Particle filter efficiency under limited communication

Summary Sequential Monte Carlo methods are typically not straightforward to implement on parallel architectures. This is because standard resampling schemes involve communication between all particles. The $$\alpha$$-sequential method was proposed recently as a potential solution this that limits limited controlled through sequence of stochastic matrices known $$\alpha$$ matrices. We study the influence structure convergence and stability properties resulting algorithms. In particular, we quantitatively show mixing play an important role in algorithm. Moreover, prove one can ensure good by using randomized structures where each particle only communicates with few neighbouring algorithms converge at usual rate. leads efficient versions distributed sequential Carlo.