[CPMD-list] energy conservation

[Magali BENOIT]

Dear all,

I do agree with all the comments that have been done previously.
My experience in simulating high temperature (3500 K) silicate melts
(SiO2, SiO2-Na2O etc.) with thermostats is that you have to
"tune" your electronic thermostat parameters such that the total
energy drift becomes small. The most important parameters are:

- the kinetic energy of the electrons: if it is too small (compared
to the kinetic energy that the electrons would normally have, without
the thermostats) then you will have a large drift.
- the number of electronic degrees of freedom (the 4th number in
the Nose-Hoover parameters). You can try to increase the scaling.
- the suzuki integrator (5th number)
- the decomposition ratio (6th number)
- if you use "MASSIVE" thermostats for the ions, you will generally
undergo a larger drift.

I was never able to totally remove the drift but only to make it small
enough to live with it...You can help to cure the drift also by "refreshing"
the thermostats (do not restart with the old ones) from time to time.

Good luck,

Magali