LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] What interactions are computed for pairwise energy?

# Re: [lammps-users] What interactions are computed for pairwise energy?

 From: Axel Kohlmeyer Date: Fri, 25 May 2018 20:18:32 -0400

On Fri, May 25, 2018 at 8:06 PM, Wujie Wang wrote:
Hey lammps users

I am trying to figure out what goes into the neighbor list.

In my toy system, I have 10 atoms of the same type. The pairwise potential is defined in the following manner:

neighbor 0.4 bin
neigh_modify every 10 one 10000
bond_style      harmonic
bond_coeff 1 350 1.53
pair_style lj/cut 10.5
pair_coeff 1 1 0.112 4.01 10.5

In theory, there are should be 45 interactions in the pairwise potential energy, is that right?

​most likely not. there is crucial information missing:
- size of the simulation cell and choice of boundary conditions
- list of bonds and special_bonds settings​

​please note, that LAMMPS does not apply minimum image conventions, but creates "ghost atoms", which are real periodic copies.​

However, I am getting gigantic energy when try to compute this energy by “hand", but the Lammps output only gives really small energy like the following.

Step PotEng KinEng E_bond E_pair Temp
0    3280.8242    134.13645    3280.8521 -0.027946372         5000
50    76.787747    233.18891    77.089967  -0.30221985    8692.2273
100    22.287439     27.21953    23.011246  -0.72380721    1014.6209
150    6.4999038    12.623158    7.3654589  -0.86555507     470.5342

If you just look at the distance, the bond distance is around 1.53, the characteristic distance LJ is 4.01, so the pairwise energy between bonded atom should be big, which explains the huge energy I have been getting by “hand”.

So does this have something to do the neighbor function? But neighbor only adds a “skin” distance to the pair cutoff and should not affect the total energy of atoms that are close to each other.

I think I generally lack of a understanding of Lammps mechanics. I would be really helpful if someone can point out what is wrong.

​you should read the LAMMPS paper and then the relevant sections of the manual. you seem to be making too many (and mostly incorrect) assumptions.

axel.​

Thank you!

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most