Shake does not accurately identify bonds and angles causing constraints to fail

LAMMPS LAMMPS Beginners
1

Dear all,

When I was performing nvt simulations of a pure water system between graphene electrodes, the water used a rigid SPC/E model.
I used de-constrained bond angles for the water molecules. However, I found that lammps shake could not accurately identify the bonds angles for constraining (this is reflected in the log file). This caused my water molecules to deform so drastically that the two hydrogen atoms were almost next to each other. In my data file, atoms 1-12 are graphene electrodes, and atoms 13 and 14 are O and H of water molecules, for a total of 2823 water molecules.

data:
LAMMPS data file via write_data, version 22 Dec 2022, timestep = 0

12501 atoms
14 atom types
5646 bonds
1 bond types
2823 angles
1 angle types

0 29.473 xlo xhi
0 29.778 ylo yhi
0 270 zlo zhi

Masses

1 12.011
2 12.011
3 12.011
4 12.011
5 12.011
6 12.011
7 12.011
8 12.011
9 12.011
10 12.011
11 12.011
12 12.011
13 15.9994
14 1.008

in:
fix shake all shake 0.0001 20 0 a 1 b 1

log:
Finding SHAKE clusters …
0 = # of size 2 clusters
2823 = # of size 3 clusters
0 = # of size 4 clusters
0 = # of frozen angles
find clusters CPU = 0.001 second
Actually, it should be 2823 angles, not key length, and frozen angles is 0.
I confirm that my force field parameters are fine. I also tested the versions lammps-23Jun2022, lammps-29Sep2021, lammps-8Feb2023, all with the same error.

Here, I attach my in and data files and the output log file.
run_nvt.zip (364.8 KB)

Any help will be appreciated!

Thanks!

2

Hello,

Your data file is weird. Your first angle is defined as :

1 1 4051 4052 4053

but 4051 is the oxygen and 4052 4053 are hydrogens:

4051 4039 13 -0.8476 5.13006185012646 1.8299576317485482 24.00998437971701 0 0 0
4052 4039 14 0.4238 5.529896225747802 2.719769102288858 23.790057075838707 0 0 0
4053 4039 14 0.4238 5.819122063028692 1.1209033844197522 23.860190855863326 0 0 0

Which means that you created H-H-O bonds instead of H-O-H. Was it intended?

Simon

1 Like
3

Dear simongravelle,

Thank you for identifying my very low-level mistake, which led to me wasting your time. I was the one who was careless! I think the problem has been found!

Thanks again!