This is very helpful advice and thank you.
I’m not really sure why they used such a low temperature (0.5 K). I would have thought that mechanical tests at room temperature on the CNT bundles would be most similar to the real situation?. But there is probably a good reason for their decision which I will ask when I talk to them. Actually what is the reason for using a thermostat during a mechanical test (tension, torsion etc.) anyway?
The reason for thermostating the atoms (on which the twisting is not applied) is to keep those atoms at certain temperature during the twise. If you don’t apply the thermostats (nvt, npt), which means you are not controlling the temperature, the only other way () to update the velocities and coordinates is to use “fix nve”, in which u assume your system is in microcanonical ensemble (constant total energy). In later one, your CNT temperature will not be constant and I am suspecting you might see significant fluctuations and possibly gradual decrease in CNT temperature over time (your potential energy will increase with increase in strain due to twisting, which in turn will decrease kinetic energy). You could try both approaches and see which one seems more physical.