[...]
The difference in diffusion is ~50%. Is it logical?
If this matter is already discussed please forgive me.
that is a statement that cannot be easily forgiven, since there is a mailing list archive where *you* can easily determine for yourself whether a topic has been discussed before (it has). so please make it a habit to search through the archives first, or you will not be forgiven at all.
axel.
1. 50% difference is rather large. But before blaming the kspace
style, you should convince yourself that your diffusion coefficient
calculation is properly converged and you should estimate bounds for
statistical and systematic errors. For example, if you sample more
time origins, how does your estimated MSD(t) change? If you increase
t, how does MSD(t) change?
2. If indeed the kspace style is affecting MSD(t), it is almost
certainly affecting other properties too. In particular, small changes
in density can have a big effect on transport properties of liquids.
Aidan
Dear Aidan,
But of course I do not believe that pppm and ewald can give such differences. The reasons are different and I will find them. Thanks a lot.
Otto
Okay. Good luck!