Compositional analysis of boolean networks using local fixed-point iterations

Abstract

We present a compositional method which allows to over-approximate the set of attractors and under-approximate the set of basins of attraction of a Boolean network (BN). This merely consists in replacing a global fixed-point computation by a composition of local fixed-point computations. Once these approximations have been computed, it becomes much more tractable to generate the exact sets of attractors and basins of attraction. We illustrate the interest of our approach on several examples, among which is a BN modeling a railway interlocking system with 50 nodes and millions of attractors.