[gmx-developers] Zero short-range energy of TIP3P

[Berk Hess]

Are you running on CPU or GPU?

/Berk

On 2021-02-12 12:23 , Igor Leontyev wrote:
> Sorry, for the confusion. I meant of course group-rest terms.
> Assuming my energy group includes 1 TIP3P molecule and has name WAT1.
> I got zeroes when I extract energy for terms:
> Coul-SR:WAT1-rest
> LJ-SR:WAT1-rest
>
> but they should include non-zero interaction with all neighbors within 
> the Cutoff.
> ------------------------------------------------------------------------
> *From:* Igor Leontyev <ileontyev at ucdavis.edu>
> *Sent:* Friday, February 12, 2021 2:09 AM
> *To:* gmx-developers at gromacs.org <gmx-developers at gromacs.org>
> *Subject:* Re: Zero short-range energy of TIP3P
> Thanks, Berk, for the response. I understand that a single water 
> molecule in the vacuum box should have no Short-Range energy. But I am 
> talking about water molecule in the BULK, i.e. there are a lot of 
> explicit H2O around. I expect that SR energy (with PME electrostatics) 
> includes at least LJ interactions with all neighbors within the 
> CutOff, and may be also Coul interactions (not sure how PME splits the 
> energy into SR and LR). But I get exact zeroes for both LJ and Coul terms.
>
> Igor
> ------------------------------------------------------------------------
> *From:* Igor Leontyev
> *Sent:* Thursday, February 11, 2021 10:00 PM
> *To:* gmx-developers at gromacs.org <gmx-developers at gromacs.org>
> *Subject:* Zero short-range energy of TIP3P
> Dear gmx developers,
> The short range energy terms (Coul and VdW) for a single water 
> molecule in the bulk (rigid TIP3P) simulated with gmx-2020 and 
> extracted from edr-file are exact zero. The result looks strange to 
> me. The same but with gmx-5.1 gives non-zero (presumably correct) 
> interaction energy of water.
>
> How I do it:
> 1. Generate an index file with indices of the specific water molecule 
> in the bulk (e.g. for the 1st H2O it is just 1 2 3) to be monitored.
> 2. Specify the energy group of the molecule to be monitored in mdp 
> file and run MD.
> 3. After MD, use gmx energy to get LJ and Coulomb (SR) interactions of 
> the energy group from the edr file.
> The computed with gmx-2020 non-bonded (SR) energies of this energy 
> group are zero.
>
> Please, let me know if I am doing something wrong. Otherwise, it looks 
> like a bug because old version (gmx-5.1) results in non-zero energies 
> that looks reasonable.
>
> Thanks,
> Igor
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20210212/0a8c1b7b/attachment.html>