xc funtional - TFW

joanyi Chen jny... at gmail.com
Sun Dec 11 03:42:17 CET 2011


Hi Juerg,

Thanks for the very important notice.

So my previous input file is not a straightforward way to do orbital-
free calculations. As you mentioned before, the OFGPW method is
probably not well-tested, and yet I'm not able to improve it.

Would you mind showing an example/input file on how to do orbital-free
calculations with TF or TFW model? It will help me to get a clearer
picture of OF. And I really appreciate your warm support.

All the best wishes from Taipei at 19 ºC :)

regards
Joan

On 12月3日, 上午1時47分, hut... at pci.uzh.ch wrote:
> Hi
>
> TF is the Thomas-Fermi kinetic energy model.TFWis TF + von Weizsaecker correction
>
> BUT, if you do  METHOD GPW
> then you always have the Kohn-Sham kinetic energy, adding to that
> TF orTFWdoes not make sense (double counting of kinetic energy).
>
> regards
>
> Juerg
>
> --------------------------------------------------------------
> Juerg Hutter                         Phone : ++41 44 635 4491
> Physical Chemistry Institute   FAX   : ++41 44 635 6838
> University of Zurich               E-mail:  hut... at pci.uzh.ch
> Winterthurerstrasse 190
> CH-8057 Zurich, Switzerland
> ---------------------------------------------------------------
>
> -----cp... at googlegroups.com wrote: -----
>
> To: cp... at googlegroups.com
> From: Joanyi Chen
> Sent by: cp... at googlegroups.com
> Date: 12/02/2011 06:33PM
> Subject: Re: [CP2K:3652] Re: xc funtional -TFW
>
> Hi Juerg,
>
> Thanks for your suggestions.
>
> I'll shorten my question into one simple setence,  so will not waste your time.
> I want to make sure that the forms of TF andTFWfuntionals are implemented from the Thomas-Fermi model and the Weizsacker correction to the Thomas-Fermi kinetic energy?
>
> Thank you!
>
> regards,
> Joan
> -----------------------------------------------------------------------
> In case anyone is interested in the reason for such a stupid quetion. I'll explain as follows.
>  I did some calculations by using TF and BLYP
> functionals. When comparing the results, the energies of using  TF functional+ BLYP are higher than BLYP.
> It's strange for me since the TF model are hypothesized at 0 K, so the kinetic energy should be underestimated. However the outcomes are exactly opposite. Or maybe it has something to do with my inputfile...?
>
> Thanks for any suggestion and hint.
>
> There are some data from my output file.
>           TF+BLYP      BLYP
> H        -0.26489     -0.49740
> He      -1.35442     -2.83187
> CO2   -22.50647   -37.96400
> CH4    -3.97605    -8.04579
>   Please allow me to post my input file again.
>
> &GLOBAL
>    PROJECT H
>    PRINT_LEVEL MEDIUM
>    RUN_TYPE ENERGY
>  &END GLOBAL
>  &FORCE_EVAL
>    METHOD QS
>    &DFT
>      BASIS_SET_FILE_NAME ~/cp2k/tests/QS/ALL_BASIS_SETS
>       POTENTIAL_FILE_NAME ~/cp2k/tests/QS/POTENTIAL
>      CHARGE 0
>      MULTIPLICITY 2
>      LSD
>      &QS
>        METHOD GPW
>      &END QS
>      &SCF
>        EPS_DIIS 1.0E-2
>         EPS_SCF 1.0E-8
>        MAX_SCF 100
>        SCF_GUESS ATOMIC
>                   &OT T
>           ALGORITHM IRAC
>          &END OT
>      &END SCF
>        &XC
>        &XC_FUNCTIONAL BLYP
>          &TF T
>         &END TF
>        &END XC_FUNCTIONAL
>      &END XC
>    &END DFT
>    &SUBSYS
>      &CELL
>        ABC 10.0 10.0 10.0
>      &END CELL
>      &COORD
>      H     0.00000000  0.00000000  0.00000000
>          &END COORD
>      &KIND H
>        BASIS_SET DZVP-GTH-BLYP
>        POTENTIAL GTH-BLYP-q1
>      &END KIND
>    &END SUBSYS
>  &END FORCE_EVAL
>
> 2011/11/21  <hut... at pci.uzh.ch>
>  Hi
>
>  the OFGPW code in CP2K is a leftover of a currently abandoned
>  project. It should be correct, although it is not well tested.
>  There are two problems with the implementation:
>  - you need local pseudopotentials
>  - convergence is very poor
>  Unless you are willing to invest a lot of time into the code
>  I would not recommend to start a project based on this implementation.
>
>  regards
>
>  Juerg
>
>  --------------------------------------------------------------
>  Juerg Hutter                         Phone : ++41 44 635 4491
>  Physical Chemistry Institute   FAX   : ++41 44 635 6838
>  University of Zurich               E-mail:  hut... at pci.uzh.ch
>
> Winterthurerstrasse 190
>  CH-8057 Zurich, Switzerland
>  ---------------------------------------------------------------
>
>  -----cp... at googlegroups.com wrote: -----
>
>  To: cp2k <cp... at googlegroups.com>
>  From: joanyi Chen
>  Sent by: cp... at googlegroups.com
>  Date: 11/20/2011 02:31PM
>  Subject: [CP2K:3619] Re: xc funtional -TFW
>
>  Hi Juerg,
>
>  Thank you for your reply.
>  Actually I want to use orbital-free method to do calculations,
>  and TF andTFWare selected kinetic energy funtionals.
>  If I add OFGPW option for orbtial free under QS, does it make sense
>  for this calculation? Are there other parameters needed
>  to be set?(Sorry to bug you like this, it's just I only find the
>  OFGPW option from the keyword "orbital-free" in the manual.)
>
>  Then, I just added the OFGPW option to see if it's ok, the notorious
>   forrtl: severe (174): SIGSEGV, segmentation fault occurred.
>  However, when I change OFGPW to GPW and GAPW, they work.
>  Wandering whether it's associated with setting of parameters?
>
>  Thanks for any tip or hint.
>  regards,
>  Joan
>  On 11月18日, 下午5時19分, hut... at pci.uzh.ch wrote:
>  > Hi
>  >
>  > although I'm not perfectly sure, this might be a bug in the code.
>  > On the other hand I doubt that you want to do such a calculation.
>  > The TF andTFWfunctionals are kinetic energy functionals and
>  > do not make sense in the setup you have. Try some of the
>  > standard XC functionals.
>  >
>  > regards
>  >
>  > Juerg
>  >
>  > --------------------------------------------------------------
>  > Juerg Hutter                         Phone : ++41 44 635 4491
>  > Physical Chemistry Institute   FAX   : ++41 44 635 6838
>  > University of Zurich               E-mail:  hut... at pci.uzh.ch
>  > Winterthurerstrasse 190
>  > CH-8057 Zurich, Switzerland
>  > ---------------------------------------------------------------
>  > > -----cp... at googlegroups.com wrote: -----
>
>  >
>  > To: cp2k <cp... at googlegroups.com>
>  > From: joanyi Chen
>  > Sent by: cp... at googlegroups.com
>  > Date: 11/16/2011 08:05AM
>  > Subject: [CP2K:3602] xc funtional -TFW
>  >
>  > Dear cp2k users,
>  > I'm trying to calculate single point energy on cp2k. According to
>  > [CP2K:2693], I encoutered almost the same issue. I was asked to add
>  > LSD under DFT, because of the odd number of electrons. Then, I added
>  > LSD, ran the script below, except the xc funtional part. I chose TF as
>  > the funtional. It works. After that, I change the xc funtional toTFW,
>  > the error comes along...
>  >
>  > Here is my input file
>  > &GLOBAL
>  >   PROJECT CCH
>  >   PRINT_LEVEL MEDIUM
>  >   RUN_TYPE ENERGY
>  > &END GLOBAL
>  > &FORCE_EVAL
>  >   METHOD QS
>  >   &DFT
>  >     BASIS_SET_FILE_NAME ~/cp2k/tests/QS/ALL_BASIS_SETS
>  >     POTENTIAL_FILE_NAME ~/cp2k/tests/QS/POTENTIAL
>  >     CHARGE 0
>  >     MULTIPLICITY 2
>  >     LSD
>  >     &QS
>  >       EPS_DEFAULT 1.0E-8
>  >     &END QS
>  >     &SCF
>  >       EPS_DIIS 1.0E-14
>  >       EPS_SCF 1.0E-8
>  >       MAX_SCF 100
>  >       SCF_GUESS ATOMIC
>  >         &OUTER_SCF T
>  >         &END OUTER_SCF
>  >         &OT T
>  >          ALGORITHM IRAC
>  >         &END OT
>  >     &END SCF
>  >       &XC
>  >       &XC_FUNCTIONAL
>  >        &TFW T
>  >        &ENDTFW
>  >       &END XC_FUNCTIONAL
>  >     &END XC
>  >   &END DFT
>  >   &SUBSYS
>  >     &CELL
>  >       ABC 8.0 4.0 4.0
>  >     &END CELL
>  >     &COORD
>  >     C     0.00000000     0.00000000    -0.47096960
>  >     C     0.00000000     0.00000000     0.73273419
>  >     H     0.00000000     0.00000000    -1.53442998
>  >     &END COORD
>  >     &KIND H
>  >       BASIS_SET DZVP-GTH-BLYP
>  >       POTENTIAL GTH-BLYP-q1
>  >     &END KIND
>  >     &KIND C
>  >       BASIS_SET DZVP-GTH-BLYP
>  >       POTENTIAL GTH-BLYP-q4
>  >     &END KIND
>  >   &END SUBSYS
>  > &END FORCE_EVAL
>  > and the error.
>  > ************************************************************************
>  >  *** 14:21:29 ERRORL2 in xc_rho_set_types:xc_rho_set_get processor 0
>  > ***
>  >  *** err=-300 condition FAILED at line 434
>  > ***
>  >
>  > ************************************************************************
>  >
>  >  ===== Routine Calling Stack =====
>  >
>  >            11 tfw_lsd_eval
>  >            10 xc_functional_eval
>  >             9 xc_rho_set_and_dset_create
>  >             8 xc_vxc_pw_create
>  >             7 qs_vxc_create
>  >             6 qs_ks_build_kohn_sham_matrix
>  >             5 qs_ks_update_qs_env
>  >             4 init_scf_loop
>  >             3 scf_env_do_scf
>  >             2 qs_energies_scf
>  >             1 CP2K
>  >  CP2K| condition FAILED at line 434
>  >  CP2K| Abnormal program termination, stopped by process number 0
>  >
>  > Any hint or suggestion will be helpful.
>  > Thank you!
>  >
>  > Joan
>  >
>  > --
>  > You received this message because you are subscribed to the Google Groups "cp2k" group.
>  > To post to this group, send email to cp... at googlegroups.com.
>  > To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
>  > For more options, visit this group athttp://groups.google.com/group/cp2k?hl=en.
>
>  --
>  You received this message because you are subscribed to the Google Groups "cp2k" group.
>  To post to this group, send email to cp... at googlegroups.com.
>  To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
>  For more options, visit this group athttp://groups.google.com/group/cp2k?hl=en.
>
>  --
>  You received this message because you are subscribed to the Google Groups "cp2k" group.
>  To post to this group, send email to cp... at googlegroups.com.
>  To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
>  For more options, visit this group athttp://groups.google.com/group/cp2k?hl=en.
>
>   --
>  You received this message because you are subscribed to the Google Groups "cp2k" group.
>  To post to this group, send email to cp... at googlegroups.com.
>  To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
>   For more options, visit this group athttp://groups.google.com/group/cp2k?hl=en.



More information about the CP2K-user mailing list