An Improvement on Tree Selection Sort
The standard Tree Selection Sort is an efficient sorting algorithm but requires extra storage for n-1 pointers and n items. The goal of this paper is to not only reduce the extra storage of Tree Selection Sort to n bits, but also keep the number of comparisons at nlogn+O(n). The improved algorithm makes at most 3n data movements. The empirical results show that the improved algorithm is efficient. In some cases, say moving one item requires at least 3 assignment operations, the algorithm is the fastest on average among known fast algorithms.
KeywordsData Movement Internal Node Space Requirement Selection Phase Sorting Algorithm
Unable to display preview. Download preview PDF.
- 1.J.W.J. Williams, Algorithm 232, Heapsort 3, Comm.ACM, 7, 1964, pp. 347–348.Google Scholar
- 2.R.W. Floyd, Algorithm 245, Treesort 3, Comm.ACM, 1964, p. 701.Google Scholar
- 4.E.H. Friend, Sorting on electronic computers, JACM 3(2), 1956, pp. 34–168.Google Scholar
- 5.D.E. Knuth, “The Art of Computer Programming Vol.3: Sorting and Searching”, Addison-Wesley, Reading, MA, 1973Google Scholar
- 7.C.J.H. McDiarmid and B.A. Reed, Building heaps fast, J. Algorithms 10, pp. 352–369.Google Scholar
- 8.S. Carlsson, A variant of heapsort with almost optimal number of comparisons, Inform. Process. Lett. 24, pp. 247–250.Google Scholar
- 9.G.H. Gonnet and J.I. Munro, Heaps on heaps, Proc. 9th ICALP, Aarhus, Denmark, July 12–16, 1982,pp. 282–291.Google Scholar
- 11.J.C. Chen, Proportion split sort, Nordic Journal of Computing 3(1996), pp. 271–279.Google Scholar
- 14.R. Sedgewick, Implementing quicksort programs. Communications of the ACM, 21(10), pp. 847–857, October, 1978.Google Scholar