The present paper examines the robustness of the average case O (nlogn) complexity on K-sort, a new version of quick sort. In our first study we reconfirm this through computer experiments. A computer experiment is a series of runs of a code for various inputs. A deterministic computer experiment is one which produces identical results if the code is re-run for identical inputs. Our second study reveals that K-sort is the better choice for discrete uniform distribution U(1, 2,...k) inputs whereas quick sort is found better for continuous uniform distribution U(0, 1) inputs. Interestingly, increasing k which decreases the ties is good for quick sort but bad for K-sort.

Login Required

Please login to access the full PDF of this article.

🔑 Author Login