An efficient algorithm for the nearest smallers problem on distributed shared memory systems with applications

We present a simple and efficient algorithm for the nearest smallers problem (NSP) (Berkman et al., 1988) on a distributed shared memory (DSM) system with applications to problems from diverse areas. We adopt the block distributed memory (BDM) model of computation as described in (Jaja and Ryn, 1996). To the best of our knowledge this is the first known algorithm for the NSP on DSM systems. Since the NSP is fundamental in many problems, a solution for it on DSM systems implies DSM-based solutions for a variety of problems in diverse areas, as discussed in this paper. Parallel algorithms known so far for the NSP are based on shared memory systems and are therefore less scalable than our algorithm