Apply shear stress by 'fix aveforce'

Your answer is really helpful. Thank you.
Actually, I try to mimic a creep deformation of a bicrystal, which is thought to be accommodate by grain boundary sliding.
As you know, I use a free surface condition normal to the grain boundary plane, so that it is difficult to use a barostat.

In experiment, grain boundary sliding happens because of the local shear stress applied on each grain.
I just want to use ‘fix aveforce’ command to achieve that constant shear stress creep, of course, with the help of thermostat.

I don’t understand why do you say ‘stresses are ill-defined’ if apply stress by stress=force/area.

What I was saying is biased from my own fluid and granular dynamics background. My thought was that since your deformation is basically discontinuous, your stresses wouldn’t be well defined. That’s just constitutive modeling problems though(which I thought you were asking about).

Sorry Yunjiang, but I am again confused, mostly in how you are dumping stress. I will also urge you to take some caution here if this is for publishing. Extracting the Cauchy stress tensor (which makes sense at the continuum level)is not always straightforward from MD (as I learned from my advisor). Stress in MD(Virial formulation) is a macroscopic quantity in that it is defined by a region/control volume, not a local quantity defined by a surface and traction vector as in the continuum case. You would find the area approach difficult to define in MD.