Parallel branch-and-bound search in Parlog

Abstract

The concurrent logic languages, of which Parlog is one, have been promoted as a new generation of software languages specifically designed for parallel programming. This paper investigates their apphcation to a search problem commonly used as an illustration of artificial intelligence techniques, the 8-puzzle. It notes that programs written in the concurrent logic languages which do not pay attention to the parallehsm can fall into two possible traps: either there is little real parallelism in them due to data dependencies, or there is too much parallelism and any practical architecture will be overwhelmed. A solution which controls the parallehsm using user-defined priorities is proposed. This solution has the advantage of being architecture-independent.

Search in concurrent logic languages

“…where each node in the tree has its own bound, but bounds are distributed through the search tree. This contrasts to the global bound version of [ 16) where there was a single upper bound held by a single process for the whole. The global bound version, though simpler, runs into the problem that the single processor holding the bound becomes a bottle-neck as all node expansions need to access it to find the latest upper bound, and all that find a lower-cost solution need to access it to update the bound.…”

“…The search order will be whatever is inherited from the underlying concurrent logic system, though it could be determined by the priority mechanism introduced in [16] and discussed later in this paper.…”