[Pw_forum] lattice constant from unit cell vs. from supercell

Chad Junkermeier junkermeier at bloch.nrl.navy.mil
Wed Mar 17 18:32:20 CET 2010


Yes, from what I wrote in my last post I can see the confusion.

In my last post I sent two representative files.  In actuality, I  
performed a series of calculations, each with a different alat value,  
on the unit cell.  The calculation that had the lowest "Final energy"  
was the was the one that I say has the equilibrium lattice constant  
for the unit cell.  I give here a listing of some of the results:

alat = 4.7
      Final energy   =     -71.3700391584 Ry
alat = 4.750
      Final energy   =     -71.3737078425 Ry
alat = 4.800                                                       
<---- Equilibrium value
      Final energy   =     -71.3753602156 Ry
alat = 4.850
      Final energy   =     -71.3751659760 Ry
alat = 4.900
      Final energy   =     -71.3732897789 Ry
alat = 4.950
      Final energy   =     -71.3698878408 Ry


I also performed a series of calculations, each with a different alat  
value, on the 2x2 supercell.  Of those calculations, the one with the  
lowest "Final energy" is the one that has the equilibrium lattice  
constant for the supercell case.  Here are some of the results for  
this case:

glat = 4.55
      Final energy   =    -285.4093082597 Ry
glat = 4.6
      Final energy   =    -285.4217551819 Ry
glat = 4.625
      Final energy   =    -285.4242025245 Ry
glat = 4.65                                                      <----  
Equilibrium value
      Final energy   =    -285.4242932895 Ry
glat = 4.675
      Final energy   =    -285.4221256935 Ry
glat = 4.700
      Final energy   =    -285.4177075378 Ry
glat = 4.725
      Final energy   =    -285.4115070010 Ry



Since the supercell case is essentially the same system as the unit  
cell case, the equilibrium value for alat should be the same in both  
cases.  I am finding that they are not.  Further, the final positions  
of the unit cell case has atom positions near where we think they  
should be, but in the 2x2 unit cell the final atom positions are not.



Chad Junkermeier, Ph.D.
NRC Post-Doctoral Fellow
Naval Research Laboratory
4555 Overlook Ave, S.W.
Washington DC 20375
Office: (202) 404-4635
Fax: (202) 767-1165







On Mar 17, 2010, at 12:52 PM, Paolo Giannozzi wrote:

> Chad Junkermeier wrote:
>
>> If I do this using the unit cell, everything seems to
>> work correctly, and the equilibrium lattice constant for the system
>> seems to be correct.  If, instead, I use a 2x2 supercell I find that
>> the equilibrium lattice constant is not the same (or even close) to  
>> the
>> unit cell result.
>
> your data are for fixed-cell structural optimization: what do you mean
> by "the equilibrium lattice constant is not the same"? You choose the
> lattice constant
>
> P.
> -- 
> Paolo Giannozzi, Democritos and University of Udine, Italy
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum




More information about the users mailing list