A scalable parallel Stokesian Dynamics method for the simulation of colloidal suspensions

We have developed a new method for the efficient numerical simulation of colloidal suspensions. This method is designed and especially well-suited for parallel code execution, but it can also be applied to single-core programs. It combines the Stokesian Dynamics method with a variant of the widely used Barnes–Hut algorithm in order to reduce computational costs. This combination and the inherent parallelization of the method make simulations of large numbers of particles within days possible. The level of accuracy can be determined by the user and is limited by the truncation of the used multipole expansion. Compared to the original Stokesian Dynamics method the complexity can be reduced from O(N2)O(N2) to linear complexity for dilute suspensions of strongly clustered particles, NN being the number of particles. In case of non-clustered particles in a dense suspension, the complexity depends on the particle configuration and is between O(N)O(N) and O(Pnp,max2), where PP is the number of used processes and np,max=⌈N/P⌉np,max=⌈N/P⌉, respectively.