Memory memory complexity on not in place in on copy

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: rations in each pass – Main disadvantage to quicksort or heapsort – In-place variant exists, but large overhead In- ! Time Time complexity (worst case): O(n log n) log – For pre-sorted sequence, worse than insertion sort pre- ! ! ! ! Can be stable if merge() is stable Easier to implement than quicksort and heapsort In In practice slower that quicksort, but faster than heapsort heapsort Advantage: sequential access of data – Excellent for external (out-of-core) sorting (e.g. disk) (out-of(e.g. – Good cache usage 14 Merge Sort: improvements ! Inefficiency of merge() – Comparisons (i > mid) and (j > ub) Comparisons evaluate evaluate to FALSE most of the time Function merge (r, lb, mid, ub): memcpy(&tmp[lb], &r[lb], mid-lb+1); memcpy(&tmp[mid+1], &r[mid+1], ub-mid); i ← lb; j ← mid+1; lb; for for k ← lb to ub lb if if (i > mid) r[k] ← tmp[j++]; else else if (j > ub) r[k] ← tmp[i++]; tmp[i++]; else if (tmp[j] < tmp[i]) r[k] ← tmp[j++]; 15 else else r[k] ← tmp[i++]; Merge Sort: bitonic sort Use sentinel (as in insertion sort) Function bsmerge (r, lb, mid, ub): memcpy(&tmp[lb], &r[lb], mid-lb+1); j ← ub; for for k ← mid+1 to ub tmp[k] ← r[j]; // copy in reverse // j--; // // order i ← lb; j ← ub; lb; for for k ← lb to ub lb if if (tmp[j] < tmp[i]) r[k] ← tmp[j--]; else...
View Full Document

This document was uploaded on 03/06/2014 for the course ECE 368 at Purdue.

Ask a homework question - tutors are online