[CP2K-user] Compatibility problem of manybody potentials with two-bodies (LJ, Williams) in FIST

[Marcella Iannuzzi]

Hi,

I am not 100% sure but it is possible that the Tersoff potential is not 
compatible with one non bonding potential.
You can solve the problem running a multiple force environment calculation, 
where you define two independent force environments, one for
Tersoff and one for LJ. The global force should be the simple sum of the 
two contributions. 
Please check for reference the examples of multiple force environments that 
you can find among the cp2k/tests
Kind regards
Marcella


On Wednesday, February 13, 2019 at 11:05:43 AM UTC+1, p clabaut wrote:
>
> Hello,
>
> I'm trying to use a manybody potential between a metallic slab and water 
> molecules in FIST but I would also need to use a simple Lennard-Jones 
> potential to govern my water/water interactions.
>
> Unfortunately, it would seems that the neighbour list building went badly 
> between those modules because as long as my manybody potential (here, 
> Tersoff) is present in the input, some of the Oxygen/Oxygen Lennard-Jones 
> interactions are ignored.
> Moreover, the verlet_skin parameters starts switching on or off some of 
> these interactions.
>
> For an example, I looked at two couples of water molecules away from my Pt 
> slab and distant for each other.
>
> - When I use 
>     &NEIGHBOR_LISTS
>       VERLET_SKIN 0.00
>     &END
> with 
>     &TERSOFF
>         ATOMS O Pt
>         RCUT [angstrom] 10
>      &END TERSOFF
> and 
>     &LENNARD-JONES
>           atoms O O
>           EPSILON [kcalmol] 0.102
>           SIGMA [angstrom] 3.188
>           RCUT 10
>      &END LENNARD-JONES
>
> the oxygens of only one of my water couple are experiencing LJ interaction 
> from each others while the others doesn'tfeel any forces at all 
>
> - When I use 
>     &NEIGHBOR_LISTS
>       VERLET_SKIN 1.00
>     &END
> with 
>     &TERSOFF
>         ATOMS O Pt
>         RCUT [angstrom] 10
>      &END TERSOFF
> and 
>     &LENNARD-JONES
>           atoms O O
>           EPSILON [kcalmol] 0.102
>           SIGMA [angstrom] 3.188
>           RCUT 10
>      &END LENNARD-JONES
>
> with no others changes, it is the second water couple that undergoes LJ 
> interaction but not the first one anymore.
>
> - When I use 
>     &NEIGHBOR_LISTS
>       VERLET_SKIN 1.00
>     &END
> and 
>     &LENNARD-JONES
>           atoms O O
>           EPSILON [kcalmol] 0.102
>           SIGMA [angstrom] 3.188
>           RCUT 10
>      &END LENNARD-JONES
>
> with no manybodypotential, both couples are resenting LJ interactions.
>
>
>
> I have observed the same exact behavior with a Williams potential instead 
> of a Lennard-Jones. I'm joining the example input in the message.
>
> Could you tell me what is going and if it is intended or a bug in the 
> programm?
>
> Thanks you by advance !
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190217/30c66c66/attachment.htm>