[lammps-users] Question about bond_style gaussian

Dear all,
Recently, I noticed that lammps new version package named"lammps-patch_30Nov2020.tar.gz" added new bond and angle style, i.e., bond_style gaussian & angle**_style gaussian**. And I create a bead-spring chain to check the bond_style gaussian. I thought the final bond length distribution should be a single gaussian shape, but the result is not perfect, and the atoms are separated finally in system. The input file shown here:

############### Gaussian Potential Test ###############
units real
atom_style full
boundary p p p

bond_style gaussian
angle_style harmonic
dihedral_style harmonic
pair_style lj/gromacs 12.0 15.0
pair_modify mix arithmetic

read_data Beads.data

bond_coeff 1 300 1 0.0128 0.375 3.0 # New gaussian bond potential
angle_coeff 1 50.0 92.1
dihedral_coeff 1 0.111111 1 3
pair_coeff 1 1 0.0957 3.03315

neighbor 2.5 bin
neigh_modify delay 5

variable TI equal 1000 # Initial Temperature
variable TF equal 300 # Final Temperature

timestep 0.5

FA9FBD4F-86F5-4257-94DC-D7D285A53E6D.png

Gaussian_test.in (2.78 KB)

unwrap-1.dcd (391 KB)

Beads.data (777 Bytes)

All.lammpstrj (147 KB)

it is a bad idea to make assumptions about functionality of a software from the name of the command alone.
the commands you are referring to are well documented (including the functional form of their potential function, which contains a gaussian term, hence the name) and contains a reference to the publication describing the potential and its intended applications.

in addition, you need to keep in mind that you are running a combination of interactions including non-bonded terms and thus what you see is the result of the combination of the interactions and will not automatically represent the distribution expected for the individual contribution unless the functional forms of those interactions are such that one is negligible for geometries when the other is significant.

for more specific inquiries, you should contact the contributing author of these styles.

axel.