A Statistical Analysis of Bubble Sort in terms of Serial and Parallel Computation

Abstract

In some recent papers, the weight based statistical bounds have arguably explained time complexity better than the count based mathematical bounds. This is definitely true for average case where for an arbitrary code it is difficult to identify the pivotal operation or pivotal region in the code for taking the expectation and/or when the probability distribution, over which expectation is taken, becomes unrealistic over the problem domain. In worst case, it can certify whether a mathematical bound is conservative or not. Here we revisit the results on Bubble sort in sequential mode and make an independent study of the same algorithm in parallel mode using statistical bound