[lammps-users] please help regarding stress tensors

I saved sxy,sxz,syz stress tensors/atom at every time step for a 100 time steps run. The purpose is to compare results from lammps with another software (written by my group-mate, which is 3 times slower). In his code he outputs sxy,sxz,syz stress tensors of the system (and not per atom) at every time step. Since, LAMMPS gives it per atom, how do I go about to get the stress tensors, sxy,sxz,syz of the system?

I have tried to add the contributions of all the atoms to obtain the corresponding component for the total system. But, there is a big difference in both the results. This tells me that there is something wrong in the way I calculate the tensor components of the system (from tensor components/atom). Please provide me with some help on this. It would be great if somebody could tell me the correct way.

Steve told me in his previous mail that the total pressure is the total stress when aptly normalized. But, thats not what I want. I want the tensor componets (sxy,sxz,syz) of the system.

Thermo custom has options to print pxx, pyy, etc with the thermodynamic
output, which are the total system pressure components you
are looking for.

More specifically, if you do a simulation of a periodic box of a given
total volume and dump per-atom sxx syy szz to a file for each snapshot
... then if you sum the sxx, syy, szz values to a single quantity sum
(3N values),
you should get

pressure of system = - sum / volume / 3

where pressure is the thermo print-out. This is in accord with the
formula for per-atoms stress on this page of the LAMMPS doc:


Also if you print (via thermo custom) pxx,pyy,pzz I think you will also
see that total pressure = pxx+pyy+pzz / 3


To get total shear stresses, you do the summation and divide it by total
Volume. It is up to you to decide whether you want to use the current
volume or the initial volume (analogous to True stress and engineering
stress respectively).

Hope this helps,

Venkat Padmanabhan wrote: