Dynamic Bounds and Transition Merging for Local First Search

Local First Search (LFS) was recently introduced as a new partial order reduction approach. It is based on the observation that local properties can be found searching on paths with a low degree (LFS-number) of concurrent progress. It has shown its potential to find sequences to local states fast. In this paper, we explore several improvements of LFS: On the one hand, we propose a replacement of the static bound on LFS-numbers by a dynamic criterion to detect exhaustion of reachable local properties faster. On the other hand, we explore the possibilities of combining LFS with other partial order reduction methods. It turns out that the combination with transition merging is possible and rewarding, while the sleep-set method is incompatible with LFS.