I made the same assumption when developing the MOLC model, i.e. to use the pressure computed with the translational temperature instead of the default pressure for the NPT/NPH time integrators. I would say that the rotational component of the temperature should not contribute to the pressure, but I, too, wondered why it was implemented in this way in the ASPHERE package.
These are the settings I currently use:
atom_style hybrid molecular ellipsoid
units real
# ...
compute temp_trasl all temp
compute temp_rototrasl all temp/asphere dof all
compute press_trasl all pressure temp_trasl
# ...
thermo_modify temp temp_rototrasl press press_trasl flush yes
# ...
timestep 10.
fix 1 all npt/asphere temp 300. 300. 1000. iso 1. 1. 10000.
fix_modify 1 press press_trasl
fix_modify 1 temp temp_rototrasl
Here, I use the full virial and the translational temperature to compute the pressure. I haven’t tested your settings with the ke keyword, but I think this is incorrect as it will not include the force field contributions to the virial.
I hope this helps.
Otello