[gmx-users] Calculating shear viscosity of a liquid - different results from different verions of Gromacs

[Mateusz Jasik]

Dear Gromacs Users,

I am trying to obtain the shear viscosity of an imidazolium-based ionic liquid by using periodic perturbation method. While recalculating some of mine old data, it turned out, that the results I get from the Gromacs 4.6.7 are different from those obtained from Gromacs 4.5.5 by two orders of magnitude.

As a test I have run a calculations using the same tpr files (created by the 4.5.5 verison) and two versions of gromacs. The mdp file I used is presented below:

title               = nvT MD
integrator          = md
constraints         = all-bonds
dt                  = 0.001
nsteps              = 12000000
nstenergy           = 100
nstlist             = 10
nstxout             = 0
nstvout             = 0
nstfout             = 0
nstlog              = 100
nstxtcout           = 1000
xtcprecision        = 1000
ns_type             = grid
coulombtype         = PME
rlist               = 1.0
rcoulomb            = 1.0
rvdw                = 1.0
tcoupl              = nose-hoover
tc_grps             = MENIM Tf2N
tau_t               = 0.1 0.1
ref_t               = 303.15 303.15
Pcoupl              = No
gen_vel             = no
gen_temp            = 298
gen_seed            = 173529
energygrps          = MENIM Tf2N
constraint_algorithm = LINCS
pbc                 = xyz
cos_acceleration        = 0.01    ; I have run simulations for five different values, ranging from 0.01 to 0.05


I obtained the 1/Viscosity values using "g_energy -f edr_file -s tpr_file -o output.xvg".

Here are the values I got from the program:

cos_acceleration    1/Viscosity (4.5.5)    1/Viscosity (4.6.7)
0.01                                   0.0583586                   5,91038    
0.02                                         0.3713                   57,4981
0.03                                       1.24583                   211.699
0.04                                       2.73402                   475.518
0.05                                       4.02399                   807.279

There have previously been numerous threads regarding the viscosity calculations, but I haven't found any with the problem similar to this. Are there any differences between the two version of program that can influence this value? And if so, which one should I treat as the more reliable? Or could this be a problem from my side (for example with the parameters used)?

Thanks and regards,
Mateusz