Implementation and Parallel Optimization of the Lees-Edwards Boundary Condition in ESPResSo++

The Lees–Edwards boundary condition (LEbc) provides the possibility to simulate molecular or coarse grained systems under non-equilibrium conditions, namely by introducing a shear flow which has potential applications in high-speed fluids, thermoplastic and other non-equilibrium processes. This paper discusses a new LEbc implementation in the molecular dynamics (MD) software package ESPResSo++that focuses on the parallel efficiency of LEbc for scale-out simulations. Using the LEbc code, shear flow simulations were carried out for model systems such as Lennard-Jones fluids and Kremer-Grest polymer melts. Some important physical properties and phenomena, including the linear profiles of shear velocities, non-layered density distribution and shear thinning, have been successfully reproduced or captured. The results are also in good agreement with those from previous literature1 even with unphysical simulation conditions, which gives a solid validation to our implementation of LEbc. We considered 1 ar X iv :2 10 9. 11 08 3v 2 [ ph ys ic s. co m pph ] 2 4 N ov 2 02 1 in depth the parallelization of the LEbc code to efficiently scale in high performance computing (HPC) environments. A modified scheme for data communication has been introduced within the domain decomposition framework inside ESPResSo++. The presented benchmarks show a linear and good scaling for LEbc simulations for up to 1, 024 processor cores in supercomputer systems.