[gmx-users] Using lennard jones and buckingham terms simultaneously
Mark.Abraham at anu.edu.au
Thu Mar 25 22:07:44 CET 2010
On 26/03/2010 7:03 AM, Gareth Tribello wrote:
> Hello again
> I have tried to do as you suggest and use tables but I have a new
> problem. First let me describe my process and then you can let me know
> if there is anything wrong in the stages:
> OK so first you include the following directives into the mdp file:
> coulombtype = pme (or whatever sort of coulomb interaction you are using)
> vdw-type = user
> energygrps = Ca CCA OCA OW HW
> energygrp_table = Ca OCA Ca CCA OCA OCA OCA OW OCA HW
> Gromacs is then (at some stage) going to look for a series of files called
> table.xvg - which is the default 6-12 Lennard Jones that will be used
> for most of the atoms
> table_Ca_OCA.xvg - which are the Buckingham interactions between your
> various atom types.
> These files have the format (and contents) described in section 6.7 of
> the manual. Finally, you define the various energy groups Ca, OCA and
> so on in your index.ndx file.
> The problem is that grompp gives me the following error:
> "atoms 1 and 2 in charge group 1 of molecule type 'SOL' are in different
> energy groups"
> (incidentally these atoms 1 and 2 are OW and HW)
> Does this mean that I cannot use different tabulated potentials for
> different atoms in a molecules? By which I mean that I can't use
> different tabulated potentials for the OW Ca and HW Ca interactions for
Charge groups are the fundamental unit GROMACS uses in constructing a
simulation. Energy groups are the next "higher" layer in the data
structures, and these must be sets of whole charge groups. With some
electrostatics models, looping over charge groups whose charge is
preferably an integer is essential for modelling correct behaviour.
GROMACS does a complex sorting of all the interactions between charge
groups into lists that allow it to iterate over charge groups and energy
groups. A user table then gets applied to a whole intra- or inter-
energy-group loop. Thus your attempt violates this precondition.
However, PME does not require the use of charge groups for accurate
results, since all inter-atomic electrostatic interactions get treated,
regardless of distance. So you could decompose your water molecules into
two charge groups, O and Hs. (Caveat, a near-brokenly bad PME
approximation might get a little worse with arbitrary charge groups)
> Final question, as its not clear to me from the manual, if you use a
> tabulated potential for Lennard Jones and you use mix type 2 (so are you
> are providing epsilon and sigma in the input rather than A and B) does
> gromacs still know that it has to manipulate the input parameters in
> order to get the coefficients of the (tabulated) g(r) and h(r)
> dispersion and repulsion functions (I mean the g(r) and h(r) defined in
> section 6.7 of the manual here)? At the same time does it also know not
> to do anything to the parameters you input for the (tabulated)
> buckingham potentials (as for a buckingham you are providing A and C)?
I expect the point of the tables is that GROMACS just uses them per
equation 6.23. Thus I'd expect C6 and C12 in that equation to be
constructed according to whatever combination rule is in force. If
you've specified them explicitly in the topology (see chapter 5), then
they will not be constructed.
You should be very careful to test your assumptions and deductions about
how all this machinery is working. Do take the time to set up a
tabulated version of a quick non-tabulated calculation to at least make
sure you've done the simple things right! To test the function of eq
6.23, do a non-table constructed-parameter calculation, a non-table
topology-specified-parameter calculation, a table constructed-parameter
calculation, etc. to be sure you understand what is going on - and that
the code works right! Please consider contributing any insights to a
wiki page on the GROMACS webpage.
Before you go to all this work please consider my advice of last email.
Unless you have an existing reason to expect some ad-hoc combination of
different interaction functions to work well together, you may find that
your best possible result is a correctly-functioning random number
> On Wed, Mar 24, 2010 at 2:25 AM, Mark Abraham <mark.abraham at anu.edu.au
> <mailto:mark.abraham at anu.edu.au>> wrote:
> ----- Original Message -----
> From: Matthew Watkins <matthew.watkins at ucl.ac.uk
> <mailto:matthew.watkins at ucl.ac.uk>>
> Date: Wednesday, March 24, 2010 2:59
> Subject: Re: [gmx-users] Using lennard jones and buckingham terms
> To: Discussion list for GROMACS users <gmx-users at gromacs.org
> <mailto:gmx-users at gromacs.org>>
> > Hi Gareth,
> > as Vitaly suggested tabulated potentials seem to be the only way
> > to go, it took me a while to get up to speed on the Gromacs way
> > of doing this, so get in touch if you wish.
> > The tables for buck potentials need to include the standard 1/r6
> > term whilst what would be the 1/r12 term needs to contain exp(-
> > Bx.rho), the C6 and C12 coefficients can then be put in a
> > standard nonbonded section. You'll need a separated table
> > for each pair of interactions that interact with buckingham
> > potential. Each pair must be an energy group as well.
> > If there is a simpler method I'd love to hear it.
> There's probably not a simpler method because it's not a widely-used
> procedure. It shouldn't be used at all unless you have established
> that the combination of functional forms is effective...
> gmx-users mailing list gmx-users at gromacs.org
> <mailto:gmx-users at gromacs.org>
> Please search the archive at http://www.gromacs.org/search before
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org
> <mailto:gmx-users-request at gromacs.org>.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
More information about the gromacs.org_gmx-users