to reduce the noise in output from compute msd I can either run multiple times and average or run one long simulation and compute msd in intervals and then average over multiple msd outputs.
say if I want to see the msd of atoms during 100 time steps, I can run a long simulation of 1000 steps and then compute msd at ts=0, 10, 20,…900. then I will have 101 msd outputs of 100 timesteps long. averaging over them would give me smoother average.
from doc page for msd I quote: “The displacement of an atom is from its original position at the time the compute command was issued.” so if I issue multiple compute msd commands in one run they will all start from t=0 and initial position thus the same output.
does anyone knows a work around to get multiple msd output from one simulation?