[gmx-users] rvdw and DispCorr

[Dommert Florian]

On Fri, 2012-05-11 at 10:23 +0200, Bernhard Knapp wrote: 
> Dear gromacs users
> 
> In a recent paper I found the following protocol of a gromacs simulation:
> 
> "All simulations were performed with the GROMACS 4.0 [12] compiled in single-precision mode at a constant temperature of 277 K in a periodic box with an edge length of approximately 8.2 nm and the default GROMOS-96 43A1 forcefield [22]. The simulation systems each contained approximately 16,500 Simple Point Charge (SPC) water molecules [23]. Short-range interactions were evaluated using a neighbor list of 1.0 nm updated at every 10 steps. Van der Waals interactions used a cutoff with a smoothing 
> function such that the interactions slowly decayed to zero between 0.75 nm and 0.90 nm. A long-range analytical dispersion correction was applied to the energy and pressure to account for the truncation of the Lennard-Jones interactions [24]. Electrostatic interactions were evaluated using the particle mesh Ewald (PME) [25] with a real space cutoff of 1.0 nm, a spline order of 6, a Fourier spacing of 0.1 m, and relative tolerance between long and short range energies of . All bonds to hydrogen 
> were constrained with LINCS [26] with an order of 12, and a time step of 2 fs was used for dynamics."
> 
> In the gromacs manual 4.5.4, page 104 it says: "The GROMOS-96 force field was parameterized with a Lennard-Jones cut-off of 1.4 nm, so be sure to use a Lennard-Jones cut-off (rvdw) of at least 1.4".
> 
> Is it a good idea to set "DispCorr" to "EnerPres" and reduce the rvdw so dramatically (almost the half value)?
> 

This depends on the sensitivity of force field to these settings. I
would try different cut-offs and compare some properties like the mass
density. In my studies I often realized that already a small change in
the cutoff ~0.1nm may give really different results, though a dispersion
correction is applied. This is a results from the procedure, the force
field is constructed. Usually all force field parameters are highly
correlated.  

> And a second question: Is there a study on the percentage of information getting lost when reducing the rvdw with and without dispcorr (e.g. to 1.2, 1.0, etc) if the forcefield was parameterized with 1.4?
> 

So I would expect, that if the system is already homogeneous at rc=0.9
and dispersion correction is applied, a further increase rc should give
the same results. However, if the system is not homogeneous at rc, an
increase in rc provides more information about the systems.


Actually, if there is no very good reason, to chose another cut-off for
LJ, as suggested by the force field authors, I would not change it.
However, if the homogeneity requirement is met, an small decrease of the
cut-off should not introduce too large artefacts. Always keep in mind,
that all force field parameters are highly correlated, which includes
also the cut-offs, switching or shifting methods, combination
rules, ... . The only parameters, where you often have some freedom
regard to electrostatics, in case no reaction field method is applied,
because then similar considerations as for the LJ cut-off apply. However
if an explicit calculation over all charges is used, like in Ewald-based
methods, you only have to assure that your settings conform to a certain
accuracy, given during the force field parametrization.

Cheers,
Flo





> best,
> Bernhard
> 
> 
> 

-- 
Florian Dommert
Dipl. - Phys.

Institute for Computational Physics
University Stuttgart

Pfaffenwaldring 27
70569 Stuttgart

EMail: dommert at icp.uni-stuttgart.de
Homepage: http://www.icp.uni-stuttgart.de/~icp/Florian_Dommert

Tel.: +49 - (0)711 - 68563613
Fax.: +49 - (0)711 - 68563658
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120511/e2cb7d49/attachment.sig>