Dear Nguyen Monh Son,

there are two ways to control the ionic temperature in cp.x, one with a simple rescaling, which likely does not lead to a 'good' canonical ensemble sampling, and the Nose-Hoover, much better from a physical point of view, but also more difficult to use. Use the first approach allows to easily tune the temperature fluctations (use the 'tolp' parameter). For the Nose-Hoover thermostat, you can use the 'fnosesep' parameter. The 'best' value depends on the system you investigate. It should be in the range of the vibrational frequencies of this system for an efficient temperature control.

HTH,

Laurent

Dear Laurent Pizzagalli, thanks for your advice.