[lammps-users] How to compute the forces between two groups of atoms?

Hello Steve:

From the LAMMPS manual, it is clear that compute group/group command id can only compute the pair potential between two groups of atoms.
Can this OK?
compute 1 group_ID2 property/atom fy
compute 2 group_ID2 reduce sum fy
or
compute 1 group_ID2 property/atom fy
compute 3 group_ID2 reduce sum c_1

from Steve Plimpton :

2011/1/28 Zhang Yan <[email protected]>:

Hello Steve:

From the LAMMPS manual, it is clear that compute group/group command id can only compute the pair potential between two groups of atoms.

yes! that typically is the only interaction that needs to be considered.
please demonstrate how and where your system is different.

Can this OK?
compute 1 group_ID2 property/atom fy
compute 2 group_ID2 reduce sum fy

or
compute 1 group_ID2 property/atom fy
compute 3 group_ID2 reduce sum c_1

no, this gives you the net force on a group,
but _not_ the force due to another group.

axel.

Hi Axel:

I want to do nanoindention MD simulation.
The indentor is treated as a rigid using the command:" fix 3 indent setforce 0 0 0".
The indentor is imposed to velocity loading.
Sample atom is Fe, and indent atom is C.
Fe-Fe and Fe-C interaction is used Finnis-Sinclair potential
C-C interaction is used Tersoff potential
Both Finnis-Sinclair and Tersoff potential are many-body potentials
Can the interact forces between the indent and the sample be commputed using the command:
“compute ID sample group/group indent”?

By the way, another 2 questions?
Velocity loading is suitable or not for indention?
Potentials between the sample Fe and indentor C is suitable or not by using Fe-C Finnis-Sinclair potential, as Fe and C could react and adhere to each other? If yes, can we get better potentials or metheds for indention?

Axel Kohlmeyer:

The fix setforce command stores (and the values can
be accessed in your script) the total force on its
atoms (the indenter in your case) before it alters
them. That appears to be what you want.

The other questions (vel loading, appropriate potential)
are not really LAMMPS questions. You should read
papers about how people model dynamic indentation
and first try to mimic those methods in LAMMPS.

Steve

2011/1/28 Zhang Yan <[email protected]>: