appear in increasing order of polar angle with respect to
point p with the lowest y-coordinate.
One way to find the convex hull is the Graham scan:
1. Choose point p with smallest y-coordinate
2. Sort points by polar angle with p
3. Consider points in order: if one point leads to a clockwise turn with
any point after it in order, discard it and connect the vertex before it
with the connected vertex. This is very naturally implemented with a
stack.

You might also like