[gmx-users] Diffusion coefficient water

[Justin Lemkul]

On 8/17/16 11:28 AM, Diez Fernandez, Amanda wrote:
> Dear Gromacs users,
>
> I was trying to reproduce the published values for the diffusion coefficient of SPCFw water (2.3 +/- 0.08 x 10^-9 m^2/s  )before using it in some of my simulations and I am failing to do so.
> More worryingly I get different results for different identical runs.
> I am using a 2nm side cubic cell containing 221 water molecules and a time step of 0.5fs (larger timestep results in a warning since I am not applying any constraints).
> I ran 6 identical simulations (only difference is the random initial velocities) and I got these different values and error magnitudes.
> The trajectories are 10ns long. I discarded the first 2ns for the msd calculation.
>
>  Nth Simulation         Diffusion Coefficient (x 10^-9 m^2/s)
>
>             1                              2.95  +/-  0.071
>             2                              2.71  +/-  0.158
>             3                              2.52  +/-   0.007
>             4                              2.96   +/-   0.26
>             5                              2.62   +/-  0.07
>             6                              2.71  +/-  0.158
>
> Is it normal that I get different results like this?  (Despite averaging over more than 7ns and 221 molecules?).
>

Diffusion constants are subject to finite-size effects.  Have you corrected for 
this phenomenon?  And perhaps more importantly, did the original authors of the 
water model correct their values?  I find that often times people don't, or 
otherwise don't say if the correction has been applied.  2 nm is a very small 
box, so the PBC correction will be quite substantial.

-Justin

> Many thanks in advance.  Below are the detailed parameters.
>
> Amanda
>
> Detailed parameters:
>
> nstlist                 = 10
>
> ns_type                 = grid
>
> rlist                   = 0.9
>
> coulombtype             = PME
>
> rcoulomb                = 0.9
>
> fourierspacing          = 0.12
>
> pme_order               = 4
>
> ewald_rtol              = 1e-6
>
> vdw-type                = cut-off
>
> rvdw                    = 0.9
>
> pbc                     = xyz
>
> tcoupl                  = nose-hoover
>
> tc_grps                 = SYSTEM
>
> tau_t                   = 0.4
>
> ref_t                   = 298
>
> nsttcouple              =-1
>
> nh-chain-length         = 1
>
> pcoupl                  = Berendsen
>
> pcoupltype              = isotropic
>
> tau_p                   = 10.0
>
> compressibility         = 4.5e-5
>
> ref_p                   = 1.01325
>
> refcoord-scaling        = all
>
> DispCorr                = EnerPres
>
> gen_vel                 = yes
>
> gen_temp                = 293
>
> gen_seed                = -1
>
> constraints             = none
>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================