Inconsistent Coulombic energies with KSpace solver

Dear all,

I am trying to use nvt+gcmc to deal with water system (yes, only a test, for more than half a year).

The code works very well with CO2 gas system, but so far not for liquid water.

For liquid water, nvt and npt simulation also agree with the results from literature. When I turn to nvt+gcmc, no correct results are obtained.

I am trying to find out why it does not work. The possible reason may be that “WARNING: Neighbor exclusions used with KSpace solver may give inconsistent Coulombic energies (…/neighbor.cpp:454)”, because this warning does not show up when I ran nvt or npt simulations.

It seems this warning is caused by or due to “fix gcmc”, but why? Since I set “pair_style lj/cut/coul/long 10.0”, for both “nvt” and “nvt+gcmc” the force and potential should be calculated in the same way, why the warning is only given for “nvt+gcmc”? Actually, when I turn to “nvt+gcmc” only a fix-gcmc command line is added to the job-script (those command lines such as “compute_modify mdtemp dynamic yes” are added, too).

Another question is about the molecule template. In the tutorial it reads about “Masses section”:

“This section is only allowed for atom styles that support per-atom mass, as opposed to per-type mass. See the mass command for details. If this section is not included, the default mass for each atom is derived from its volume (see Diameters section) and a default density of 1.0, in units of mass/volume.”

According to my understanding, masses section have to be provided otherwise an estimation of atom-mass will be used. So, I give the following masses section for water:

Masses

(blank line)

1 15.9994
2 1.00794
3 1.00794

However, I got the following warning, “WARNING: Molecule attributes do not match system attributes (…/molecule.cpp:1350)”

If I delete Masses section, this warning does not come out.

I am really confused about it.

I will be very appreciated if you can help me with these questions!

Best wishes,

Yongbiao

Regarding the question about masses, the atom_style doc page is pretty clear:

“All of the point-particle styles assign mass to particles on a per-type basis, using the mass command, The finite-size particle styles assign mass to individual particles on a per-particle basis.”

Regarding your question about the neighborlist exclusion warning for Coulombics, you are right to be concerned, but our testing indicates that the particular exclusion used by fix gcmc gives consistent results for Coulombics, because the charges on the excluded atoms are also temporarily set to zero.

Regarding your question about liquid water “no correct results are obtained,” there is no question here that I can answer.

Aidan

Dear Aidan,

Thanks for your help!

As far as “our testing indicates that the particular exclusion used by fix gcmc” (mentioned in your email) is concerned, it means you do have some testing systems for fix-gcmc. Are they gases or liquids? Is it possible to share the input files (especially the liquid system)? I will be very appreciated if you can share them.

Best regards,

Yongbiao