[CP2K:4575] BAND calculation with 16 frames

[Jörg Saßmannshausen]

Hi Carlos,

thanks for adding some more 'spice' here!

It helped me a lot to understand what I need to do better.

All the best from London!

Jörg

On Mittwoch 21 August 2013 Carlos Campana wrote:
> A bit more spice to Teo's main course,
> One of the reasons why your calculation will have better chances of
> succeeding when a TS candidate is provided
> is because the band should display minimum lateral motion at the TS
> (assuming not very stiff springs). The tension at the TS should help
> stabilizing the band motion. As to the interpolation, it is a bit of a
> guessing game that cannot be won without using your physical/chemical
> intuition. Gradients (forces) in Cartesian coordinates tend to vary
> abruptly from one point to another point in space. Thus, using collective
> variables (when known) can help a lot to achieve convergence.
> Carlos
> 
> On Wed, Aug 21, 2013 at 12:19 AM, Teodoro Laino 
<teodor... at gmail.com>wrote:
> > Jörg,
> > 
> > On Aug 21, 2013, at 12:30 AM, Jörg Saßmannshausen <
> > 
> > j.sassma... at ucl.ac.uk> wrote:
> > > Dear all,
> > > 
> > > I am in the process of setting up a BAND calculation with 16 frames.
> > > Each frame got 82 atoms and if I recall it correctly I need to have the
> > > cartesian coordinates in all frames in the same order. I assume that is
> > 
> > still
> > 
> > > correct? As this is a rather large molecule, doing it manually is a bit
> > 
> > correct - still, this should not be a limitation at all : during a
> > geometry optimisation or during an MD, the coordinates are never
> > swapped.. of course if you build your frames with different atoms
> > order.. well.. then you are asking for trouble… (or at least for more
> > additional work :) )
> > 
> > > cumbersome. I have tried to use GaussView for that but but it appears I
> > 
> > need a
> > 
> > > larger screen to get all 16 molecules in one window. :-(
> > > Is there a better program to do that task?
> > 
> > ?
> > 
> > > Also, I want to investigate a dissociation reaction, i.e.:
> > > 
> > > A-B ---> A      B
> > > 
> > > Due to the flexibility of the molecules A and B, the forward reaction
> > 
> > might be
> > 
> > > different from the reverse reaction. Is there a way to influence that
> > 
> > when I am
> > 
> > > doing the BAND?
> > 
> > A BAND is just a geometry optimisation is a wider space. What you put in,
> > is what you will get out optimised… Rarely there is only one path
> > connecting two points. This means that the success (in finding the
> > minimum energy path) is entirely due to quality of the interpolating
> > frames that you are feeding in.
> > 
> > Having said that, if you provide different frames for the opposite
> > reaction you will get probably a different path - unless the landscape is
> > relatively simple.
> > 
> > > I got the frames generated by pulling (pushing) A and B apart
> > > (together)
> > 
> > so I
> > 
> > > know the geometry is optimised here. I am doing steps of 0.3 A so that
> > 
> > should
> > 
> > > be ok as well (I hope). So, if I am using the optimised geometries for
> > 
> > the
> > 
> > > forward and reverse reaction I should be able to get what I am
> > > expecting
> > 
> > or is
> > 
> > > there something else which I need to bear in mind?
> > 
> > Just pulling apart the molecule and optimising them is quite risky:
> > In the rare event that the reaction is only driven by this distance you
> > will have automatically an optimised MEP. But, if the reaction is driven
> > by much more complex reaction coordinates, then you are severely biasing
> > the process.
> > The best approach is to provide A, B and something similar to your TS (or
> > any other crucial point, that you know is characterising your reaction).
> > Let all other frames being interpolated and let the BAND optimisation
> > relax the MEP.
> > 
> > Again, the final quality will be very much dependent on how good you'll
> > be able to provide your TS (or these crucial intermediate points).
> > If you do not have any idea about the TS, you can still try with a linear
> > interpolation (or something similar to your procedure) and see what
> > happens, analysing critically the results: most of the time it is easy to
> > understand if you are just throwing ropes between minima, or you really
> > got a MEP.
> > 
> > 
> > Teo
> > 
> > > I hope that makes sense to you.
> > > 
> > > All the best from 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
> > > 
> > > --
> > > 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/groups/opt_out.
> > 
> > --
> > 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/groups/opt_out.


-- 
*************************************************************
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