The ABINIT Users Mailing List ( CLOSED )

["Allan, Douglas C Dr" <AllanDC@corning.com>]

Thanks Xavier.  It is not a bad idea to wait "until the dust settles" before 
sending your comments. 

I would like to add one more thought for forum readers.  You can test which 
term(s) in the total energy are responsible for breaking translational 
invariance by running your favorite test case as a ground state calculation 
only, then re-running exactly the same structure but with all atomic 
coordinates shifted to the center of the "voxel" or small fft grid box 
associated with your fft grid.  Compare the different contributions to the 
total energy.  Many should agree to machine precision (especially if you set 
the tolerances to small values), while others (presumably those related to 
exchange-correlation energy) will not.  This separates out the calculation of 
energy terms from the calculation of derivatives, so is a useful test when 
looking for possible bugs.  The breaking of translational symmetry by the 
exchange-correlation integrals is not a bug.  It is simply an approximate 
integral that might need to be improved.  The nonlinearity of this expression 
in density rho(r) is the source of breaking translati
-Doug Allan

-----Original Message-----
From: Xavier Gonze [mailto:gonze@pcpm.ucl.ac.be] 
Sent: Saturday, December 14, 2002 12:14 PM
To: forum@abinit.org
Subject: Re: [abinit-forum] Phonons at Gamma


Dear Doug and Artem,

I have been following your exchange of mail. Sorry not to have taken the time 
to answer until now ... Actually, Doug has correctly summarized the expected 
behaviour of such computation :
- slight breaking of the translational invariance by the XC term
  only.
- should be better with non-zero intxc, but the latter has not
  been implemented for response-function
And Artem has observed the same behaviour as some other persons (but this was 
before the forum mailing list) :
- the deviation is larger for GGA than LDA
- the k points seem to have an effect
Also, the effect on Gamma phonon frequencies can be very large when Hydrogen 
is present, because (likely) of the small mass of H (frequencies of 100-150 
cm-1 have been observed sometimes !). (seen by Dennis Klug, Gian-Marco 
Rignanese, Xavier Rocquefelte, and maybe others).

That's the situation now :
- this might be normal, and improvable by implementing intxc 1
  in Response-Function
- this might be a small bug ?!

I simply do not know yet

In practice, people impose the acoustic sum rule in IFC,
which forces the phonon frequencies to be zero at Gamma,
and the other frequencies can be modified accordingly as well.

Good continuation,
Xavier

Allan, Douglas C Dr wrote:
 > Dear Artem,
 >
 > This is interesting.  I see the effect you describe in your output  > 
 > files.  Let me offer a quick comment and hope I can think more  > 
 > carefully about this later (after other obligations).  My comment:  > when 
 > thinking about the failure of translational invariance, I don't  > think 
 > it matters whether the actual calculation closely resembles a  > physical 
 > system (with sensible pseudopotentials, etc.).  I think what  > matters is 
 > the source of breaking translational invariance.  Without  > xc, I think I 
 > can prove that there are no terms that break  > translational invariance.  
 > I think I can even prove, without xc, that  > the k-point sampling can't 
 > break translational invariance.  I am  > working again from memory here, 
 > but this is what I recall from when I  > last saw this problem about 6 
 > years ago.  When you have a term that  > clearly breaks translational 
 > invariance, then the magnitude of the  > problem may well depend on things 
 > like k-point sampling.  It should  > be possible to show that mathematical
 > <http://slamdunk.geol.ucl.ac.uk/~artem> phone: +44 (0)20-7679-3424  > 
 > **************************  >  > -----Original Message----- From: Allan, 
 > Douglas C Dr  > [mailto:AllanDC@corning.com] Sent: Wednesday, December 11, 
 > 2002 10:17  > PM To: 'forum@abinit.org' Subject: RE: [abinit-forum] 
 > Phonons at  > Gamma  >  >  > One more comment: I see on more careful 
 > reading of the help files  > that the variable we introduced for this 
 > purpose, "intxc", was  > deprecated during further development work 
 > because it was too much  > trouble to implement this method within the 
 > response function work.  > The other developments were considered higher 
 > priority.  Thus, only  > the coarser integration is available (intxc=0) 
 > for response function  > work, while the more accurate (intxc=1) 
 > integration is available for  > ground state calculations.  >  > This does 
 > not explain why lda works better than gga for response  > functions, as 
 > both use the coarse grid, unless the nonlinearity of  > the gga (in 
 > density) makes more demands on
 > <http://slamdunk.geol.ucl.ac.uk/~artem> phone: +44 (0)20-7679-3424  > 
 > **************************  >  >  >  >