Bucket sort and its parallel implementations suffer for poor performance if the numbers are not uniformly distributed, because more numbers will fall into the same bucket for subsequent sorting....

Bucket sort and its parallel implementations suffer for poor performance if the numbers are not uniformly distributed, because more numbers will fall into the same bucket for subsequent sorting. Modify the algorithm so that the regions that each bucket collects are altered. This could be done as the algorithm is executed or before in a preprocessing step. Implement your algorithm.