Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


JT
Posts:
1,434
Registered:
4/7/12


Re: Just finished the fastest ever, general purpose sorting algorithm.
Posted:
Jan 5, 2013 5:22 PM


On 5 Jan, 18:39, forbisga...@gmail.com wrote: > http://stackoverflow.com/questions/3074861/binarysortalgorithmi > > Algorithmi? That's sorta correct. it points to: > > http://www.brpreiss.com/books/opus5/html/page487.html > > It says: > Whereas a linear search requires O(n) comparisons in the worst case, a binary search only requires comparisons > > and gives this caveat: > The number of comparisons required by the straight insertion sort is in the worst case as well as on average. Therefore on average, the binary insertion sort uses fewer comparisons than straight insertion sort. On the other hand, the previous section shows that in the best case the running time for straight insertion is O(n). Since the binary insertion sort method always does the binary search, its best case running time is . Table summarizes the asymptotic running times for the two insertion sorts. > > (sorry that didn't all copy. quicksort is probably better in most cases.) I am not sure if you looked at my countsort algorithm using arrays, it is further down in sci.math. But my algorithms do not compare anything it just read in the values in an ordered fashion, it just look for if the digit value is zero or one. So each digit have to be stepped thru but there is none comparissons. See my JT countsort algorithm. I just move in the valuesand read out the table. This would be so as fast relative quicksort as JT countsort is with the difference that this algorith actually sort values of any size or length This probably the only algorithm being used for sorting in future since it is general purpose.



