The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

mixing relaxation and MD will not work: how do you decide which steps are 
relaxation and which MD? Each time you relax, you lose energy and 
temperature... You could do an MD at fixed Nose-Hoover or Langevin 
thermostat temperature, but then you have to look at all the time steps 
and do statistical analysis, see if your system looks like one or the 
other phase. Also consider applying pressure (Gilles Zerah was 
implementing constant P MD - what is the status on that?)

What you should be doing, in your case, is the calculation of free 
energies versus temperature. For each of the competing phases you do a 
standard GS run (eventually at different lattice parameters for the 
thermal expansion - this is where it gets painful), calculate the phonon 
spectrum, and anaddb allows you to plot the free energies vs T. The F(T) 
or G(T) curves should intersect and one phase becomes more stable at T_1 
then another at T_2 etc... No MD needed.

One example using abinit: "Ab initio thermodynamics of MgSiO3 perovskite 
at high pressures and temperatures." in J. Chem. Phys. 122,pp 124501-6 
(2005)

And one not using abinit: Correa et al. PNAS 103, 1204 (2006)

Matthieu

--
================================================================
Dr. Matthieu Verstraete                 mailto:mjv500@york.ac.uk
Dept. of Physics, University of York,     tel: +44 1904 43 22 08
Heslington, YO10 5DD York, United Kingdom fax: +44 1904 43 22 14