[CPMD-list] Temperature simulations

[Ari.P.Seitsonen at iki.fi]

Dear Veronique,

> thanks for your suggestions.  First of all I gave the atoms a random 
> displacement of 0.05 by the option randomize in the inputfile and then 
> did again a geometry optimization.  The programs ends normally, but the 
> atoms do not return to their original position, the energy however is 
> converged.  I must conclude then that the system is indeed unstable. 

So indeed the 4+4 atom cell is too small for the Gamma point only. The
next bigger (simple cubic) cell is 32+32 atoms. There are also
intermediate cells but they have either BCC or FCC basis vectors.

>  The input file is given below of this simulations.  By the way I did 
> not optimize the cell parameters, is this necessary?  

The lattice parameters are not important - as long as e.g. the pseudos
are good and the basis set more or less converged. Usually one should
be within about 2 % of the experimental lattice constant, and in such
a small range there are no competing structures; even if there would
be one completely different lattice strcture there are too few atoms
for a phase transition to occur (and the basis vectors are anyway
fixed).

> Is it an option to use the nose keyword for the electrons to keep them 
> close to the BO surface?  

I think that KCl should be an insulator with a large electronic gap,
right? Then the electrons should not deviate far from the
Born-Oppenheimer surface unless the time step is too large or the
electronic mass is inapproriate.

> Then the question remains how to choose this electronic mass?  

Well, if one already has a value for the maximal time step you can try
a couple of values for the \mu and take the "average" one which
conserves the total energy (KS+ionic kinetic+electron kinetic). If you
are not sure about the time step you can take a smaller one, run the
tests for the electronic mass and then increase them so that dt^2/\mu
is a constant, again as far as the total energy is well preserved.
(How to choose the time step: about 1/6th ... 1/10th of the fastest
ionic period unless the fastest electronic oscillations = E_cut limit
the value)

    Greetings,

       apsi

-- 
-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
 Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
 Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35      
 Anschrift: Physikalisch Chemisches Institut (PCI), Universität Zürich (UniZh)
 Indirizzo: Winterthurerstraße 190, CH-8057 Zürich
 Address:   Schweiz / Svizzera / Suisse / Svizra / Switzerland