Multi-threaded Explicit State Space Exploration with State Reconstruction

This article introduces a parallel state space exploration algorithm for shared memory multi-core architectures using state compression and state reconstruction to reduce memory consumption. The algorithm proceeds in rounds each consisting of three phases: concurrent expansion of open states, concurrent reduction of potentially new states, and concurrent duplicate detection. An important feature of the algorithm is that it requires little inter-thread synchronisation making it highly scalable. This is confirmed by an experimental evaluation that demonstrates good speed up at a low overhead in workload and with little waiting time caused by synchronisation.