Discontinuous spikes in MSD timeseries

[emilyviolet]

Plotting the mean-square displacement over time occasionally has large spikes, which are unphysical.

The timeseries (for either x-, y- or z-components) starts out continuously and monotonically increasing before leveling out at around t=0.3 (for default parameters). After this point, it sort of sits in a steady-state of around 0.02, but occasionally spikes up to around 0.14 for one or two timesteps before dropping down to the steady-state value.

[emilyviolet]

This is likely due to the effects of periodic boundary conditions. Currently, the mean-squared displacement is calculated using the "wrapped" coordinates (i.e. with PBCs applied). The MSD plots are much smoother if we use unwrapped coordinates (i.e. adding or subtracting a box-length if the particle moves outside the box in a time-step), but the differences between phases of matter are less obvious (assuming I'm implementing it correctly).