CS29002 Algorithms Laboratory

Assignment No: 7
Last date of submission: 07–September–2016

So far, you have seen running times of algorithms expressed in the big-Oh notation. In order to put these
theoretical metrics to practical tests, we take two sorting algorithms in this assignment (quick sort and merge
sort), and measure their actual running times on inputs of various sizes. Let n denote the size of an array A
(of integers) which we want to sort.
The performance of quick sort is very sensitive to the initial distribution of A and also to the choice of the
pivot. You will work with random, sorted, and almost sorted arrays. Also, the following ways of choosing
the pivot for partitioning should be experimented with.

FIRST Choose the first element A[0] as the pivot.

RANDOM Choose A[r] as the pivot for a random r ∈ {0, 1, 2, . . . , n − 1}.

MEDIAN OF THREE Choose r, s,t ∈ {0, 1, 2, . . . , n − 1}, and take the median of A[r], A[s], A[t] as the
pivot. Use both the following choices for r, s,t.

(1) r = 0, s = n/2, and t = n − 1.

(2) r = n/4, s = n/2, and t = 3n/4.

• Write a function quicksort(A, n, pivot type) to sort an array A with n (non-negative) integers. The third
argument pivot type indicates how you choose the pivot for partitioning: 0 means FIRST, 1 means
RANDOM, 2 means MEDIAN OF THREE (1), and 3 means MEDIAN OF THREE (2).

The working of merge sort is not theoretically sensitive to the distribution of A or other parameters (merge
sort has nothing like a pivot).

• Write a function mergesort(A, n) to sort an array A of n (non-negative) integers.

The main() function

(a) For n = 10k , k = 4, 5, 6, 7, and for each choice of the pivot type, repeat the following:

Populate an array A with n random integers in the range 0 to 109 −1. The random choices of elements
may lead to repetitions. You do not need to avoid repetitions. Sort A by calling quicksort. Now, A
is sorted. Call quicksort again on this sorted array. Now, swap a few elements in A (take k = n/100,
and swap k randomly chosen pairs of A). This gives you an almost sorted array. Run quicksort again
on this array. Report the times taken by the three calls. Do not include the array-preparation times.

(b) For n = 10k , k = 4, 5, 6, 7, repeat the following:

Populate A with n random integers in the range 0 to 109 − 1. Sort A by calling mergesort. Call
mergesort again on this sorted array. Take k = n/100, and swap k randomly chosen pairs of A. Run
mergesort again on this almost sorted array. Report the times taken by the three calls.

A random integer in the range [0, 109 − 1] can be generated by the call (#include <stdlib.h>):

A[i] = rand() % 1000000000;

Different runs of your program should handle different random sequences. In order to achieve that, write the
following line at the beginning of your main() function. You need to include <stdlib.h> and <time.h>.

srand((unsigned int)time(NULL));

— Page 1 of 2 —
Finally, this is how you can measure (calendar) times in C/C++ programs (include <time.h>).

clock_t c1, c2;

double runtime;

c1 = clock();
- - - Code for which the running time is to be measured - - -
c2 = clock();
runtime = (double)(c2 - c1) / (double)CLOCKS_PER_SEC; /* Time in seconds */

Sample output

Present your output in the following format. You may compile your program with the -O2 optimization flag.

+++ Performance of Quick Sort

n Pivot type Random Sorted Almost sorted

10000 FIRST 0.001 0.002 0.002

10000 RANDOM 0.001 0.001 0.001
10000 MEDIAN OF THREE 1 0.001 0.002 0.001
10000 MEDIAN OF THREE 2 0.001 0.000 0.001

100000 FIRST - - -
100000 RANDOM - - -
100000 MEDIAN OF THREE 1 - - -
100000 MEDIAN OF THREE 2 - - -

1000000 FIRST - - -
1000000 RANDOM - - -
1000000 MEDIAN OF THREE 1 - - -
1000000 MEDIAN OF THREE 2 - - -

10000000 FIRST - - -
10000000 RANDOM - - -
10000000 MEDIAN OF THREE 1 - - -
10000000 MEDIAN OF THREE 2 - - -

+++ Performance of Merge Sort

n Random Sorted Almost sorted

10000 0.001 0.001 0.001

100000 - - -
1000000 - - -
10000000 - - -

Submit a single C/C++ source file. Do not use global/static variables.

— Page 2 of 2 —