[CP2K:6967] Re: Re-optimising transition state using dimer method

[luca]

Dear Tobias,
you are refining a TS state obtained from NEB,  thus, you have a "good"
starting point.
Nevertheless, the dimer-algorithm assigns the directions of the dimer in
random way, thus, in this case,  the random assignement affects
negatively the convergence of the refininement, preventing the
convergence.

Provide the right dimer-directions in agreement with the imaginary mode 
is the right choice, however, I would suggest to use a proper set-up of
the input parameters to avoid extra work.

-1- See J. Org. Chem., 2013, 78 (5), pp 2118–2127
DOI: 10.1021/jo3023439
I this work I used this set-up:

    &TRANSITION_STATE
      METHOD DIMER
      &DIMER
        DR [angstrom] 0.01
        ANGLE_TOLERANCE 1.5
        INTERPOLATE_GRADIENT  T
        &ROT_OPT
          OPTIMIZER CG
          MAX_ITER  12           #INCREASE!
          &CG
            MAX_STEEP_STEPS 0
            &LINE_SEARCH
              TYPE 2PNT
            &END
          &END
        &END
      &END
    &END

Anyway, the parameters are system dependent, so you have to try
different set-up.

-2- In your case, to obtain the right directions of the dimers, you can
increase the MAX_ITER keyword, at least for the first few search steps.
Because you are near the TS, increasing the MAX_ITER, the algorithm
should find the dimer directions for the local imaginary mode, which
should be the imaginary mode obtained by the vibrational analysis.


Luca  


> 
> thanks for the quick reply. I was thinking along similar lines, since
> I was wondering how the optimiser would know which direction to follow
> (for other quantum chemical codes you would have to provide force
> constants in one way or other). Now, having not used this keyword in
> CP2K before, could you be a bit more specific about the format?
> According to the manual, one would explicitly specify the dimer vector
> in this input section, i.e. ideally I would use the imaginary mode for
> this.I simply copy this from the _frequenxies.mol output file (the one
> I have specified to read with molden). Hence the section would look
> like this:
> 
> &MOTION
>   &GEO_OPT
>     TYPE TRANSITION_STATE
>      &TRANSITION_STATE
>        METHOD DIMER
>         &DIMER
>         &ROT_OPT
>           OPTIMIZER CG
>         &END ROT_OPT
>         &DIMER_VECTOR
>             -0.000030      -0.000001      -0.000005
>              0.000019       0.000008       0.000023
>             -0.000008      -0.000017      -0.000007
>          <remaining lines>
>         &END DIMER_VECTOR
>        &END DIMER
>      &END TRANSITION_STATE
>     OPTIMIZER CG
>     MAX_FORCE 8.0E-4
>     MAX_ITER 500
>   &END GEO_OPT
>   &PRINT
>      &TRAJECTORY
>       &EACH
>          GEO_OPT 1
>       &END EACH
>     &END TRAJECTORY
>   &END
> &END MOTION
> 
>   
> Tobias
> 
> -- 
> You received this message because you are subscribed to the Google
> Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to cp2k+uns... at googlegroups.com.
> To post to this group, send email to cp... at googlegroups.com.
> Visit this group at http://groups.google.com/group/cp2k.
> For more options, visit https://groups.google.com/d/optout.