Dear Lammps users,

I’ve run simulations of simple shear flow with LAMMPS ( basically the fix deform command + DPD pair force + fix NVE which guarantees linear velocity profile as also confirmed by other papers ) . the problem I encountered is the very large noise in the non-diagonal stress tensor element, i.e., Txy. the average of stress values at steady state is logical (compare to other reports of the same system), how ever the error bars are so large; the same degree as the average values which makes accuracy of these results questionable. my colleague who is also running a NEMD simulation ( fix deform + SLLOD equation + lj potential) observed the same large noise in the stress values. therefore, it seems regardless of the approach and potential, the noise is very large which is not reported by other research groups for the same systems(actually their error bars are very small). I was wondering if anyone had the same problem with LAMMPS shear simulations. any help is much appreciated.


Just a thought! Could it be your system size? Have you check if the error bars you are getting increase if you run your simulations with twice the size?