# [lammps-users] Pressure tensor fluctuation

Dear all,

To compute viscosity, I am running NEMD by deforming the box and using SLLOD

integration.

One thing I noticed is that for LJ fluid in triple point condition, without

shear, the average pressure tensor (Pxy Pxz Pxz) is always around 0.0001, while

fluctuation is around 0.1. When I introduced shear, the fluctuation level is still

around 0.1. I just wonder if this high fluctation is normal. If yes, how to estimate

the deviation when computing viscosity?

best wishes
Zhenlong

Pressures fluctuate, often by very large amplitudes. The
off-diagonal terms are worse b/c they are typically
small, even in shear for a fluid. You can use fix
viscosity (Muller-Plathe) method which doesn't
require monitoring the pressure.

Steve

Thanks, Steve.

I know velocity exchange is another way to calculate viscosity. But as for viscosity calculation by shear flow, I still want to make it clear how to do it within LAMMPS.

1. how to estimate the error bar for viscosity, considering the high magnitude fluctuation

of stress tensor?

1. To say, if the shear flow is introduced in x direction, and the velocity profile is
therefore generated in y direction, Pxy is required to calculate the viscosity. With available
options in LAMMPS, we can get Pxx, Pyy, Pzz, Pxy, Pxz, Pyz. I just wonder if the Pxy is exactly the one we need, or it is the average of Pxy and Pyx (they are different in this case).

Thanks

with best wishes

Zhenlong

Re: error estimate - that's a general Q, not really a LAMMPS Q, so I
won't attempt to answer.

Re: Pxy - LAMMPS does not compute Pyx. I think you want Pxy for
this viscosity calc. And I think you should always be able to setup
your simulation geometry so that Pxy (not Pyx) is what you want.

Steve