I wonder whether anyone knows how the pressure tensor is calculated when the SLLOD equations for planar Couette flow are applied. Then it its important to use the peculiar velocities of the molecules, i.e. the velocity relative to the streaming velocity. However, I have the feeling that this is not the case because I get strange values for the viscosity which is equal to the ratio of the off-diagonal element of the pressure tensor and the shear rate.

I wonder whether anyone knows how the pressure tensor is calculated when the SLLOD equations for planar Couette flow are applied. Then it its important to use the peculiar velocities of the molecules, i.e. the velocity relative to the streaming velocity. However, I have the feeling that this is not the case because I get strange values for the viscosity which is equal to the ratio of the off-diagonal element of the pressure tensor and the shear rate.

well, fix nvt/sllod doesn't pay any attention to accumulating the
virial (on which the pressure tensor is based) and similarly compute
pressure doesn't pay any attention to how time integration is
performed. however, the temperature that fix nvt/sllod uses is
computed with compute temp/deform and you can tell compute pressure to
use this computed instead of the default to compute the kinetic
contribution to the virial and thus the pressure tensor (its id is the
fix-id with _temp appended). for example like this:

fix f1 all nvt/sllod temp 300.0 300.0 100.0
compute c1 all pressure f1_temp

that is how these kind of things are supposed to work in LAMMPS in
general. however, since i have not done any simulations of that kind
myself, i cannot confirm that this is sufficient for what you are
looking for. a more elaborate approach would be to run the simulation
and record the trajectory with velocities and box included, then
remove the flow bias with a self-written tool/script and analyze the
resulting unbiased trajectory with rerun.

Why would you need LAMMPS to? As far as I know you donâ€™t need information about the stress tensor in the SLLOD algorithm - it only works with your imposed rate of deformation. To ask a simpler question, if you prescribe both the deformation rate and the stress, are you simulating something or just making an animation?

By default, the pressure uses the full velocity, not the peculiar velocity. This can be changed using the thermo_modify temp command. There are other pitfalls too. You should closely study examples/VISCOSITY/in.nemd.2d