On Hash-Based Work Distribution Methods for Parallel Best-First Search
This work addresses communication inefficiencies in parallel search algorithms, which is an incremental improvement for AI planning and optimization domains.
The paper tackled the high communication overhead in hash-based parallel best-first search algorithms by proposing Abstract Zobrist hashing, which uses feature projection to increase locality and reduce transfers, showing performance improvements with hand-coded functions and further gains with GRAZHDA* in domain-independent planning.
Parallel best-first search algorithms such as Hash Distributed A* (HDA*) distribute work among the processes using a global hash function. We analyze the search and communication overheads of state-of-the-art hash-based parallel best-first search algorithms, and show that although Zobrist hashing, the standard hash function used by HDA*, achieves good load balance for many domains, it incurs significant communication overhead since almost all generated nodes are transferred to a different processor than their parents. We propose Abstract Zobrist hashing, a new work distribution method for parallel search which, instead of computing a hash value based on the raw features of a state, uses a feature projection function to generate a set of abstract features which results in a higher locality, resulting in reduced communications overhead. We show that Abstract Zobrist hashing outperforms previous methods on search domains using hand-coded, domain specific feature projection functions. We then propose GRAZHDA*, a graph-partitioning based approach to automatically generating feature projection functions. GRAZHDA* seeks to approximate the partitioning of the actual search space graph by partitioning the domain transition graph, an abstraction of the state space graph. We show that GRAZHDA* outperforms previous methods on domain-independent planning.