[gmx-users] Viscosity calculations

Justin A. Lemkul jalemkul at vt.edu
Mon Jan 17 13:11:29 CET 2011

Xiaohu Li wrote:
> Hi, 
>    I'm trying to calculate viscosities of a few ionic liquid and has 
> roughly read about Hess's paper JCP 116 209, 2002.
> Follows are the questions that I have
> (1) For the method which uses the fluctuations of the pressure tensor 
> using Green-Kubo relation. I used g_energy -f *.edr -s *.tpr -vis option 
> and has obtained
> a few files such as evisco.xvg, eviscoi.xvg and visco.xvg
> The evisco.xvg seems like the one I'm interested, since it has the title as
> @    title "Shear viscosity using Einstein relation" 
> and the unit seems also correct. This file has 5 columns, if I 
> understand it correctly(by extrapolating from Hess's paper), the second 
> to fourth columns are the results from the off-diagonal elements of the 
> pressure
> tensor and the fifth one being the average. Hess's paper seems to be 
> this one(Figure 5). However, the results are way to low for the ionic 
> liquid, since I'm getting 1e-7 kg/m.s and way too low for a ionic liquid.
> the eviscoi.xvg file seems to be some kind of integral of evisco.xvg 
> which I'm not quite getting what it is.
> visco.xvg contains the shear viscosity and bulk viscosity, which again 
> seems to be the one I'm looking for. However, the shear viscosity I'm 
> looking is(the region which roughly is constant) about 3~5 times higher 
> than the experiment. The simulation is 12ns long. Of course, this could 
> mean (1) the force field is bad or (2) the Green-kudo converges too 
> slowly as Hess pointed out.

Either of these are possible.  Have you tried (1) different parameters or (2) 
longer simulations?  If the model does not reproduce the experiment, it is not a 
very good model...

> Since it is not getting any useful result, I turned to non-equilibrium 
> MD by applying a shear external force, which is in (2)
> (2) I tried both NVT and NPT simulations which uses berendsen coupling 
> bath. and set up the cos_acceleration = 0.1 nm/ps^2.
> However, I tried to use g_energy -f *edr -s *tpr and by selecting 
> the 1/Viscosity or 2CosZ*Vel-X option, the output are giving me zero 
> data points on all times.

This could be related to a bug that was fixed just a few days ago.  Try pulling 
the latest stable release from git and trying again.


> I tried really hard on searching  results on the user list of gromacs 
> but it seems that there are quite a lot questions on this but not a 
> single satisfactory answer has been given, the most popular answer 
> people gave is refer to Hess's paper, but even though Hess was giving a 
> good theoretical background on this paper, practically on how to set up 
> MD jobs is not given. 
> So I would appreciate anyone's hint.
> Cheers,
> Xiaohu


Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list