[lammps-users] using fix/nvt with fix/uniaxial


I'm running a simulation to imitate a tensile test by using the following commands (My ensemble is NVT and I want a strain of 0.05 along z at the end of the simulation):

fix 1 all nvt 150.0 150.0 100.0
fix 2 all uniaxial 100 z 1.05

The stress values (that are calculated by lammps) fluctuate around a constant value and do not increase with increasing strain. I also calculated stress components that lammps do not include in its calculation (intra-molecular interactions such as bond, angle and dihedral potentials) and they also fluctuate around a constant value.

Is it wrong to use fix/nvt with fix/uniaxial in the way that I'm using?
Also, in order to calculate the total stress, is it ok to add stresses of individual atoms along a certain direction (i.e., sxx, syy, szz etc.) for a saved configuration?

Thank you for your help,


I see no problem with thermostatting (nvt) while using fix uniaxial.

If you dump per-atom stresses (sxx, etc), you should be able
to sum them or otherwise compute statistics on them however
you wish. The sum of sxx,syy,szz is related to the total pressure
of the system for any snapshot.