[gmx-developers] Fw: [gmx-users] Fw: Two questions about PME-User in coulombtype

[Berk Hess]

Hi,

I don't know if I can still mail to the gmx-users list, so I answer 
here, but please post my answers below to the list.

Cheers,

Berk

Answers:

1) Correct.
2) Correct.
3) PME splits the Coulomb interaction into two parts and this splitting 
depends on beta. All you need to know is that up to the cut-off 
distance, you exactly get the interaction you set in the table. Beyond 
the cut-off you get 1/r.

Your potential values should have 10 decimals to be sure that rounding 
errors are negligible. In principle, also r should have 10 decimals, but 
0.0020000000 is identical to 0.002 (and 0.0000000000 to 0.000 or 0).

Cheers,

Berk

On 2017-10-17 15:15, Du, Yu wrote:
> Dear GMX Developers,
>
> I'm sorry for repeting sending this email to the mail list. I think the PME-User is not common for users to use and the related explanation is scarce.
>
> Hope you gurus can help me. Thanks a lot.
>
> Yu
>
>
> -----Original Messages-----
> From: "Du, Yu" <duyu at sioc.ac.cn>
> Sent Time: 2017-10-15 17:17:14 (Sunday)
> To: GMX-user <gromacs.org_gmx-users at maillist.sys.kth.se>
> Subject: [gmx-users] Two questions about PME-User in coulombtype
>
> Hi GMX Users,
>
>
> I'm using tabulized potentials between energygrp-table = Protein Ligand. I'm writing to inquiring the details of PME-User parameter.
>
>
> I have seen some discussions in the GMX User mail list and the explanation in user guide. I still can't ensure I have grasped the meaning of PME-User in coulombtype option. The following is my understanding. If anything is wrong please point out. Thanks a lot.
>
>
> 1) Because the mesh part also contributes below the cut-off, the PME mesh contribution is subtracted from the user table by gmx mdrun to get the short-range coulombic interaction below the rcoulomb.
>
>
> 2) The tabulized potential is short-range interaction. If I use coulombtype=User, I only have the short-range interaction defined in the tabulized potetial files and no PME between any components (i.e. Protein, Ligand, SOL, ions) in the system.
>
>
> 3)If I use coulombtype=PME-User instead, I will get the short-range interaction defined in the tabulized potetial files and PME between any components (i.e. Protein, Ligand, SOL, ions) in the system like normal coulombtype=PME.
>
>
> --Next Question--
> "The PME mesh contribution is subtracted from the user table by gmx mdrun. Because of this subtraction the user tables should contain about 10 decimal places."
>
> The second question is how to deal with the PME-User's "10 decimal places" in user guide of the coulombtype option. Does it mean that the original step of r = 0.002 nm in mixed precision should be 0.000000000# nm? What should be the detailed step size of r?
>
>
>