19990617: different randomization wrt 19990614

GOAL

Run exactly the same as 19990614, but with a different seed for randomization. To see if it makes any difference.

Results

It looks like different randomizations give different diffusion coefficients.

Actually, this is well documented in [6,7].

Here are the results for this run:

From the slope of oxygen MSD's the following values for the diffusion coefficient are obtained (for procedure see 19990614):

$$\begin{array}{lll} &\mbox{slope}\ (\mbox{\textit{\AA}}^2/ps)... ..._{\mathit{shell}} &0.0110220193 & 1.83700321667 \\ \end{array}$$

(MSD data fitted in the interval $[20.0:140.0]\;ps$)

And here is the comparison of oxygen MSD's (core only) between 19990617 and 19990614:

Following the two papers [6,7], we should setup a criterion for building up the starting configuration. Something like: ``no $Ce^{\prime}_{Ce}$ point defects in nearest neighbor position'' or ``no $V^{\cdot \cdot}_O$ centers with $Ce^{\prime}_{Ce}$ impurities in nearest neighbor position'' or whatever. In other words, we have to ``standardize'' the starting configuration in order to make results from different runs comparable.