The multilevel principle applied to sorting

Sorting algorithms are developed in the setting of iterative multilevel methods. These algorithms borrow aggregation techniques from algorithms used for the numerical solution of elliptic partial differential equations which are of optimal order in running time and storage space for structured problems. A computationally inexpensive preconditioner drives random data chosen from known distributions towards a special case for which the new sorting algorithms are of optimal order. © 1990 BIT Foundations.