[gmx-users] Re: g_energy and viscosity

[Vitaly V. Chaban]

> Viscosity in g_energy is computed from the pressure tensor which is
> stored in the energy file. Unfortunately is seems like the acflen is
> ignored in this case. Please submit a bugzilla.
>
> On the other hand you need all the data you can get for this quantity,
> because it converges extremely slowly.
>

Dear Prof. van der Spoel,

Could you please give me a hint in what units the viscosity value is output?
Looking at the output of g_energy, the pressure is in [bar] and the
time is in [ps]. So for the viscosity units:
10e+5Pa*10-12s=10e-7Pa*s=10e-4mPa*s. Testing SPC water with quite a
big system (~6000 mols) and for a long time (~1 ns)
g_energy -vis
g_analyze -f enecorr.xvg -integrate
I get 176.8 (integral value). As follows from the above calculations
this number should be divided by 10000 to get the value in [mPa*s], so
it is 0.01768 from the calculation. However, the experimental one is
1.05e-3Pa*s...

Please give me a hint where I am wrong here.

Thanks,
Vitaly