[CP2K-user] [CP2K:18356] Re: Guided AIMD reaction, CONSTRAINT, RESTRAINT, METADYNAMICS

[Marcella Iannuzzi]

Dear Dobromir

The TARGET is the initial value that is imposed to the variable, as it is 
defined. 
The TARGET_GROWTH is the rate of change of that target value, which in the 
example you posted is ridiculously high:

TARGET_GROWTH {Real}
*Specifies the growth speed of the target value of the constrained 
collective variable.*

The constraint with the value determined by the growth rate will be 
imposed, no matter which are the other forces in place.
At least until the calculation does not converge anymore and everything 
explodes.

Nothing of this is a sampling in the NVT ensemble, anyway.

If the goal is to sample the final state after the reaction has occurred, 
just start a simulation from the configuration of the reacted molecule.
If the goal is to find the reaction mechanism, there is no way out of 
studying the free energy surface or, in simpler cases, the potential energy 
surface. 
An efficient way to find  the potential minimal energy pathway in  cases 
like yours is in general the nudged elastic band approach.

Regards
Marcella


On Thursday, January 19, 2023 at 9:42:15 PM UTC+1 Dobromir A Kalchevski 
wrote:

> Hello everyone!
>
> I want to make a molecule occasionally approach a surface and react with 
> it in a NVT ensemble production run. I don't want it fixed in predetermined 
> coordinates and I don't want it to fly away or to wait for too long of a 
> trajectory.
> For now I don't want to study the reaction's thermodynamics. 
> Experimentally the reaction occurs.
>
> I tried the following approaches without success:
>
> I defined a COLVAR for the distance between one surface atom (Si) and the 
> center atom (C) of the gas phase reagent molecule (methane)
>
>         &COLVAR
>             &DISTANCE
>                 AXIS XYZ                !DEF XYZ, ANY COMBINATION
>                 ATOMS 50 81
>             &END DISTANCE
>         &END COLVAR
>
> and then I defined a CONSTRAINT
>
>     &CONSTRAINT
>         &COLLECTIVE
>             COLVAR 2
>             INTERMOLECULAR
>             TARGET [angstrom] 1.87
>             TARGET_GROWTH [angstrom*fs^-1] 0.25
>             TARGET_LIMIT [angstrom] 5.0
>         &END COLLECTIVE
>     &END CONSTRAINT
>
> The result is that both atoms get closer too quickly and the system gets 
> an unphysical geometry - the Si kind of moves too far away from it's 
> lattice and the C abandons its hydrogens. The remperature rises too quickly 
> to unbelievable degrees, the calculation slows down and eventually the 
> program dies.
> My cell is defined just fine, with enough space for the gas reagent to 
> move around. I use a XYZ PBC with space left in one dimension so it 
> effectively becomes 2D. In that space is the reacting molecule.
> I tried playing with TARGET_GROWTH with many values from 0.01 to 5.0 and 
> the result is always the same.
> In this group I read that the moment TARGET is set the atoms are also set 
> at that distance, so I tried switching TARGET and TARGET_LIMIT, but even 
> then the result is the same - wrong geometry, they get too close too 
> quickly even with very low TARGET_GROWTH, too high temperature - program 
> dies.
> - Is this a bug, because it makes no sense as expected behavior ?
>
> I also tried using a RESTRAINT:
>
>     &CONSTRAINT
>         &COLLECTIVE
>             COLVAR 2
>             INTERMOLECULAR
>             TARGET [angstrom] 3.0
>             &RESTRAINT
>                 K [kcalmol] 1.0
>             &END RESTRAINT
>         &END COLLECTIVE
>     &END CONSTRAINT
>
> with the COLVAR being a plane between 3 surface atoms:
>
>         &COLVAR
>             &DISTANCE_POINT_PLANE
>                 &POINT
>                     TYPE GEO_CENTER
>                     ATOMS 8
>                 &END POINT
>                 &POINT
>                     TYPE GEO_CENTER
>                     ATOMS 21
>                 &END POINT
>                 &POINT
>                     TYPE GEO_CENTER
>                     ATOMS 28
>                 &END POINT
>                 &POINT
>                     TYPE GEO_CENTER
>                     ATOMS 81
>                 &END POINT
>                 ATOMS_PLANE 1 2 3
>                 ATOM_POINT 4
>             &END DISTANCE_POINT_PLANE
>         &END COLVAR
>
> and the gas phase reagent does not stop at the TARGET distance, instead it 
> gets inserted into the Si lattice, pushing Si atoms away, inserting both C 
> and H from the methane and one of the hydrogens flied far away. Temperature 
> goes anomalously high, program slows down, dies. At least it took some time 
> and steps to get to the wrong point.
> - Why didn't the molecule stop at the surface, while experiencing the 
> forces of the lattice atoms ?
> - Why did it do beyond the TARGET distance ?
> - Is this a bug ?
>
> Can I use metadynamics in any way to push the molecule towards the surface 
> for a reaction to occur ? Once again, I'm not interested in studying the 
> FES or the rest of the thermodynamics.
>
> I just need a way to get the gas phase reagent to occasionally get close 
> enough for a reaction.
>
> I think the answers to those question can be very useful for many.
>
> I attach some pictures of the anomalous results. The first two are from 
> the CONSTRAINT, the 3rd is from the RESTRAINT.
>
> Best Regards,
> Dobromir
>

-- 
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+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/3553d9d6-c461-4e6b-bd6c-8140bf36796cn%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20230120/3f144761/attachment-0001.htm>