Difference in electrostatic energy calculated by MS and LAMMPS

Dear Lammps experts,

I simulated a cross-linked epoxy resin system using MS and lammps based on Dreiding and obtained almost the same energy components except the electrostatic energy. The electrostatic energy in MS is about 700 and in lammps it is about 5000.The special bond is selected as Dreading so the 1-2 and 1-3 interaction should be excluded, and I also think the MS did the same thing. Another point confusing me is that when I change the accuray of the Ewald summation, Lammps is very sensitive but MS is stable. For example, lammps provide electrostatic energy of about 2500, 5000 and 7000 corresponding to Ewald accuracy of 1e-3, 1e-4 and 1e-5, but for MS, the energies are always around 700. Could you help to suggest some way that I can solve this problem?

Many thanks and regards

Xu Hao

Dalian University of Technology,PRC

Dear Lammps experts,

I simulated a cross-linked epoxy resin system using MS and lammps based on
Dreiding and obtained almost the same energy components except the
electrostatic energy. The electrostatic energy in MS is about 700 and in
lammps it is about 5000.The special bond is selected as Dreading so the 1-2
and 1-3 interaction should be excluded, and I also think the MS did the same
thing. Another point confusing me is that when I change the accuray of the
Ewald summation, Lammps is very sensitive but MS is stable. For example,
lammps provide electrostatic energy of about 2500, 5000 and 7000
corresponding to Ewald accuracy of 1e-3, 1e-4 and 1e-5, but for MS, the
energies are always around 700. Could you help to suggest some way that I
can solve this problem?

there is little that can be said about this, since the information you
provide is too vague and cannot be verified. most likely, you are
making some mistakes that you don't realize. for example, that you are
not computing the energy correctly. the fact that you compare numbers
without units is not inspiring much confidence. the fact, that you
don't mention which specific LAMMPS version you are using or what
platform you are running on, and that you are not providing a simple
way to reproduce your observations, makes your inquiry pretty useless.

here is an example for your reference.
if i amend the input of the bench/in.rhodo example by replacing the
"run 100" command with the following:

kspace_style pppm 1.0e-3
run 0 post no
print "E_coulomb = \(ecoul\) \+ (elong) = $(ecoul+elong)"

kspace_style pppm 1.0e-4
run 0 post no
print "E_coulomb = \(ecoul\) \+ (elong) = $(ecoul+elong)"

kspace_style pppm 1.0e-5
run 0 post no
print "E_coulomb = \(ecoul\) \+ (elong) = $(ecoul+elong)"

kspace_style pppm 1.0e-6
run 0 post no
print "E_coulomb = \(ecoul\) \+ (elong) = $(ecoul+elong)"

and then run using the current 16 March 2018 version of LAMMPS like
this: lmp_mpi -in in.rhodo | grep E_coulomb

i get the following:

E_coulomb = 154357.76227086147992 + -217703.9004825487209 =
-63346.138211687240982
E_coulomb = 207025.53004534213687 + -270403.7324560698471 =
-63378.20241072771023
E_coulomb = 252888.32078145124251 + -316269.11402794177411 =
-63380.793246490531601
E_coulomb = 294293.65748948295368 + -357674.78043339133728 =
-63381.122943908383604

as you can see, the electrostatic energy is consistently around -63380
kcal/mol and by far not changing to the degree you mention.

axel.