Talk:AVL tree/C++: Difference between revisions

From Rosetta Code
Content added Content deleted
mNo edit summary
No edit summary
 
Line 10: Line 10:


The abbreviated Java and D samples also have the same problem in that they descend the tree to ascertain the depth. You must be careful not to do anything to upset the performance of RotateLeft and RotateRight.[[User:NNcNannara|NNcNannara]] ([[User talk:NNcNannara|talk]]) 12:26, 16 July 2016 (UTC)
The abbreviated Java and D samples also have the same problem in that they descend the tree to ascertain the depth. You must be careful not to do anything to upset the performance of RotateLeft and RotateRight.[[User:NNcNannara|NNcNannara]] ([[User talk:NNcNannara|talk]]) 12:26, 16 July 2016 (UTC)

[https://en.wikipedia.org/wiki/AVL_tree Wikipedia]'s pseudo code updates the balance factor in the RotateLeft and RotateRight functions. Note that they use a few if statements but do not descend the tree. I suppose that this is OK, although the Rosetta Code leaves the task of updating the balance factor to outer layers. The rotation algorithms supplied here are correct and of optimum performance.[[User:NNcNannara|NNcNannara]] ([[User talk:NNcNannara|talk]]) 08:27, 17 July 2016 (UTC)

Latest revision as of 15:59, 11 April 2017

The algorithms started out life in Pascal in a book called "Data Structures and Program Design" by Robert L Kruse. These were top/down (i.e. root to leaf) recursive algorithms without parent pointers. In 2006, the algorithms were made bottom/up (i.e. leaf to root) and the recursion was abolished (using parent pointers). These algorithms may be compared to the red/black set algorithms of STL. NNcNannara (talk) 12:30, 13 July 2016 (UTC)

I tested the elaborate code versus the shorter C++ version (on the main page) with the following results:

AVLtree insertions took: 00:00:02.5097816
Set insertions took: 00:00:00.0080027

Clearly, AVLtree is more like O(N2) or worse than O(log N). This is because it descends the tree during rotations (to adjust the balance factor). Set is clearly very fast at O(log N). The test was for 10000 insertions. If 100000 or 1000000 insertions are used Set rips through it but AVLtree stalls the machine.NNcNannara (talk) 12:45, 15 July 2016 (UTC)

The abbreviated Java and D samples also have the same problem in that they descend the tree to ascertain the depth. You must be careful not to do anything to upset the performance of RotateLeft and RotateRight.NNcNannara (talk) 12:26, 16 July 2016 (UTC)