Error while using "Grid" option

[Vardhan]

Dear Professor,

1. I am doing gas adsorption studies using GCMC.  To speed up my simulations, I have used the Grid option. By using the following input file, I am able to make energy grid.

Input file_1
SimulationType  MakeGrid

Forcefield      test
UseChargesFromCIFFile   yes

CutOff                        14.0

ChargeMethod                  Ewald
EwaldPrecision                1e-10

Framework 0
FrameworkName  framework
UnitCells 2 2 5
ExternalTemperature 298.0

NumberOfGrids 3
GridTypes H_s S_s S_com
SpacingVDWGrid 0.1
SpacingCoulombGrid 0.1
Input file end

I have used H2S molecule (4-3 Trappe model). There is one lone pair which is defined as an atom. So, in the .def file I have included "0 3 RIGID_BOND" command, to define the third bond. I have given S_com  notation for it.
Now, I used these grids for adsorption studies (GCMC), to speed the simulations. But I am getting a warning something like "Energy density interpolation probably not accurate enough". As I checked, Boltzmann relative error was very high for S_com only.  Other atoms doesn't have any error. The input file is as follows.

Input file_2
SimulationType                MonteCarlo
NumberOfCycles                50000
NumberOfInitializationCycles  50000
PrintEvery                    1000

Forcefield                    test
UseChargesFromCIFFile         yes

CutOff                        14.0

ChargeMethod                  Ewald
EwaldPrecision                1e-6

Framework 0
FrameworkName framework
UnitCells 2 2 5
ExternalTemperature 298.0
ExternalPressure 1e5

NumberOfGrids 3
GridTypes H_s S_s S_com
SpacingVDWGrid 0.1
SpacingCoulombGrid 0.1
UseTabularGrid yes

Component 0 MoleculeName                     H2S
            MoleculeDefinition               TraPPE
            TranslationProbability           0.5
            RotationProbability              0.5
            SwapProbability                  1.0
            CreateNumberOfMolecules          0
Input file end

I have used the correct grid, and the parameters are also the same. Can you suggest a way to overcome this error?


Thank you.

[David Dubbeldam]

These are warnings, it is up to you to judge how good the interpolation is.
For atom-types that have no VDW interaction, the atoms can be placed on top of other atoms during the test. This lead to large errors in the energy/force interpolation, but this would not occur during a real simulation.
Also, you can select which atom types to use with grid-interpolation (and which not).

[Vardhan]

Thank you for the reply professor.

"For atom-types that have no VDW interaction, the atoms can be placed on top of other atoms during the test. This lead to large errors in the energy/force interpolation, but this would not occur during a real simulation."  Could you please elaborate on this point??

I am not able to understand, why the lone pair pseudo atom alone is showing error while using grid energy interpolation.

Also, for example, let us say I have made a grid for H, S, and L (of H2S), and C and O (of CO2). Now, in the final simulation, I call only H, S, C, and O grids alone, for interpolation. Is this approach appropriate? Can I expect that the energies (VDW and coloumbic) of 'L' with the framework are calculated separately?

Awaiting for your reply.

Thank you.

[David Dubbeldam]

You can select which atoms to use for interpolation. The other ones will be computed in the usual way without interpolation.
The test at the beginning is limited and just meant to catch obvious wrong interpolations. But it is up to the user to judge whether the interpolation is sufficiently accurate or not. A few test of e.g. isotherms with and without grids can shows whether the grid is accurate enough.