[CP2K:4634] restarting NEB calculations.

Jörg Saßmannshausen j.sassma... at ucl.ac.uk
Wed Oct 2 19:44:51 UTC 2013


Hi Teo,

thanks for the quick feedback!

On Donnerstag 26 September 2013 Teodoro Laino wrote:
> On Sep 26, 2013, at 1:31 PM, Jörg Saßmannshausen <j.sassma... at ucl.ac.uk> wrote:
> > in the same directory where I done the calculation but alas! it appears
> > it started again from scratch. The file sm-scan3-1_275.restart is the
> > last restart file I got so I assumed it contains all the required
> > information.
> 
> the energies that you will get from your NEB  must be identical (up to the
> SCF convergence) to the one of the frame corresponding to the restart
> file. Is this the case?

No, it is not. That is the last frame which was computed in the first run:

 *******************************************************************************
 BAND TYPE                     =                                              SM
 BAND TYPE OPTIMIZATION        =                                            DIIS
 STEP NUMBER                   =                                             275
 NUMBER OF NEB REPLICA         =                                              16
 DISTANCES REP =        2.530978        2.530977        2.530977        2.530976
                        2.530975        2.530975        2.530977        2.530978
                        2.530978        2.530977        2.530977        2.530977
                        2.530977        2.530978        2.530978
 ENERGIES [au] =     -648.478940     -648.467777     -648.450539     -648.436808
                     -648.424076     -648.411575     -648.401455     -648.395091
                     -648.394535     -648.397824     -648.402001     -648.407490
                     -648.415564     -648.423551     -648.425201     -648.422192
 BAND TOTAL ENERGY [au]        =                           -10374.75462048212466
 *******************************************************************************
 *******************************************************************************
 RMS DISPLACEMENT =   0.00053     [ 0.00010]                               ( NO)
 MAX DISPLACEMENT =   0.00885     [ 0.00020]                               ( NO)
 RMS FORCE        =   0.00274     [ 0.00050]                               ( NO)
 MAX FORCE        =   0.02016     [ 0.00100]                               ( NO)
 *******************************************************************************

and that is the first frame after the restart:

 *******************************************************************************
 BAND TYPE                     =                                              SM
 BAND TYPE OPTIMIZATION        =                                              SD
 STEP NUMBER                   =                                               1
 NUMBER OF NEB REPLICA         =                                              16
 DISTANCES REP =        4.611085        4.611083        4.611081        4.611071
                        4.611053        4.611042        4.611057        4.608585
                        4.363173        4.583945        4.610633        4.610733
                        4.611017        4.611071        4.611085
 ENERGIES [au] =     -648.478940     -648.388234     -648.213414     -648.033912
                     -647.902317     -647.837233     -647.892035     -648.067775
                     -648.274512     -648.336806     -648.031108     -647.791068
                     -647.757054     -647.962000     -648.273950     -648.422192
 BAND TOTAL ENERGY [au]        =                           -10369.66255104071388
 *******************************************************************************
 *******************************************************************************
 RMS DISPLACEMENT =   0.05464     [ 0.00010]                               ( NO)
 MAX DISPLACEMENT =   0.26217     [ 0.00020]                               ( NO)
 RMS FORCE        =   0.02618     [ 0.00050]                               ( NO)
 MAX FORCE        =   0.18566     [ 0.00100]                               ( NO)
 *******************************************************************************

So there is a large difference in the engergy and it does no get any better after 6 more frames:

 *******************************************************************************
 BAND TYPE                     =                                              SM
 BAND TYPE OPTIMIZATION        =                                              SD
 STEP NUMBER                   =                                               6
 NUMBER OF NEB REPLICA         =                                              16
 DISTANCES REP =        4.167181        4.167174        4.167158        4.167129
                        4.167112        4.167054        4.166423        4.159775
                        4.127302        4.118230        4.162667        4.166636
                        4.167056        4.167160        4.167181
 ENERGIES [au] =     -648.478940     -648.433865     -648.352353     -648.268776
                     -648.195851     -648.133911     -648.104171     -648.144445
                     -648.236878     -648.266347     -648.180945     -648.131997
                     -648.148447     -648.242456     -648.366991     -648.422192
 BAND TOTAL ENERGY [au]        =                           -10372.10856378791868
 *******************************************************************************
 *******************************************************************************
 RMS DISPLACEMENT =   0.02265     [ 0.00010]                               ( NO)
 MAX DISPLACEMENT =   0.12833     [ 0.00020]                               ( NO)
 RMS FORCE        =   0.01330     [ 0.00050]                               ( NO)
 MAX FORCE        =   0.10403     [ 0.00100]                               ( NO)
 *******************************************************************************

So I think I got a problem here.
 
> If not, check the keywords that control aligment..

I did not set the ALIGN_FRAMES keyword and the default is .T. here.
I am doing a SM band type here and in that section I have set 
ROTATE_FRAMES F
Was that correct or is that where the problem is?

> Also, not all information are stored in the restart file. For instance, the
> DIIS vectors are not. This means that the first steps may be offline
> regarding the gradients, but the energy of the first point MUST be
> identical to the one of the frame when restart was written.

That is what I would have expected. However, unless I am blind and do not see something, 
as the energy is different between step 275 and the first step in the restart I got a problem here.

> 
> > If I cannot restart it that way, and as I am close to the convergence
> > criteria, can I simply use the last frames and use that as a starting
> > point? I
> 
> yes

Are there any problems associated with that method?

All the best from a mild London!

Jörg

-- 
*************************************************************
Jörg Saßmannshausen
University College London
Department of Chemistry
Gordon Street
London
WC1H 0AJ 

email: j.sassma... at ucl.ac.uk
web: http://sassy.formativ.net

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html




More information about the CP2K-user mailing list