[gmx-developers] comb-rule=2 support for forcetable

Sun Dec 1 10:13:29 CET 2019

Hi Yu,

Yes. The reason for this is that tabulated potentials require you to provide the form of the repulsion and dispersion parts of the interaction separately. Otherwise it would never be possible to tabulated e.g. a switched or shifted interaction.

At first it sounds like a solution to rather tabulated

x^-12 - x^-6

And then use x=sigma/r, but then all cutoffs and shifts in the table would be applied to x rather than r, which would violate a bunch of physics.

So, in order to apply dispersion and repulsion separately we must have separate coefficients to apply for each interaction, and that way of specifying combination rules corresponds to normal c6/c12 with geometric combinations.

However, as Berk wrote there's a trick you can use: 

If you read the topology section of the manual (in detail, sorry, it's somewhat condensed, but all the info is there), just as we have a section for pairrtypes (1,4 parameters) you can create a separate section for nonbonded parameters and explicitly specify c6 and c12 for each pair of atomtypes (have a look at the old Gromos force fields for an example).

You will have to translate from sigma/epsilon with a small script, and the section might be large if you have a ton of atomtypes, but it will work.

We might be able to automate this better when enabling the verlet tables, but that's still a bit away ;-)

Cheers,

Erik

--
Erik Lindahl <erik.lindahl at scilifelab.se>
Professor of Biophysics
Science for Life Laboratory
Stockholm University & KTH
Office (SciLifeLab): +46 8 524 81567
Cell (Sweden): +46 73 4618050 
Cell (US): 1 267 307 8746

> On Dec 1, 2019, at 02:42, Du, Yu <duyu at sioc.ac.cn> wrote:
> 
> Hi Prof. Hess, 
> 
> Thanks for your reply.
> 
> It is because user table requires comb-rule=1 that I have to change the comb-rule of AMBER and CHARMM 
> from 2 to 1. I realy have no special need for comb-rule!=1. So could you give some hints of "You can 
> list parameters for all combinations of atoms types you need/want explicitly."? I don't know the way 
> to explicitly list the combined values for all atom pairs.
> 
> Yes, I will take your caveat of larger errors.
> 
> Cheers, 
> 
> Yu
> 
> -----Original Messages-----
> From:"Berk Hess" <hess at kth.se>
> Sent Time:2019-12-01 02:26:54 (Sunday)
> To: gmx-developers at gromacs.org
> Cc: 
> Subject: Re: [gmx-developers] comb-rule=2 support for forcetable
> 
> Hi,
> 
> But you don't need to change the combination rule. You can list parameters for all combinations of atoms types you need/want explicitly.
> 
> Note that changing the combination rule for a force field likely results in much larger errors than the differences between different force fields.
> 
> Cheers,
> 
> Berk
> 
> 
> On Nov 30, 2019 14:36, "Du, Yu" <duyu at sioc.ac.cn> wrote:
> Hi Prof. Lindahl, 
> 
> Thanks for your reply.
> 
> If user table is used e.g. coulombtype=PME-user and vdwtype=user, I think that the only difference 
> of geometric (comb-rule=1) and arithmetic (comb-rule=2) average of sigma that can't be eliminated 
> by editing the value in force field.
> 
> comb-rule=1 matches GROMOS force field which is deprecated for its physically incorrect scheme.
> comb-rule=2 matches AMBER and CHARMM force fields which I want to use in the a compatible comb-rule
> under circumstances of user table.
> 
> As you suggested, I indeed have copied and modified AMBER and CHARMM ff with a new name :), but in 
> comb-rule=1 for user specified nb potentials between groups. This is why I'm asking for a comb-rule=2
> support.
> 
> By the way, I notice that forcetable in Verlet scheme has been under development. I'm excited about
> it and hope it can move to GPU in the future.
> 
> Regards,
> Yu
> 
> -----Original Messages-----
> From:"Erik Lindahl" <erik.lindahl at gmail.com>
> Sent Time:2019-11-30 17:08:41 (Saturday)
> To: gmx-developers at gromacs.org
> Cc: 
> Subject: Re: [gmx-developers] comb-rule=2 support for forcetable
> 
> Hi Yu,
> 
> That won't really require any changes in the code.
> 
> First, the reason we can't "support" it is that force field Lennard-Jones parameters have been determined for a specific set of combination rules and cutoffs, and when you change the combination rules you will change the result of the simulation. How good or bad that is depends on what you are doing, but it's something you'd have to determine and be ready to defend when publishing.
> 
> Having said that, if you absolutely want to change the setting for a particular force field, you can just edit the value in the force field file. I would strongly advise making a copy of the entire force field directory and changing its name to reflect this, so you don't forget and have an altered installed version of the force field ruin other simulations later.
> 
> Cheers,
> 
> Erik
> 
> On Sat, Nov 30, 2019 at 3:11 AM Du, Yu <duyu at sioc.ac.cn> wrote:
> Dear GMX Developers, 
> 
> The force table function (src/gromacs/tables/) in GROMACS is a magic, it helps to achieve a lot of work.
> Could you please support it in comb-rule=2 for AMBER and CHARMM FF? 
> 
> Sincerely, 
> Yu
> -- 
> Gromacs Developers mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before posting!
> 
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> 
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers or send a mail to gmx-developers-request at gromacs.org.
> 
> 
> -- 
> Erik Lindahl <erik.lindahl at dbb.su.se>
> Professor of Biophysics, Dept. Biochemistry & Biophysics, Stockholm University
> Science for Life Laboratory, Box 1031, 17121 Solna, Sweden
> 
> 
> --
> Du, Yu
> PhD Student,
> Shanghai Institute of Organic Chemistry
> 345 Ling Ling Rd., Shanghai, China. 
> Zip: 200032, Tel: (86) 021 5492 5275
> -- 
> Gromacs Developers mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before posting!
> 
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> 
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers or send a mail to gmx-developers-request at gromacs.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20191201/d476836b/attachment.html>