# [lammps-users] how to extract the pair-wise energy from airebo

Dear LAMMPS Users,

I have a question about how to get pair interaction energy from AI_REBO
potential. First, I try to use compute group/group command by defining
two atoms in two groups respectively using the script below:

pair_style airebo 3.0 1 0
pair_coeff * * \$LAMMPS/potentials/CH.airebo C
group sub1 id = 150
group sub2 id = 148

compute 1 sub1 group/group sub2

thermo_style custom step temp etotal press evdwl c_1 c_1[1] c_1[2]
c_1[3]
neigh_modify delay 0 every 1 check yes
minimize 1.0e-8 0.00001 1000 10000

The atoms, ID 150 and 148, are carbon atoms with the distance at around
1.4A, which is within the cut-off range of REBO potential. But the
output is 0 for c_1 and the forces are incredibly small.

I notice the doc for group/group says that it can not be used to
calculate the pair energy with 3-body terms, so i suppose the airebo
falls in this case, as it has cos(theta_ijk) terms in bij.

So I am asking is there any easier way to get this information (pair
wise interaction energy) out from lammps other than by measuring the
angles and distance between these related atoms and calculate the Vij
from original REBO equation. I know the evdwl in airebo.cpp is the total
energy from all atoms. Is there anyway to extract two atom interaction
from these?

Any suggestions are welcome. Thanks in advance!

Regards
Tianyi

So I am asking is there any easier way to get this information (pair
wise interaction energy) out from lammps other than by measuring the
angles and distance between these related atoms and calculate the Vij
from original REBO equation. I know the evdwl in airebo.cpp is the total
energy from all atoms. Is there anyway to extract two atom interaction
from these?

LAMMPS will compute per-atom energy for AI-REBO which is the
energy of an atom due to all its REBO interactions. LAMMPS cannot
compute just the VDWL piece for you. I'm not sure it even makes sense
for just 2 atoms. You could try to instrument the pair_airebo.cpp code
yourself to get what you want. Or if the formula is simple, then define
a variable that computes it.

Steve