Showing posts with label sorting. Show all posts
Showing posts with label sorting. Show all posts
Tuesday, April 12, 2011
Dance and Sorting == Awesome
This is brilliant! According to the poster, quicksort's coming soon. Can't wait.
Wednesday, March 16, 2011
Mergesort, Quicksort and Heapsort growth
These three sorting algorithms grow at approximately the same rate according to this graph.
They are O(nlogn) on average. Selection sort and bubble sort are O(n^2)
Here's a graph of the O(nlogn) algorithms.
They are O(nlogn) on average. Selection sort and bubble sort are O(n^2)
Here's a graph of the O(nlogn) algorithms.
Wednesday, March 9, 2011
Mergesort performance
Updated Sorting performance graph. My Quicksort implementation has weird variations. I'll have to look at the algorithm to figure out what's going on. Mergesort and Quicksort are practically equivalent for these set sizes.
Tuesday, March 1, 2011
Sorting in awk
I became interested in playing around with simple sorting algorithms and learning awk at the same time. Here is a quick roundup of the first three.
Divide and conquer rules the roost especially with large datasets. It would be nice to update it to graph these stats using gnuplot
Update: Graphing can now be done by graph_performance.sh. It generates a couple of stats files and the graph in tmp. It expects to find gnuplot in your path. Sort size can be changed by editing make_stats.sh directly.
$ bash make_stats.sh Doing 10 rounds of bubble_sort.awk with a dataset of 100 ints Average real time in seconds: 0.0151 Doing 10 rounds of quick_sort.awk with a dataset of 100 ints Average real time in seconds: 0.0476 Doing 10 rounds of selection_sort.awk with a dataset of 100 ints Average real time in seconds: 0.0048 Doing 10 rounds of bubble_sort.awk with a dataset of 500 ints Average real time in seconds: 0.3663 Doing 10 rounds of quick_sort.awk with a dataset of 500 ints Average real time in seconds: 0.2443 Doing 10 rounds of selection_sort.awk with a dataset of 500 ints Average real time in seconds: 0.0795 Doing 10 rounds of bubble_sort.awk with a dataset of 2500 ints Average real time in seconds: 9.7188 Doing 10 rounds of quick_sort.awk with a dataset of 2500 ints Average real time in seconds: 0.335 Doing 10 rounds of selection_sort.awk with a dataset of 2500 ints Average real time in seconds: 2.1556
Divide and conquer rules the roost especially with large datasets.
Selection sort is obviously much faster than bubblesort.
Probable reasons:
- Selection sort works on progressively smaller arrays with each iteration
- Comparisons in bubble sort are expensive compared to moving an element to the end of the array
- My implementation of bubble sort sucks.. (I'll check it again :( )
Update: Graphing can now be done by graph_performance.sh. It generates a couple of stats files and the graph in tmp. It expects to find gnuplot in your path. Sort size can be changed by editing make_stats.sh directly.
Subscribe to:
Posts (Atom)



