Program of sorting algorithms , C/C++ Programming

The program sorting.cpp contains a main function for testing the operation of several sort algorithms over various data sizes and data set organisations. The program understands the following options:

-i insertion sort

-s selection sort (the default)

-q quicksort

-a (almost) sorted data set

-v reverse-sorted data set

-r random data set (the default)

n generate a data set of size n (default 10000)

Background

In part 3 you will need to gather information about the number of times that various parts of the code is executed. On Unix-based systems (including Linux and cygwin), the simplest approach is to use execution profiling tools. To gather profiling data, first compile the program with the -pg option, which will cause it to write information about its execution into a file (called gmon.out).

Then each time you run the program, you can use the profile analysis tool (gprof) to report on the execution counts of each called function.

g++ -pg -o sort sort.cpp

./sort // run with default arguments

gprof -b // -b means "brief" output

./sort -qr 1000000 // quicksort on one million random values

gprof -b // profile for second run

...

If the system that you're using doesn't support profiling, a work around is to add code to the program to increment a counter at each point of interest, then print the values of the counters at the end of execution.

Selection Sort

1. Implement the selectionSort and insertionSort functions. The program checks that the final list is sorted correctly, so you can test your code using try with task  sorting.

Quicksort

2. Implement the quickSort function. The real work of quicksort happens in the partition operation, so it's probably best to define a separate function to do the partitioning.

 Profiling

3. Gather data about the number of times that the innermost loops of each sort algorithm is executed. A simple way to count inner loop executions is to call a "do nothing" function from the inner loop, then look for the execution count of that function in the profiler output.

void inner() {} // for gprof to count

void ...Sort() {

for (...) {

for (...) {

...

inner();

}

}

}

When you have profiling working to your liking, run the program and record the inner loop counts for each combination of algorithm (-i, -s, and -q) and data organisation (-a, -v, and -r), beginning with a data size of 1000 and increasing progressively until execution takes around 30 seconds. A suitable set of size increments would be 1000, 2000, 5000, 10000, 20000 and so on.

For each run, compute a normalised execution count by dividing the actual count by the data set size. For example, if quicksort with a random data set containing 10,000 elements executed its inner loop 161,619 times, then the normalised execution count would be 161610/10000, or about 16.1

Finally, plot the 9 sets of data (for the options -ia, -iv, -ir, -sa, -sv, -sr, -qa, -qv, and -qr) as separate lines on a single set of axes, with data set size on the horizontal axis and normalised execution count on the vertical axis. You'll need to choose scales so that the full range of values can be displayed, and label the lines so you know which one is which. You may also find it worthwhile using a log-log plot to better show the values over the full range.

Improving Quicksort

4. A simple implementation of quicksort is likely to perform badly for certain data sets. For example, the implementation used as an example in lectures performs very poorly if the data set is in reverse-sorted order.

The poor performance can be avoided by choosing a better pivot for the partitioning step.

Common strategies include selecting a random element or finding the median of the first, middle, and last elements. In any case, the chosen pivot is swapped with the last element before proceeding to do the partition as before.

Read about and implement one of the improvements to quicksort and show (by adding its profile data to the graph of part 3) how much improvement you have achieved.

Posted Date: 3/15/2013 3:21:38 AM | Location : United States







Related Discussions:- Program of sorting algorithms , Assignment Help, Ask Question on Program of sorting algorithms , Get Answer, Expert's Help, Program of sorting algorithms Discussions

Write discussion on Program of sorting algorithms
Your posts are moderated
Related Questions
A string S is said to be "Super ASCII", if it contains the character frequency equal to their ascii values. String will contain only lower case alphabets (''a''-''z'') and the asci

I need Website with built in scraper Project Description: The coding skills mentioned above are just possibly ones that will be used. Skills required are ASP, C++ Programm

The ability to verify at run time the type of an object by using the typed operator or the dynamic_cast operator.

A palindrome is a string that reads the same from both the ends. Given a string S convert it to a palindrome by doing character replacement. Your task is to convert S to palindrome


Define a complex number in c program: class complex                                 {                                   private:                                 in

In this project, you will write a software for the Registrar's office of a university called YouOIT to maintain information about students and courses at the university. The reg

Write a program that reads data into an array of type int. You must use an array. The numbers input are attendance numbers at a conference. Keep a running average, number of events

Program to define an array in c: Write a program to define an array and print the value of array. void main() { int a[10]={0,11,21,34,44,75,46,57,88,89},i,j,k; clr

For this program you will add and test 2 new member functions to the IntSLList class posted on the website. The two member functions are: insertByPosn(int el, int pos) Assuming t