Question on: The results about coul interaction energy obtained by Lammps and MS are very different

Dear All,
I want to simulate molecular diffusion properties in zeolites with lammps. I created .car and .mdf file of the structure with Material Studio and then I used them to create data file with msi2lmp and pcff. But I found the coul interaction energy obtained by Lammps and MS are very different. E-coul obtained by lammps is positive but MS is negative. A simple example about CHA zeolite is given as follows,just do minimization.The data and in files are in the attachment. Thank you very much!

Lammps results:

LAMMPS (31 Mar 2017)
units real
atom_style full
pair_style lj/class2/coul/long 12
bond_style class2
angle_style class2
dihedral_style class2
improper_style class2
kspace_style ewald 0.0001
pair_modify mix arithmetic

read_data CHA-36T.data
triclinic box = (-6.96522 2.09529 -5.05796) to (6.75128 13.9755 9.80515) with tilt (-6.85539 0.000726348 -0.00035957)
1 by 1 by 1 MPI processor grid
reading atoms …
108 atoms
scanning bonds …
2 = max bonds/atom
scanning angles …
6 = max angles/atom
scanning dihedrals …
6 = max dihedrals/atom
scanning impropers …
4 = max impropers/atom
reading bonds …
144 bonds
reading angles …
288 angles
reading dihedrals …
432 dihedrals
reading impropers …
144 impropers
4 = max # of 1-2 neighbors
6 = max # of 1-3 neighbors
16 = max # of 1-4 neighbors
20 = max # of special neighbors
neighbor 0.5 bin
neigh_modify delay 10

thermo_style multi
minimize 0.0001 0.005 100 500
WARNING: Resetting reneighboring criteria during minimization (…/min.cpp:168)
Ewald initialization …
WARNING: Using 12-bit tables for long-range coulomb (…/kspace.cpp:321)
G vector (1/distance) = 0.188538
estimated absolute RMS force accuracy = 0.0387266
estimated relative force accuracy = 0.000116624
KSpace vectors: actual max1d max3d = 28 2 62
kxmax kymax kzmax = 2 2 2
Neighbor list info …
update every 1 steps, delay 0 steps, check yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 12.5
ghost atom cutoff = 12.5
binsize = 6.25, bins = 4 2 3
1 neighbor lists, perpetual/occasional/extra = 1 0 0
(1) pair lj/class2/coul/long, perpetual
attributes: half, newton on
pair build: half/bin/newton/tri
stencil: half/bin/3d/newton/tri
bin: standard
WARNING: Inconsistent image flags (…/domain.cpp:785)
Per MPI rank memory allocation (min/avg/max) = 17.15 | 17.15 | 17.15 Mbytes
---------------- Step 0 ----- CPU = 0.0000 (sec) ----------------
TotEng = 2977.5571 KinEng = 0.0000 Temp = 0.0000
PotEng = 2977.5571 E_bond = 3.2879 E_angle = 195.1358
E_dihed = -23.1013 E_impro = 1.1617 E_vdwl = -50.4755
E_coul = 3373.9913 E_long = -522.4428 Press = 31153.9109
---------------- Step 8 ----- CPU = 0.0272 (sec) ----------------
TotEng = 2943.1443 KinEng = 0.0000 Temp = 0.0000
PotEng = 2943.1443 E_bond = 3.8698 E_angle = 182.1567
E_dihed = -43.8653 E_impro = 1.5510 E_vdwl = -49.1766
E_coul = 3371.1048 E_long = -522.4960 Press = 30627.9005
Loop time of 0.027262 on 1 procs for 8 steps with 108 atoms

99.0% CPU use with 1 MPI tasks x no OpenMP threads

Minimization stats:
Stopping criterion = energy tolerance
Energy initial, next-to-last, final =
2977.55711384 2943.3268794 2943.14432948
Force two-norm initial, final = 128.004 17.3911
Force max component initial, final = 16.8661 2.50815
Final line search alpha, max atom move = 0.0368907 0.0925273
Iterations, force evaluations = 8 16

MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total

CHA-36T.data (46.3 KB)

run_minimization.in (365 Bytes)

Dear All,
I want to simulate molecular diffusion properties in zeolites with lammps. I
created .car and .mdf file of the structure with Material Studio and then I
used them to create data file with msi2lmp and pcff. But I found the coul
interaction energy obtained by Lammps and MS are very different. E-coul
obtained by lammps is positive but MS is negative. A simple example about
CHA zeolite is given as follows,just do minimization.The data and in files
are in the attachment. Thank you very much!

absolute energies have no meaning in this context, as they can be
offset by arbitrary values (see pair_modify shift yes). also, ecoul
only contains the short range contribution, for systems with
long-range coulomb support, you have to add elong. however, those
still can have a different reference energy.

if you want to compare the interactions of the two codes, you need to
compare forces at step 0, and the geometries when converged.

axel.