• 0

Shell Sort Vs Heap Sort Vs Quick Sort


Question

i am having a debate as to which of the sort is more efficient in 2 cases:

when a 2d array of size 50, 100, or 1000 of randomly generated integers

and the same 2d array with all numbers sorted but only 2 indexes where their values have been exchanged.

for these 2 cases which of the 3 sorts seems more efficient.. i am pretty sure the quick sort is the best for the randomly generated array..

but what about the other case... the problem that is bugging me is that when the size is small the difference in the number of comparisons made with quick sort is very similar to the shell sort ...

any suggestions welcomed

:wacko:

Link to comment
https://www.neowin.net/forum/topic/51421-shell-sort-vs-heap-sort-vs-quick-sort/
Share on other sites

9 answers to this question

Recommended Posts

  • 0

In general, quicksort is the fastest known sort method.

However, if the array is presorted, or nearly presorted, then quicksort is no longer the best (the time taken is not much different from random data which is O(n log n)). Shell sort should be quite a bit quicker. Strangely enough, insertion sort and bubble sort - the two algorithms that are taught first 'cause they're so simple - do really well with presorted input (both O(n)). I suspect that in your case ("all numbers sorted but only 2 indexes where their values have been exchanged") insertion sort will be fastest, even faster than shell sort.

Of course I could be wrong ;)

BTW, Heap sort is never any good - don't use it :)

  • 0
  Quote
Big O notation isnt as exact as you are talking about.

Of course it's not exact. On the other hand, for large data sets, any difference in complexity (ie. big-oh) makes a huge difference. Anyway, for the case he's talking about (sorting an almost sorted array) I'm pretty sure shell sort, bubble sort and insertion sort are all O(n). Which makes big-oh notation somewhat useless really. But my guess is that insertion sort will perform the best, due to the simplicity of the algorithm.

  • 0

Also of note: according to my data structures book, shellsort (with Sedgewick's increments) performs better than quicksort when the data set is small (about 100 items or less).

The full results are posted below (the input data is random and the times are in seconds).

post-47-1038033898.gif

  • 0

yeah, what are they

i am doing data structure and i have done discrete mathematics

but what is discrete structures?? :D

from my program it seems like quick sort performed the best in all the cases.

heap sort is disastrous

its time just explodes ...

btw, here is the program which performs these sorts and counts the number of times each sort does a compare and a move ..

#include <stdio.h>
#include <stdlib.h>
#include <memory.h>

#define ARYSIZE 1000
#define ROW 5
#define COL 10
#define SWAP1 19
#define	SWAP2 20
#define NUMTOPRINT 50
#define S "Shell Sort"
#define H "Heap Sort"
#define Q "Quick Sort"

void printSortLine (char *str);
void createAry (int *ary);
void printAry (int *ary, int size);
void shellSort (int list [], int last, int *compare, int *moves);
void printCmpAndMov (int compare, int moves, char *str);
void swap (int *ary);
void heapSort (int list[], int last, int *compare, int *move);
void quickSort (int sortData [], int left, int right, int *compare, int *move);

int main (void)
{
	int ary[ARYSIZE], heapAry [ARYSIZE], quickAry [ARYSIZE], compare = 0, moves = 0;

	createAry (ary);
	memcpy (heapAry, ary, ARYSIZE * sizeof (int));
	memcpy (quickAry, ary, ARYSIZE * sizeof (int));

	printf("Array size:\t%d\n\n", ARYSIZE);

	printSortLine (S);
	compare = moves = 0;
	shellSort (ary, ARYSIZE, &compare, &moves);
	printAry (ary, ARYSIZE);
	printCmpAndMov (compare, moves, "Random list Shell Sort:");
	swap (ary);
	compare = moves = 0;
	shellSort (ary, ARYSIZE, &compare, &moves);
	printCmpAndMov (compare, moves, "Nearly ordered list Shell Sort:");

	printSortLine (H);
	compare = moves = 0;
	heapSort (heapAry, ARYSIZE - 1, &compare, &moves);
	printAry (heapAry, ARYSIZE);
	printCmpAndMov (compare, moves, "Random list Heap Sort:");
	swap (heapAry);
	compare = moves = 0;
	heapSort (heapAry, ARYSIZE - 1, &compare, &moves);
	printCmpAndMov (compare, moves, "Nearly ordered list Heap Sort:");

	printSortLine (Q);
	compare = moves = 0;
	quickSort (quickAry, 0, ARYSIZE - 1, &compare, &moves);
	printAry (quickAry, ARYSIZE);
	printCmpAndMov (compare, moves, "Random list Quick Sort:");
	swap (quickAry);
	compare = moves = 0;
	quickSort (quickAry, 0, ARYSIZE - 1, &compare, &moves);
	printCmpAndMov (compare, moves, "Nearly ordered list Quick Sort:");

	return 0;
}

void printSortLine (char *str)
{
	printf("\nPerforming %s....\n\n", str);
	return;
}

void createAry (int *ary)
{
	int i;

	for (i = 0; i < ARYSIZE; i++)
	{
  ary[i] = rand () % 1001;
	}
	return;
}

void printAry (int *ary, int size)
{
	int i, j;

	if (size <= NUMTOPRINT)
	{
  for (i = 0; i < ROW; i++)
  {
 	 for (j = 0; j < COL; j++)
 	 {
    printf("%4d   ", ary[(i * COL) + j]);
 	 }
 	 printf("\n");
  }
	}
	printf("\n");
	return;
}

void printCmpAndMov (int compare, int moves, char *str)
{
	printf("%-25s\t|compares = %6d  |  moves = %6d|\n", str, compare, moves);
	return;
}

void swap (int *ary)
{
	int temp;

	temp = ary[SWAP1];
	ary [SWAP1] = ary [SWAP2];
	ary [SWAP2] = temp;
	return;
}

void shellSort (int list [], int last, int *compare, int *moves)
{
	int hold, incre, curr, walker;

	incre = last / 2;
	while (incre != 0)
	{
  for (curr = incre; curr <= last; curr++)
  {
 	 hold = list [curr];
 	 *moves += 1;
 	 walker = curr - incre;
 	 while (((*compare)++,walker >= 0 && hold < list [walker]))
 	 {
    list [walker + incre] = list [walker];
    *moves += 1;

    walker = (walker - incre);
 	 }

 	 list [walker + incre] = hold;
 	 *moves += 1;
  }
  incre /= 2;
	}
	return;
}


void heapSort (int list[], int last, int *compare, int *move)
{
	void reheapUp (int heap[], int newElem, int *compare, int *move);
	void reheapDown (int heap[], int current, int last, int *compare, int *move);

	int sorted, holdData, walker;

	for (walker = 1; walker <= last; walker++)
	{
  reheapUp (list, walker, compare, move);
	}

	sorted = last;
	while (sorted > 0)
	{
  holdData = list [0];
  list [0] = list [sorted];
  list [sorted] = holdData;
  *move += 3;
  sorted--;
  reheapDown (list, 0, sorted, compare, move);
	}
	return;
}

void reheapUp (int *heap, int newNode, int *compare, int *move)
{
	int parent, hold;

	if (newNode)
	{
  parent = (newNode - 1)/2;
  if ( *compare += 1, heap [newNode] > heap [parent])
  {
 	 hold = heap[parent];
 	 heap[parent] = heap[newNode];
 	 heap[newNode] = hold;
 	 *move += 3;
 	 reheapUp (heap, parent, compare, move);
  }
	}
	return;
}

void reheapDown (int *heap, int root, int last, int *compare, int *move)
{
	int hold, leftKey, rightKey, largeChildKey, largeChildIndex;

	if ((root * 2 + 1) <= last)
	{
  leftKey = heap [root * 2 + 1];
  *move += 1;

  if ((root * 2 + 2) <= last)
  {
 	 rightKey = heap [root * 2 + 2];
 	 *move += 1;
  }
  else
  {
 	 rightKey = -1;
  }

  if (leftKey > rightKey)
  {
 	 largeChildKey = leftKey;
 	 largeChildIndex = root * 2 + 1;
  }
  else
  {
 	 largeChildKey = rightKey;
 	 largeChildIndex = root * 2 + 2;
  }

  if (*compare += 1, heap[root] < heap [largeChildIndex])
  {
 	 hold = heap [root];
 	 heap [root] = heap [largeChildIndex];
 	 heap [largeChildIndex] = hold;
 	 *move += 3;
 	 reheapDown (heap, largeChildIndex, last, compare, move);
  }
	}
	return;
}

void quickSort (int sortData [], int left, int right, int *compare, int *move)
{
#define MIN_SIZE 16

	void quickInsertion (int sortData[], int first, int last, int *compare, int *move);
	void medianLeft (int sortData[], int left, int right, int *compare, int *move);

	int sortLeft;
	int sortRight;
	int pivot;
	int hold;

	if ( (right - left) > MIN_SIZE)
	{
  medianLeft (sortData, left, right, compare, move);
  pivot = sortData [left];
  *move += 1;
  sortLeft = left + 1;
  sortRight = right;

  while (sortLeft <= sortRight)
  {
 	 while (*compare += 1, sortData [sortLeft] < pivot)
    sortLeft = sortLeft + 1;

 	 while (*compare += 1, sortData[sortRight] >= pivot)
    sortRight = sortRight - 1;

 	 if (sortLeft <= sortRight)
 	 {
    hold = sortData [sortLeft];
    sortData [sortLeft] = sortData [sortRight];
    sortData [sortRight] = hold;
    *move += 3;
    sortLeft = sortLeft + 1;
    sortRight = sortRight - 1;
 	 }
  }

  sortData [left] = sortData [sortLeft - 1];

  sortData [sortLeft - 1] = pivot;

  *move += 2;

  if (left < sortRight)
 	 quickSort (sortData, left, sortRight - 1, compare, move);
  if (sortLeft < right)
 	 quickSort (sortData, sortLeft, right, compare, move);
	}
	else
  quickInsertion (sortData, left, right, compare, move);
	return;
}

void quickInsertion (int sortData[], int first, int last, int *compare, int *move)
{
	int current;
	int hold;
	int walker;

	for (current = first + 1; current <= last; current++)
	{
  hold = sortData [current];
  *move += 1;
  walker = current - 1;
  while (*compare += 1, walker >= first && hold < sortData [walker])
  {
 	 sortData [walker + 1] = sortData [walker];
 	 *move += 1;
 	 walker = walker - 1;
  }
  sortData[walker + 1] = hold;
  *move += 1;
	}
	return;
}

void medianLeft (int sortData[], int left, int right, int *compare, int *move)
{
	int mid;
	int hold;

	mid = (left + right) / 2;

	if (*compare += 1, sortData [left] > sortData [mid])
	{
  hold = sortData [left];
  sortData [left] = sortData [mid];
  sortData [mid] = hold;  
  *move += 3;
	}
	if (*compare += 1, sortData [left] > sortData [right])
	{
  hold = sortData [left];
  sortData [left] = sortData [right];
  sortData [right] = hold;
  *move += 3;
	}
	if (*compare += 1, sortData [mid] > sortData [right])
	{
  hold = sortData [mid];
  sortData [mid] = sortData [right];
  sortData [right] = hold;
  *move += 3;
	}

	hold = sortData [left];
	sortData [left] = sortData [mid];
	sortData [mid] = hold;
	*move += 3;

	return;
}

here is the output file

  Quote
output# 1

Array size:     50

Performing Shell Sort....

 18     41    117    126    150    153    225    258    273    291

292    303    328    329    359    370    404    431    440    442

449    467    474    491    527    590    642    656    660    675

698    699    700    704    705    709    757    788    811    823

862    868    876    885    893    899    931    936    952    993

Random list Shell Sort:         |compares =    372  |  moves =    580|

Nearly ordered list Shell Sort: |compares =    209  |  moves =    417|

Performing Heap Sort....

 18     41    117    126    150    153    225    258    273    291

292    303    328    329    359    370    404    431    440    442

449    467    474    491    527    590    642    656    660    675

698    699    700    704    705    709    757    788    811    823

862    868    876    885    893    899    931    936    952    993

Random list Heap Sort:          |compares =    269  |  moves =   1140|

Nearly ordered list Heap Sort:  |compares =    361  |  moves =   1467|

Performing Quick Sort....

 18     41    117    126    150    153    225    258    273    291

292    303    328    329    359    370    404    431    440    442

449    467    474    491    527    590    642    656    660    675

698    699    700    704    705    709    757    788    811    823

862    868    876    885    893    899    931    936    952    993

Random list Quick Sort:         |compares =    294  |  moves =    306|

Nearly ordered list Quick Sort: |compares =    155  |  moves =    105|

Output# 2

Array size:     100

Performing Shell Sort....

Random list Shell Sort:         |compares =    916  |  moves =   1425|

Nearly ordered list Shell Sort: |compares =    510  |  moves =   1019|

Performing Heap Sort....

Random list Heap Sort:          |compares =    642  |  moves =   2756|

Nearly ordered list Heap Sort:  |compares =    910  |  moves =   3685|

Performing Quick Sort....

Random list Quick Sort:         |compares =    712  |  moves =    731|

Nearly ordered list Quick Sort: |compares =    410  |  moves =    213|

Output# 3

Array size:     1000

Performing Shell Sort....

Random list Shell Sort:         |compares =  15493  |  moves =  23508|

Nearly ordered list Shell Sort: |compares =   8015  |  moves =  16030|

Performing Heap Sort....

Random list Heap Sort:          |compares =   9733  |  moves =  43757|

Nearly ordered list Heap Sort:  |compares =  15325  |  moves =  61177|

Performing Quick Sort....

Random list Quick Sort:         |compares =  10982  |  moves =   9177|

Nearly ordered list Quick Sort: |compares =   7032  |  moves =   2205|

  • 0

quick sort is the fastest when u have huge test cases. (i learnt that half way through a programming competition, bad day that was)

shell sort and merge sort would serve ur purpose the rest of the time.

heapsort, bubble sort, selection sort are practically useless.

thats wat i feel.

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • WinSCP 6.5.2 by Razvan Serea  WinSCP is an open source free SFTP client, FTP client, WebDAV client and SCP client for Windows. Its main function is file transfer between a local and a remote computer. Beyond this, WinSCP offers scripting and basic file manager functionality. WinSCP features: Graphical user interface Translated into several languages Integration with Windows (drag&drop, URL, shortcut icons) U3 support All common operations with files Support for SFTP and SCP protocols over SSH-1 and SSH-2 and plain old FTP protocol Batch file scripting and command-line interface Directory synchronization in several semi or fully automatic ways Integrated text editor Support for SSH password, keyboard-interactive, public key and Kerberos (GSS) authentication Integrates with Pageant (PuTTY authentication agent) for full support of public key authentication with SSH Explorer and Commander interfaces Optionally stores session information Optionally supports portable operation using a configuration file in place of registry entries, suitable for operation from removable media WinSCP 6.5.2 changelog: Thumbnail view in file panels. Three selectable sizes of toolbar icons, showing slightly larger size by default. Switching to Segoe UI font with slightly larger size. Improvements to Synchronization checklist window, including resolving file moves and pushing synchronization to background queue. Ongoing local delete operation can be moved to a background queue. Optimized working with large local directories. Compatibility with new OneDrive WebDAV interface. Dark theme for session tabs. Improvements to S3 support, including more options to authentication and display and modification of S3 file/object tags. List of all changes. Download: WinSCP 6.5.2 | 11.6 MB (Open Source) Download: WinSCP MSI | 28.7 MB Download: Standalone Executable | 8.4 MB Link: WinSCP Home page | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • QOwnNotes 25.6.2 by Razvan Serea QOwnNotes is a open source (GPL) plain-text file notepad with markdown support and todo list manager for GNU/Linux, Mac OS X and Windows, that (optionally) works together with the notes application of ownCloud (or Nextcloud). So you are able to write down your thoughts with QOwnNotes and edit or search for them later from your mobile device (like with CloudNotes) or the ownCloud web-service. The notes are stored as plain text files and you can sync them with your ownCloud sync client. Of course other software, like Dropbox, Syncthing, Seafile or BitTorrent Sync can be used too. Features: the notes folder can be freely chosen (multiple note folders can be used) sub-string searching of notes is possible and search results are highlighted in the notes application can be operated with customizable keyboard shortcuts external changes of note files are watched (notes or note list are reloaded) older versions of your notes can be restored from your ownCloud server trashed notes can be restored from your ownCloud server differences between current note and externally changed note are showed in a dialog markdown highlighting of notes and a markdown preview mode notes are getting their name from the first line of the note text (just like in the ownCloud notes web-application) and the note text files are automatically renamed, if the the first line changes compatible with the notes web-application of ownCloud and mobile ownCloud notes applications compatible with ownCloud's selective sync feature by supporting an unlimited amount of note folders with the ability to choose the respective folder on your server manage your ownCloud todo lists (ownCloud tasks or Tasks Plus / Calendar Plus) or use an other CalDAV server to sync your tasks to encryption of notes (AES-256 is built in or you can use custom encryption methods like Keybase.io (encryption-keybase.qml) or PGP (encryption-pgp.qml)) dark mode theme support theming support for the markdown syntax highlighting all panels can be placed wherever you want, they can even float or stack (fully dockable) support for freedesktop theme icons, you can use QOwnNotes with your native desktop icons and with your favorite dark desktop theme support for hierarchical note tagging and note subfolders support for sharing notes on your ownCloud server portable mode for carrying QOwnNotes around on USB sticks Evernote import QOwnNotes is available in many different languages like English, German, French, Polish, Chinese, Japanese, Russian, Portuguese, Hungarian, Dutch and Spanish Changes in QOwnNotes 25.6.2: The Find action dialog is now working again (for #3294) Added more French translation (thank you, jd-develop) Download: QOwnNotes 25.6.2 | 37.3 MB (Open Source) Download: QOwnNotes for Other Operating Systems View: QOwnNotes Home Page | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • So it’s fine he scams people to troll others? 🙄 This is a repeat of his watch fiasco.
    • That explains my recent issues with classic Outlook. I can see the FORMS2 folder has resolved the issue on my system.
    • Brave 1.79.126 by Razvan Serea Brave Browser is a lightning-fast, secure web browser that stands out from the competition with its focus on privacy, security, and speed. With features like HTTPS Everywhere and built-in tracker blocking, Brave keeps your online activities safe from prying eyes. Brave is one of the safest browsers on the market today. It blocks third-party data storage. It protects from browser fingerprinting. And it does all this by default. Speed - Brave is built on Chromium, the same technology that powers Google Chrome, and is optimized for speed, providing a fast and responsive browsing experience. Brave Browser also features Brave Rewards, a system that rewards users with Basic Attention Tokens (BAT) for viewing opt-in ads. This innovative system provides an alternative revenue model for content creators and a way to support the Brave community. Brave 1.79.126 changelog: Fixed crash when clicking on shields panel in certain cases. Fixed crash on navigation when a Google sign-in request has been triggered. (#46769) Fixed crash when Leo generates empty list when clicking on “Suggest questions” while using BYOM (Bring Your Own Model). (#46843) Fixed issue importing from Trezor wallet when Trezor Connect is outdated. (#46660) Upgraded Chromium to 137.0.7151.119. (#46911) Download: Brave Browser 64-bit | 1.2 MB (Freeware) Download: Brave Browser 32-bit View: Brave Homepage | Offline Installers | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
  • Recent Achievements

    • First Post
      Fuzz_c earned a badge
      First Post
    • First Post
      TIGOSS earned a badge
      First Post
    • Week One Done
      slackerzz earned a badge
      Week One Done
    • Week One Done
      vivetool earned a badge
      Week One Done
    • Reacting Well
      pnajbar earned a badge
      Reacting Well
  • Popular Contributors

    1. 1
      +primortal
      698
    2. 2
      ATLien_0
      279
    3. 3
      Michael Scrip
      208
    4. 4
      +FloatingFatMan
      195
    5. 5
      Steven P.
      130
  • Tell a friend

    Love Neowin? Tell a friend!