[gmx-users] Comparison of diffusion constant from Einstein and Green Kubo

[abhishek khetan]

Dear gmx-users,

I wanted to calculate the diffusion coefficient of lithium ions is a
solution. I tried using the gmx msd routine first with the options -nomw
and -mol. I want to know if -nomw should be used or not if I want to
compare my results with experiments. Anyway The result I obtained from here
was :

$ gmx_mpi msd -f dms216_16lino3_npt.xtc -s npt.tpr -n index.ndx -o li.xvg
Fitting from 100 to 900 ps
D[        LI] 0.1093 (+/- 0.0086) 1e-5 cm^2/s

Next, i tried using the velacc routine as:
gmx_mpi velacc -f dms216_16lino3_npt.trr -s npt.tpr -n index.ndx -o li.xvg
and then integrated it using the analyze routine to get the result
$  gmx_mpi analyze -f li.xvg -integrate

Read 1 sets of 101 points, dt = 5

Calculating the integral using the trapezium rule
Integral 1     2.36293  +/-    0.00000
                                      std. dev.    relative deviation of
                       standard       ---------   cumulants from those of
set      average       deviation      sqrt(n-1)   a Gaussian distribition
                                                      cum. 3   cum. 4
SS1   9.644851e-03   9.989920e-02   9.989920e-03       6.042   30.877

When I multiply the obtained result ( 2.36293) by 10^(-2) and divide by 3
to get the final answer in cm2/s, the value is 7.87 x 1e-3 . This is orders
of magnitude different from the einstein style calculated D = 0.1093 (+/-
0.0086) 1e-5 cm^2/s. I tried increasing the -dt value to 100 and then the
output I get is:

Read 1 sets of 6 points, dt = 100

Calculating the integral using the trapezium rule
Integral 1    64.40000  +/-    0.00000
                                      std. dev.    relative deviation of
                       standard       ---------   cumulants from those of
set      average       deviation      sqrt(n-1)   a Gaussian distribition
                                                      cum. 3   cum. 4
SS1   1.990000e-01   3.611937e-01   1.615307e-01       1.079    0.362

Which is much worse off. Can anyone please suggest what I might be doing
wrong.

Best,

-- 
MfG,
abhishek