The ABINIT Users Mailing List ( CLOSED )

[Marcel Mohr <marcel@physik.tu-berlin.de>]

Hi Xang

On Tue, 17 Apr 2007, Zhang, Waigen wrote:

> Maybe the difference is from the pseudopential.But what I need is to find a
> relationship between total energy and volume(pressure).Therefore,the exact 
> energy
> is important for me.
The absolute value of the total energy is of no importance. I even would 
go to say it has no real physical meaning.

However relative energy values have.
Total energy vs. volume looks a bit like bulk modulus, which corresponds 
a curvature, the absolute value is not important.
Said another way, the parabola x^2+c has the same curvature, independant 
of c.

Cheers
Marcel

> One thing I want to know most is that which result should I choose and should 
> I use
> GGA in other calculation for consitent,or LDA is OK?


> Best,
> Zhang,Waigen
>
>
> -----Original Message-----
> From: paul.boulanger@umontreal.ca [mailto:paul.boulanger@umontreal.ca]
> Sent: Tue 4/17/2007 2:50 AM
> To: forum@abinit.org
> Subject: Re: RE: [abinit-forum] LDA and GGA
>
> Hi,
>
> I don't really understand this big difference in total energy, but maybe the 
> key to comprehension comes from the two different pseudopotentials. In fact, 
> two different pseudopotentials should give two different total energies, we 
> can change the number of electrons, change the interaction between core and 
> valence, etc.
>
> After all, a pseudopotential is not suppose to conserve total energy but to 
> conserve the physics. All physical quantities are variation of the total 
> energy: binding energy, bulk modulus, phonon band structure, response 
> functions, etc. Hence comparing only total energies is a bad test to judge 
> the quality of a pseudopotential or for functionals. One should compare 
> physically mesurable quantities.
>
> Could you provide the two pseudopotential files (and source files)? Could you 
> calculate the binding energy or different propertie of your system with LDA 
> and GGA? These different quantities should give better agreement (but some 
> differences are expected, if I remember correctly GGA tend to overestimate 
> the binding).
>
> Good luck,
> Paul