The ABINIT Users Mailing List ( CLOSED )

[Thierry Deutsch <thierry.deutsch@cea.fr>]

Dear Deyu,

      The wavelet implementation in ABINIT 5.2 is based on an article 
(L. Genovese et al.) which will be available at the end of this month in 
J. Chem. Phys. The Poisson solver is also available at 
http://pages.unibas.ch/comphys/comphys/SOFTWARE.

The Poisson solver is stable (based on Green function) and the 
computational time is comparable to the periodic Poisson solver. 
nscforder equal to 14 is enough for the accuracy.

As Pierre-Matthieu told you,  this is experimental : This is why the k 
point is not the Gamma point.
We need more tests especially for charged systems.

Best Regards,

       Thierry Deutsch
-

dylu@ucdavis.edu a écrit :
> Hello, dear abinit developers:
>        I have a few questions regarding the wavelets implementation in Abinit5.2 through "icoultrtmt" and "nscforder". After reading the release notes and the BigDFT webpage, though not much information available, I figured that it might be good for calculations of non-periodic systems, since the Hartree and PSP local terms  are evalulated through Poisson solver using open boundary conditions. 
>       
> 1. As for the example, v5_t07, why the K-point is shifted to 0.5 0.5 0.5 but not on the Gama point, which seems to me a natural choice for cluster calculations?
>
> 2. I expect the new method has a much better convergence regarding Etotal and 
> EHartree, if I understand it correct, since there is no image Coulomb 
> interaction. I modified the super cell size as the following in t07.in:
>
> ndtset 5
> acell: 3*5
> acell+ 3*2.5
>
> After a standard plane wave calculation using Abinit5.1.2, I got reasonable 
> numbers:
> [deyulu@zinfandel test512]$ grep Etotal *out
>     >>>>>>>>> Etotal= -3.24266940241069E-01
>     >>>>>>>>> Etotal= -4.26599328552894E-01
>     >>>>>>>>> Etotal= -4.42120273713403E-01
>     >>>>>>>>> Etotal= -4.44678592202608E-01
>     >>>>>>>>> Etotal= -4.45102876652833E-01
> [deyulu@zinfandel test512]$ grep "Hartree energy " *out
>     Hartree energy  =  1.32705836887717E-01
>     Hartree energy  =  1.38492352304226E-01
>     Hartree energy  =  1.56214176716576E-01
>     Hartree energy  =  1.74895165654710E-01
>     Hartree energy  =  1.90593035878911E-01
>
> Playing with the wavelets, the results seem to be very unstable. 
> Can someone provide more information about this implementation and
> tell me what is going wrong?
>
> nscforder  8     # Order of scaling used by kernel
> !nscforder2 14
> !nscforder3 16
> !nscforder4 20
> !nscforder5 24
>
> [deyulu@zinfandel tmp]$ grep Etotal *out
>     >>>>>>>>> Etotal= -3.24272603801180E-01
>     >>>>>>>>> Etotal= -1.00569442580796E+00
>     >>>>>>>>> Etotal= -1.69531799128392E+00
>     >>>>>>>>> Etotal= -2.37427341162027E+00
>     >>>>>>>>> Etotal= -2.97559451789859E+00
>
>
> [deyulu@zinfandel tmp]$ grep "Hartree energy " *out
>     Hartree energy  =  3.89261939710595E-01
>     Hartree energy  =  4.31111370942495E-01
>     Hartree energy  =  5.26095331415071E-01
>     Hartree energy  =  5.72879491830780E-01
>     Hartree energy  =  5.85224698919510E-01
>
> Thanks
> Deyu Lu
>   


--
-----------------------------------------------------------------------
Thierry Deutsch                                                        
Laboratoire de simulation atomistique (L_Sim)
DRFMC/SP2M                                      Tél:(33) 04 38 78 34 06
C.E.A.Grenoble                                  Fax:(33) 04 38 78 51 97
17, Avenue des Martyrs                           mailto:TDeutsch@cea.fr
38054 GRENOBLE CEDEX 9 FRANCE        http://www-drfmc.cea.fr/SP2M/L_Sim
-----------------------------------------------------------------------